.

. .

39!“). .-

~ you: A

v ....L.m..u“r
1 ‘
:5

 

 

This is to certify that the
dissertation entitled

Adolescents’ Bonding To School: The Impact Of
Perceived School Climate Pathways On Middle School
Students’ Bonding To School

presented by
Zora Robab Ziazi

has been accepted towards fulfillment
of the requirements for the

degree in Family and Child Ecolggi

Wfim

/ MajorvPRfessor’s Sigiature
S’fiw/a 7

Dee

PhD.

MSU Is an Afiinnative Action/Equal Opportunity Institution

 

 

ADOLESCENTS’ BONDING TO SCHOOL: THE IMPACT OF PERCEIVED
SCHOOL CLIMATE PATHWAYS ON MIDDLE SCHOOL STUDENTS’
BONDING TO SCHOOL
By

Zora Robab Ziazi

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Family and Child Ecology

2004

ABSTRACT
ADOLESCENTS’ BONDING TO SCHOOL: THE IMPACT OF PERCEIVED
SCHOOL CLIMATE PATHWAYS ON MIDDLE SCHOOL STUDENTS’
BONDING TO SCHOOL
By
Zora Robab Ziazi

This study investigated the impact of school climate on students’ bonding to
school among a sample Of middle school students from two schools in an urban district in
Michigan, with a high percentage of minority and economically disadvantaged
population. The variables of interests in this study consisted of five constructs of school
climate measuring students’ perceptions of the environment of their school and four
constructs of school bonding that measured the extend of students’ bonding to school.
Data for this study were collected during Spring 2003 as part of a larger project
(Michigan Character Education Project, MCEP).

The main research questions in this study were posed to determine a) how sub-
groups as identified by gender, race/ethnicity, grade level and school varied in their
bonding to school and b) the pathways that linked students’ perceptions of school climate
to students’ bonding to school and to find out whether or not there were any differences
among students by gender, race/ethnicity and school. Multivariate Analysis of
Covariance (MANCOVA) was employed to analyze test the research hypotheses relevant
to school bonding, controlling for students background characteristics. Structural
Equation Modeling (SEM) technique was used to examine the research hypotheses

related to pathways that linked the indicators of school climate to students’ bonding to

school.

Results showed significant differences among sub-groups for both sets of
hypotheses. Major findings of the study indicated a steady decrease in students’ bonding
to school from grade 6th to grade 8‘“. Perceived support and care for teachers and staff in
school as well as respect for all members of the school and for school property were
major contributors to students’ bonding to school. Sub-group differences and estimates
of the pathways were discussed. Findings seemed to point and encourage school
characteristics that provide a healthy context in which students can develop healthy
attachment to adults, students and to school.

Limitations of the study included a non-random sample and small sample sizes for
some of the sub-group analyses. Future directions included addressing the
methodological limitations of the study by drawing random sample and including a larger
sample for sub-group by race/ethnicity. Furthermore, future research should conduct

grade level by gender comparisons for school bonding.

DEDICATION

I dedicate this dissertation to my parents, Hassan and F atemeh Ziazi, who taught
me the value of education and high ideals in life; to my daughters Saqi and Dina Salehi,
who encouraged me to continue even when things were difficult; and to all children and

youth who have suffered through an imperfect education system.

iv

ACKNOWLEDGEMENTS

First, and foremost, I thank the Creator who enabled me in this long process,
always giving me ideas, hope, and inspiration, and bringing so many wonderful people
into my life who supported me in this work. I also want to thank my lovely family and
my two daughters Saqi and Dina Salehi who have always been there for me during times
of difficulty when I needed support. Dreams do come true, folks .....

Yes, I did my part by working hard and persevering. But, I could not have
succeeded without the dedicated support of so many wonderful people. I had the best
dissertation committee any doctoral student could wish for. I want to thank my
dissertation Director Dr. Joanne Keith whose leadership and direction, as well as innate,
creative thinking, Opened doors for me all along the way. I also want to thank my
committee members Dr. Francisco Villarruel, Dr. June Youatt, and Dr. Ken Frank for
their support, directions, valuable suggestions, and guidance in this process. Another
committee member 1 wish to give a special thanks to is Dr. Melissa Quon Huber, who is
also my mentor and friend. Her positive thinking, encouragement, and endless support,
as well as her excellent intellectual ability, paved the way and got me through. Thank
you , Melissa. You are awesome.

A very special thanks to my friend and colleague Dr. Linda Patriarca whose
genuine support and friendship, when I needed it the most, gave me hope and
encouragement to continue. I thank my dear friend and editor, MS. Sarra Baraily for her
excellent editorial input and for her support whenever it was needed. I also want to thank
Dr. Randall Fotiu, Dr. Comic Page, Dr. Alexander von Eye, and Dr. Stan Kaplowitz at

Michigan State University for their valuable time and statistical consultation. Last, but

not least, a Special thanks to my dear friends Dr. Brenda Bryant, Ms. Adela Pena, Ms.
Jodi Castillo, and Mr. Don LaGuire for their friendship and support. Thank you, guys,

believe it or not, the work is finished and I’m done!

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................... x
LIST OF FIGURES ................................................................................................. xii
CHAPTER 1
INTRODUCTION .................................................................................................... 1
Social Significance of the Study ........................................................................ 4
Underlying Theories ...................................................................... 5
CHAPTER 2
LITERATURE REVIEW .................................................................. 7
Purpose of the Study ................................................................... 7
Rationale for the Study ...................................................................................... 7
Focus of Literature Review ................................................................................ 8
Definition and Measurement of School Climate ................................................ 9
School Bonding .................................................................................................. 12
Importance of Understanding School Bonding .......................................... 12
Background on School Bonding ........................................................................ 12
Roots in Attachment Theory ....................................................................... 12
Expanded Use of Attachment Theory ......................................................... 12
Definition and Use of the Term School Bonding ....................................... 13
Importance of School Bonding in Promoting Positive Student Outcomes. 14
Summary of Importance of School Bonding in Student Outcomes ............ 18
Predictors of School Bonding ..................................................................... l9
Limitation of Previous Research ........................................................................ 23
Addressing Gaps in the Literature ...................................................................... 24
Research Hypotheses ......................................................................................... 25
Research Hypotheses Related to School Bonding ...................................... 25
CHAPTER 3
METHODOLOGY .................................................................................................. 30
Participant Recruitment ....................................................................................... 30
Participants ............................................................... ' ............................................ 31
Data Collection Procedure ................................................................................... 34
Measurements ...................................................................................................... 35
School Bonding .............................................................................................. 35
School Climate ............................................................................................... 36
Demographic Variables ...................................................................................... 38
CHAPTER 4
ANALYSIS ............................................................................................................... 39
Background Analyses ........................................................................................ 39
Descriptive Analyses .......................................................................................... 39

vii

School Climate Descriptive Analysis ............................................................ 39

School Bonding Descriptive Analysis ........................................................... 40
Reliability Analysis .............................................................................................. 44
Testing Research Hypotheses .............................................................................. 45
Analysis of Group Differences ............................................................................ 45
Interpretation Of MANCOVA Tables .................................................................. 47
Presentation Of MANCOVA Analyses ................................................................ 47

Hypothesis 1 ................................................................................................... 48

Hypothesis 2 ................................................................................................... 49

Hypothesis 3 ................................................................................................... 51

Hypothesis 4 ................................................................................................... 54
Structural Equation Modeling (SEM) Analysis ................................................... 56

Interpretation of the SEM Analyses ............................................................... 57

Presentation of SEM Analyses ....................................................................... 57

Initial Hypothesized Model for all Analyses ................................................. 58

Hypothesis 5 ................................................................................................... 60

Hypothesis 6 ................................................................................................... 65

Hypothesis 7 ................................................................................................... 72

Hypothesis 8 ................................................................................................... 81

Power Analysis for Tests Of Model Fit .......................................................... 89

CHAPTER 5
DISCUSSION ........................................................................................................... 92
Group Differences on Measures of School Bonding ........................................... 92

School Bonding by Gender ............................................................................ 92

School Bonding by Race/Ethnicity ................................................................ 93

School Bonding by Grade Level .................................................................... 95

School Bonding by School Effect .................................................................. 97
Structural Equation Modeling (SEM) Analysis ................................................... 99

Pathways of School Climate to School Bonding for the Entire Sample ........ 99

Pathways for Sub-Groups by Gender ............................................................ 100

Pathways for Sub-Groups by Race ................................................................ 102

Pathways for Sub-Groups by School ............................................................. 103
Summary of Findings and Conclusions ............................................................... 105

Summary of Findings for the Entire Sample ................................................. 105

Summary of Findings by Grade Level ........................................................... 105

Summary of Findings by Gender ................................................................... 106

Summary of Findings by Race ....................................................................... 106

Summary of Findings by School .................................................................... 107

Conclusions .................................................................................................... 107
Methodological Problems/Limitations ................................................................. 108
Implications and Future Directions ...................................................................... 109

APPENDICES .......................................................................................................... 1 13
Appendix A - Survey Items for Students’ Demographic Characteristics
and Perceived School Climate ....................................................................... 1 l4

viii

Appendix B — Survey Items for School Bonding ................................................ 120
Appendix C — Results Of the Multivariate Analysis of Variance ......................... 123

REFERENCES ......................................................................................................... 126

ix

LIST OF TABLES

Number Page
Table 3.1 .................................................................................................................... 31
Table 3.2 .................................................................................................................... 32
Table 3.3 .................................................................................................................... 33
Table 4.1 .................................................................................................................... 40
Table 4.2 .................................................................................................................... 43
Table 4.3 .................................................................................................................... 44
Table 4.4 .................................................................................................................... 49
Table 4.5 .................................................................................................................... 50
Table 4.6 .................................................................................................................... 53
Table 4.7 .................................................................................................................... 55
Table 4.8 .................................................................................................................... 61
Table 4.9 .................................................................................................................... 61
Table 4.10 .................................................................................................................. 63
Table 4.1 l .................................................................................................................. 64
Table 4.12 .................................................................................................................. 66
Table 4.13 .................................................................................................................. 67
Table 4.14 .................................................................................................................. 68
Table 4.15 .................................................................................................................. 70
Table 4.16 .................................................................................................................. 71
Table 4.17 .................................................................................................................. 71
Table 4.18 .................................................................................................................. 74

Table 4.19 .................................................................................................................. 75

Table 4.20 .................................................................................................................. 75
Table 4.21 .................................................................................................................. 76
Table 4.22 .................................................................................................................. 79
Table 4.23 .................................................................................................................. 80
Table 4.24 .................................................................................................................. 80
Table 4.25 .................................................................................................................. 81
Table 4.26 .................................................................................................................. 83
Table 4.27 .................................................................................................................. 84
Table 4.28 .................................................................................................................. 84
Table 4.29 .................................................................................................................. 87
Table 4.30 .................................................................................................................. 88
Table 4.31 .................................................................................................................. 88
Table 4.32 .................................................................................................................. 90

xi

LIST OF FIGURES

Number Page
Figure 4.1 ................................................................................................................... 59
Figure 4.2 ................................................................................................................... 60
Figure 4.3 .................................................................................................................. 63
Figure 4.4 ................................................................................................................... 65
Figure 4.5 ................................................................................................................... 66
Figure 4.6 ..................................................................... ' .............................................. 6 9
Figure 4.7 ................................................................................................................... 69
Figure 4.8 ................................................................................................................... 72
Figure 4.9 ................................................................................................................... 73
Figure 4.10 ................................................................................................................. 73
Figure 4.1 1 ................................................................................................................. 77
Figure 4.12 ................................................................................................................. 78
Figure 4.13 ................................................................................................................. 78
Figure 4.14 ................................................................................................................. 82
Figure 4.15 ................................................................................................................. 82
Figure 4.16 ................................................................................................................. 86
Figure 4.17 ................................................................................................................. 86

xii

CHAPTER 1

INTRODUCTION

This study investigated the impact of perceived school climate on students’
bonding to school among a sample of middle school students from two schools in an
urban district in Michigan, with a high percentage Of minority students. The variables Of
interest in this study consisted of five constructs of school climate measuring students’
perceptions of the environment Of their school and four constructs Of school bonding that
measured the extent of students’ bonding to school. Data for this study were collected
during the spring Of 2003, as part Of a larger grant study (Michigan Character Education
Project, MCEP). To examine the research hypotheses, several techniques for analysis
were utilized in this study. These included descriptive methods, reliability analysis, the
Multivariate Analysis of Covariance (MANCOVA), and the Structural Equation
Modeling (SEM) technique. The main research questions in this study were posed in
order to determine the pathways that links students’ perceptions of school climate to
students’ bonding to school and to find out whether or not there were any differences
among the students by gender, race/ethnicity, and schools to which they were enrolled.
Differences among students by gender, race/ethnicity, grade level, and school on
measures of school bonding, were also examined.

Convincing evidence in the literature has identified bonding to school as a
significant protective factor against risks for adolescents. Studies have indicated that
positive and strong bonding to school increases students’ positive outcomes such as

physical and mental health, achievement, increased motivation and self-esteem, while

decreasing the likelihood of adolescents’ involvement with anti-social behavior (i.e., drug
abuse, smoking, drinking and sexual activities) (McBride, Curry, Cheadle, Andennan,
Wagner, Diehr & Psaty, 1995). Bonding to school has also been associated with a
decrease in school violence (Simons-Morton, Crump, Haynie & Saylor, 1999). Najaka
(2001) reported school bonding as the most convincing evidence of a relationship
between risk and problem behavior. Positive changes in bonding and commitment to
school were consistently related to positive changes, or decreases, in problem behavior.
Weak or low bonding to school, however, is associated with negative school outcomes
for youth. In addition, it has been reported as a risk factor for adolescent substance
abuse.

Bonding to school as a protective factor is especially important during
adolescence. This is because during this stage of life many individuals have difficulty
dealing with turmoil and conflicts that are caused by internal and external factors such as
psychological, physical and social changes, as well as changes in the roles and
expectations (Lerner, Lerner, von Eye, Ostrom, Nitz, Talwar-Soni, & Tubman, 1996;
Lanni, & Orr, 1996). These factors, in turn, add to the vulnerability of the adolescents to
potential risks in their environment, particularly for those who do not have the internal
stamina or do not receive the necessary support from their families and social circle to
buffer these risks (Anson, 1995). Therefore, bonding to school can serve an important
role in protecting the youth from potential risk factors.

Among several other factors, school climate has been reported as a significant
predictor of students’ bonding to school. Previous literature has indicated that social

Opportunities for students to feel a sense of belonging to participate in their social

environment can act as a protective factor against stress and decreases adolescents’ risk
taking behavior (Hawkins, Catalano & Miller, 1992; Jessor, 1994). Following past
studies, McBride et a1. ( I 995) conducted an investigation and found that pro-social
bonding environment was associated with a Significant increase in bonding among ninth
grade students and a reduced likelihood of their engaging in risk-taking behaviors.
Several other studies have also confirmed the association between school climate and
school bonding and a few studies have investigated the impact of school climate on
students’ bonding among adolescents (e. g., Anson, 1995; Kumpfer & Turner, 1990-
1991). While these studies have provided valuable information, there still remain a
number of important areas that are not addressed and need further investigation.

First, given the significance of school bonding as a protective factor for
adolescents, and the impact it has in increasing positive outcomes and decreasing the
negative outcomes, it is important that this construct be fully understood. However,
studies on school bonding have not addressed differences among students by
demographic variables of gender, race/ethnicity, grade level or school. Second, pathways
from students’ perceptions of school climate to students’ bonding are not well understood
and to date only few studies have indicated how school climate variables measured at the
individual level (i.e., individual’s perceptions of school climate) affect variables of
students’ bonding to school. Third, very few studies have considered how pathways of
school climate to school bonding may differ in student sub-groups as identified by
gender, race/ethnicity and school. Fourth, only few of these studies have been conducted
on samples of students from urban areas or included middle school students. Therefore,

this study seeks to extend and improve the previous literature by addressing the following

questions:

Question I. What are the differences between students by gender, race/ethnicity, grade
level and school on measures Of school bonding?

Question 2. What are the pathways from perceived school climate to school bonding?
Question 3. Are sub-groups of students as identified by students’ gender, race/ethnicity
and school different with respect to the pathways that links perceived school climate to

school bonding? And if so how are the sub-groups different?

Social Significance of the Study

The value of this study contributes to the field of positive youth development in
several ways. First, this study will include comparisons between several minority groups
and their Caucasian counterparts, particularly those economically disadvantaged from an
urban district, who are most underserved and are generally underrepresented in the
population. Evidence suggests that minority students are also underrepresented in this
area of research as to date, only few studies concerning students’ perceptions of school
climate and students’ bonding have included a comparison between minority and other
students (e.g., Chase, 2000) and there is a need for further investigation. Second, this
research seeks to determine if race, gender, or school impacts the relationship between
perceptions of school climate and school bonding. The potential differences have not
been fully understood previously. Third, the results of this study can help to add to the
knowledge base a) from which interventions can be further developed such as programs
to improve those aspects of school climate that significantly contribute to students’

positive bonding to school, thus increasing positive school outcomes while decreasing the

likelihood of students’ involvement in anti-social and negative behaviors; b) to inform
school practitioners to put forth more efforts and to invest more resources in developing
the aspects of school climate that lead to students’ positive bonding tO their schools thus

serves for allocation of resources for greatest student outcomes.

Underlying Theories

The theoretical foundation of this research is rooted in Bronfenbrenner’s
ecological system’s theory (1979), and Hirschi’s Social Control Theory (1969).
Bronfenbrenner’s theory holds that human development is affected by the interactions of
person and the context or environment over time. Growth of an individual cannot be
separated from the context but it must be examined within the environment in which the
developing individual grows. The growing individual’s development is affected by a set
of inter-related, nested structures (micro, meso, exo and macro systems).

Hirschi’s Social Control Theory asserts that feelings of attachment to school and
teachers have an important role to play in decreasing deviance and delinquency.
Furthermore, it refers to attachment to family and school as one of the primary elements
of establishing a social bond. From this perspective, other elements are commitment to
conventional pathways of achievement and beliefs in the legitimacy of societal order.
Hirschi’s theory originated from Bowlby’s Attachment Theory (1979), which was
concerned about the importance of the attachment bond between infants and young
children and their primary care takers in their healthy development. This idea of
attachment has been extended from primary caretakers and family environments to

include other social and religious institutions such as schools and churches.

In formulating the research topic presented in this thesis, both of these theories
seem to be most appropriate to consider. Hirschi’s theory provided the grounds for
extension of the concept of attachment to schools, allowing for studying school bonding.
To address the impact of school climate on students’ bonding to school,
Bronfenbrenner’s theory allowed for an examination of the climate of the school micro-
system and how it affected the growing individual (i.e., the middle school students). As
a context, middle schools have a potential to impact students’ development towards
positive pathways that lead to success or failure. Schools can impact students’
development towards positive trajectories by providing accepting and affirming climate
in which students can grow in positive self-regard and self-esteem (Garbarino, 1985).
Students’ perceptions of school climate impacts their attitude and behavior, and
consequently it affects their outcomes. Supportive and caring school environments can
provide the appropriate context and opportunity for positive growth Of students while
serving as a protective factor against potential risks. While the role of middle schools are
very important in creating a positive climate for development, a student’s character and
deveIOpment is also important in responding to the conditions in schools.

Thus in understanding and examining the impact of school as a context for middle
school students’ bonding to school both students’ perception of school climate as well as

individual’s characteristics must be considered which are addressed in this study.

CHAPTER 2

LITERATURE REVIEW

Purpose of the Study—Statement of the Problem

Using Bronfenbrenner’s (1979) ecological model of human development and
Hirschi’s (1969) Social Control Theory as the theoretical frameworks, this study seeks to
extend and improve on previous research by considering a) the differences in school
bonding by students’ characteristics of gender, race/ethnicity, and grade level as well as
school to which they belong, b) the dimensions of perceived school climate that
contributes to students’ bonding to school among a sample of middle school students
from an urban district Michigan, and c) pathways of perceived school climate to student
bonding among students for subgroups of students as identified by students’ gender,
race/ethnicity and school.

Rationale for the Study

Adolescents are vulnerable to potential risks due to many factors that impact their
lives during this transitional stage. The importance of bonding to school as a protective
factor against many risks, especially for adolescents has been well documented. It is
shown that bonding to school contributes to students’ positive outcomes such as
achievement, motivation, self-esteem physical and mental health, while decreasing
negative outcomes such as smoking, drinking, drug use and sexual activities.

Furthermore, the Significance of school climate in relation to students’ bonding to
school has been established. Previous literature has indicated school climate as one Of

the significant predictors Of students’ bonding to school. It is important to consider the

environment of school when studying school bonding because it has been documented
that student-school bonding can be highly increased by improving the climate of the
school.

Focus of Literature Review

A comprehensive electronic search of two major databases, “PSYCHINFO” and
“ERIC,” was conducted between the years 1992 to 2004 in order to find the relevant
literature on the topic of this study. While the search was limited to including only
studies from the past 12 years (in order to avoid studies that were out—dated), when
resources were limited, the search was extended to identify studies that were conducted
prior to 1992 as well (i.e., from 1966 to 1992).

Two separate literature searches were conducted to identify the studies relevant to
the topic of interest for this investigation. First, a search was undertaken to identify
studies that linked school bonding to school climate. The key terms used were school
climate, school bonding, school belonging, attachment, school environment or some
combination Of these terms such as school environment and attachment. Result of the
search yielded only 45 studies on the topic. These studies were further filtered out so that
only those with school climate as an independent variable and school bonding as a
dependent variable were kept. Second, a broad search was conducted to identify studies
related to school bonding using the key terms school bonding, attachment to school,
attachment, and school belonging. This search was further narrowed to include only
studies that predicted school bonding. Both of these groups of studies are summarized in

the literature review section of this study.

Definition and Measurement Of School Climate

There are various opinions in the literature as to how the term school climate is
defined and at what level it is actually measured. Tagiuri (1968) defined school climate
as the total environmental quality within an organization and stated that it is consists of
four dimensions: ecology, milieu, culture, and social system. A more recent definition of
the term has referred to school climate as the sum of the values, cultures, safety practices,
and organizational structures within a school that cause it to function and react in
particular ways (McBrien & Brandt, 1997). These definitions reflect an aggregate
definition of school climate and suggest that both physical and social aspects of the
climate are important.

Although school climate may be conceptualized as the aggregation of the physical
and social aspects at the school level, several studies (e. g., Chase, 2002; Simons-Morton,
et al., 1999; Kurnpfer, & Turner, 1990-1991) have measured school climate at the
individual level. That is, they have assessed perceptions of the research participants
about the climate of their school when studying topics relevant to school climate. Fewer
studies (e.g., Anson, 1995) have used an aggregate measure to reflect the school level
climate of school, particularly given the large sample of schools needed in order to
conduct powerful statistical analyses at the building level.

In addition to differences in the level (individual vs. school) at which school
climate is measured, differences also exist in what is included in a school level analysis.
For some studies (e. g. Partnership for Academic and Career Education, 1992), measuring
school climate at the school level includes inventory of school level materials and

activities that impacts or reflects the climate of the school, such as activities to involve

parents, career planning guides, award ceremonies and other recognition activities to
celebrate success in school, etc. While other studies (e.g., Welsh, Wayne, Jenkins, &
Greene, 1998) have measured school culture as part of the school level climate. Still
Others (e.g., Gonder & Hymes, 1994) have emphasized the importance Of distinguishing
between school culture and school climate. They have argued that school climate and
school culture are two different concepts and should be measured separately. Similarly
some individuals (e.g., Johnson, 1998-1999) have suggested a distinction between
affective and experiential features of the setting (such as respect, high morale, caring, and
trust) and the cognitive and managerial components (such as Opportunity for input,
continuous academic and social growth, cohesiveness, and school renewal) of school
climate.

Measurement of school climate has also varied in how it is collected. One type of
aggregate school level assessments includes self-assessments. These assessments have
been designed to measure organizational or managerial climate of school and are
intended for internal organizational improvement. These are designed to be filled at the
district or school level by key informants within the school and are school building level
observational ratings. (e.g. ‘Self-Assessment Inventory’, Partnership for Academic and
Career Education, 1992). However, a large number of school climate assessment tools
have been developed to assess students’ perceptions of their personal learning
environment, and many of them are designed and used primarily for research purposes
(e.g. Classroom Environment Scale (CES), Moos & Trickett, 1995; CFK, Ltd School
Climate Profile, Howard, Howell, & Brainard, 1987; Education, Training, Research

Character Education Survey, ETR Associates, 2000). These surveys include items about

10

students’ perceptions Of the “actual” school climate as it exists, or about their “preferred”
climate as they would like it to be (e.g., Individualized Classroom Environment
Questionnaire (ICEQ), Fraser, 1990). Still other assessments rely on the individual
perceptions of not only students, but also teachers, parents and others (Huber & Baraily,
2002) such as ‘Comprehensive Assessment of School Environments’ “CASE” (National
Association of Secondary School Principals, 1987). These authors claim that this method
of collecting perceptions Of climate from all groups (students, parents, teachers, and staff)
tends to be persistent and stable over time. It should be noted that although many
different surveys are designed to assess individuals’ personal experience of the climate
and are usually analyzed at the individual level, these type of surveys have also been used
to assess school level climate at a higher level through use of school-level or classroom
level mean perceptions. However, the individual level of analysis is a more common use
of the survey data (Huber & Baraily, 2002).

In this study, school climate referred to each individual student’s perceptions Of
the environment of the school as they experienced it for themselves and as they observed
it in other fellow students or adults in school. Therefore, the items included in the school
climate survey inquire about individual’s perceptions of their own experience in the
school as well as that of other individuals in the school. The term perceived school
climate is preferred in this study as it captures the essence of what this construct actually

measures.

11

School Bonding

Importance of Understanding School Bonding

Positive bonding to school has been identified as a Significant protective factor
against many risks, especially during the adolescent stage of development (Simons-
Morton, et al., 1999). This period Of growth is considered to be a time of risk and
upheaval and a dramatic increase in occurrence of problem behavior which is rare prior to
transition to middle school (Eccles, Lord & Buchanan, 1996). Although currently there is
consensus that adolescence is not a time of extreme psychological turmoil as research
shows that only 5 to 20 percent of youth go through severe emotional disturbance, a great
deal of research shows that adolescence period is filled with emotional tension, conflict
and turmoil especially with parents, siblings and peers (Smentana, 1995). During this
period of development most individuals also experience internal and external shifts such
as physical, and psychological changes of puberty, transition from elementary school to
middle school, or the new role demands by parents, peers and teachers, which adds to
their tension and make them vulnerable to potential risks in their environment.
Therefore, positive bonding to school can have an important buffering effect against the

risk factors adolescents may face.

Background on School Bonding
Roots in Attachment Theory
The school bonding literature is an extension of attachment theory. The concept
of attachment was originally referred to the relationship between caretaker and child

(Bowlby, 1969). It suggests that infants establish strong affectionate bonds, or

12

attachment, to the caretaker(s) who meets the needs of the child (Ainsworth, 1989,
Bronfenbrenner, 1979). This bond provides the context in which dramatic development
occurs, allowing the infant to gradually move towards increasing separateness and
individuation within the context of a meaningful and secure relational base. Attachment
theory is based on the collaborative work of John Bowlby and Mary Salter Ainsworth. In
their joint work together, Bowlby came up with the idea, while Ainsworth contributed by
developing a way to measure infant-mother’s attachment (Ainsworth, 1989).

Expanded Use of A ttachment Theory

Over the years, the concept of attachment has been expanded to include other
attachment relationships besides mother-infant (e. g., adolescent peer relationship) and
other social settings such as schools, churches, and other adults such as teachers (Anson,
1995). Therefore, an attachment process similar to the one occurs with the primary
caretaker may occur with other family members, special friends, and kin which serves to
socialize children. Since then researchers have explored the notion of attachment to
school and have conducted studies to inquire the impact Of student’s feelings Of
attachment or bonding to school.

Definition and Use of the Term ‘School Bonding'

The concept of students’ bonding to school has been defined and used in the
literature inconsistently by various authors. Some studies have referred to this concept as
school attachment or attachment to school (Hoppe, Wells, Haggerty, Simpson, Gainey &
Catalano, 1998; Anson, 1995), while others have used the terms school bonding (e.g.,
Simons-Morton, et al., 1999; Kumpfer & Turner, 1990-19991), social bonding (McBride

et al., 1995), or school belonging (Goodenow, 1991).

13

School attachment has been defined as a sense Of belonging at school (Dworkin,
1987; Goodenow, 1991), a network of relationships with peers and other school
personnel (Valverde, 1987; Williams, 1987); and a sense of inherent value for the
learning process as it relates to students’ lives (Dworkin, 1987).

With regard to school bonding, Hawkins and Weis (1985) defined this concept as
attachment to pro-social peers, commitment to conventional academic and social
activities at school and belief in the established norms fro school behavior. They viewed
bonding to school through Hirschi’s Social Control Theory (1969). Moreover, McBride
et al. (1995) identified key elements of social bonding as: “attachment, the quality of
affective relationships that the adolescent has with family and friends; commitment, the
degree to which adolescents’ aspire to pro-social goals; and involvement, the degree to
which the adolescent is integrated in conventional social activities” (p. 63).

For the ease of reading, in this study the term “school bonding” is chosen as the
preferred term to refer to the concept of students’ feelings of attachment to school.
“School bonding” is also used in this document to refer to what other studies might have
referenced as “attachment to school”, “social bonding”, “belonging to school” or other
similar terms.

Importance of School Bonding in Promoting Positive Student Outcomes

Previous literature has indicated that positive bonding to school is associated to
positive student outcomes. For example, Anson (1995) reported that in a sample of
middle schools students’ positive bonding to school promoted their pro-social behaviors
and positive mental health as well as contributing to their academic achievement.

Mouton, Hawkins, McPherson and Copley (1996) also found an association between

14

positive bonding and students’ to academic achievement. In addition, these authors
showed that positive school bonding was related to student’s self-esteem and motivation.

Furthermore, there has been theoretical evidence to Show that students’ bonding
to school is related to reducing negative student outcomes. Social Control Theory
(Hirschi, 1969) asserts that feelings of bonding to school play an important role in
decreasing deviance and delinquency. Researchers (e.g., Hawkins & Weis, 1985) have
applied this theory to their studies in which they indicated that youth who bond to pro-
social institutions such as home and schools are less vulnerable to negative peer
influences and negative outcomes. Kirby (2002) reviewed several studies and found
school bonding to be significantly related to less sexual behavior and lower pregnancy
rates among adolescents. School programs, that were designed to increase school
bonding, delayed students’ sexual activities or reduced pregnancy rates even when they
did not address sexuality. School bonding has also been associated with reduced school
violence and decreased drug use among adolescents (Simon-Morton, et al., 1999).
Another study Hoppe, Wells, Haggerty, Simpson, Gainey, & Catalano, 1998) revealed
that bonding to school decreased the odds of smoking and drinking in youth. In these
studies, school bonding, as a single variable, was important in reducing risk behaviors in
adolescents.

Additional studies have demonstrated the importance of school bonding in
reducing these risk behaviors, even when multiple predictors are considered. In some
cases, school bonding is a direct predictor and in other cases it is a mediator. Bryant,
Schulenberg, Bachman, O’Malley, and Johnston (2000) reported that school bonding, as

one of several indicators Of school experience, had an impact on decreasing cigarette use

15

among the students of eighth through twelfth grades. School bonding contributed to a
decrease in cigarette use and acted as a buffer against cigarette smoking. In another study
(Hawkins, Graham, Maguin, Abbott, Hill & Catalano, 1997) authors found that school
bonding, among other factors, mediated the relationship between alcohol-use at younger
age (10-11) to a higher level of alcohol misuse at age 17—18. Finally, Najaka (2001)
conducted a meta-analysis and found school bonding to be the most convincing evidence
of a relationship between risk and problem behavior. Positive changes in bonding and
commitment to school were consistently related to positive changes, or decrease in
problem behavior. These three studies indicate the importance of school bonding in
reducing risk behaviors, even when other possible predictors are considered.

Additionally, the degree of bonding to school, identified as strong or high and
weak or low was investigated by some authors. It is indicated that strong bond to school
decreases negative outcomes in youth. For example, McBride, et a1. (1995) reported that
in a sample of ninth grade students, individuals with high level of bonding to school were
less likely to exhibit high-risk behavior such as smoking, substance abuse, and sexual
activities. Low or weak bonding to school is associated with negative school outcomes
for youth. Results of a study (Hallfors & Van-Dom, 2002) showed that low school
bonding was a risk factor for adolescence substance abuse.

Some authors have focused on examining school bonding by Students’
characteristics such as grade level and gender. One study (Simons-Morton, et al., 1999)
compared middle school students’ bonding to school by their grade level. School
bonding was significantly higher among 6th graders than for 7m graders and 8m graders.

Similarly, 7"I graders had a higher score on school bonding than their peers in 8‘h grade.

16

Authors reported that students in higher grades liked school less, and felt less closely
affiliated with school. Furthermore, this study also reported significant gender
differences on measures of school bonding and found a higher bonding to school for
females than the male students. McDonald and Wright (2002) investigated gender
differences in relation to low school bonding and showed that female students who
smoked cigarettes were more likely to feel less bonded or attached to school than either
males or nonsmoking females. The relationship between low school bonding and poor
student outcomes has been Observed among students from a variety of backgrounds.
Storino (2001) showed that weak school bonding was a significant predictor Of problem
behaviors in a sample of Hispanic students. Robertson, Harding and Morrison (1998)
also reported that low school bonding was among the variables that contributed to poor
peer self-concept and alienation among at-risk Latino youth with learning disabilities.
Low levels of school bonding appear to be linked to problem behaviors at school as well
as feelings of alienation and poor self-image for these students of color.

Low levels of school bonding appear to be connected to feelings of isolation for
other students as well. Another study with a sample of mostly Caucasian students
(Mouton, et al., 1996) also revealed that adolescents who had low bonding to school
described their lives in school as lonely and isolated and saw themselves as being
alienated from school community. They perceived a low level of support and
encouragement from the school personnel and from their peers.

Based on the evidence Shown in the above, it is clear that bonding to school plays
a significant role in the success or failure of students. Positive or strong bonding

increases positive outcomes in youth and therefore contributes to success of students.

17

Bonding to school promotes adolescents’ pro-social behaviors and academic achievement
in addition to contributing to their physical and mental health. Also, it is shown that
positive bonding to school increases students’ self-esteem and motivation. It acts as a
buffer against adolescents’ risk behaviors and is associated with reducing negative
student outcomes such as drug use, smoking, drinking, and sexual activities. Moreover,
bonding to school decreases the likelihood Of students’ involvement in anti-social
behaviors and is effective in preventing school violence. Youth who bond to school are
less vulnerable to negative peer influences and are less likely to engage in high-risk
behaviors.

Summary of Importance of School Bonding in Student Outcomes

While positive and strong bonding is associated with positive outcomes, weak or
low bonding to school is shown to increase negative outcomes and contribute to students’
failure. Evidence indicated that low school bonding is also a risk factor for adolescent
substance abuse and appear to be connected to feelings of isolation and poor self- concept
and low self- image for students.

The above literature also indicated that students’ bonding to school varies based
on students’ characteristics. There were some indications that school bonding may affect
female students differently than the male students. Also, students in higher grades
showed a lower bonding to school than those in lower grades. However, with regard to
race/ethnicity it was shown that school bonding might impact minority and other students
Similarly. In sum, bonding to school is identified to be the most convincing evidence of

a relationship between risk and problem behaviors during adolescence. Given the

18

importance of school bonding presented here, it is reasonable to explore factors that may
impact students’ bonding to schools.
Predictors of School Bonding

Given the significance of school bonding in predicting student outcomes, it is
important to understand what factors might help to promote school bonding. While
several variables such as parent-school bonding (Pryor, 1994), positive attachment to
parents (Anson, 1995), and school misbehavior (e.g., Bryant, et al., 2000) are referred to
as having an impact on school bonding, studies have indicated school climate as one of
the significant predictors of students’ bonding to school (e.g., McBride, et. al., 1995;
Anson, 1995; Kumpfer & Turner, I990-l991; Pyper, Freiberg, Ginsburg, & Spuck,
1987). Some authors (e.g., Anson, 1995; Hawkins, & Catalano, 1990; Schaps, &
Battistich, 1991) revealed that student-school bonding can be highly increased by
improving school climate, developing social skills and social competence and
encouraging authoritative parenting practices. Students, who like their classmates,
believe their teachers are supportive and fair, have good peer relations, and accept their
school’s mission, values and standards are more likely to bond to the school. Also,
student bonding to school may be improved by establishing school policies and practices
that are fair, increasing opportunities for students to participate in school activities and
cultivating positive culture in school (Gottfredson, 1986) that in turn affect students’
experience and perceptions of school climate.

Some authors have conducted correlation and regression studies to investigate the
relationship between school bonding and school climate. For example, Simons-Morton,

et a1. (1999) indicated a significant positive association between perceived school climate

19

and school bonding among a sample of middle school students. They also examined the
relationship between these variable and several other variables and concluded that
schools can improve students’ bonding by improving classroom climate and by
modifying school environment. In addition, Anson (1995) used hierarchical multiple
regression in examining the impact of school climate on students’ bonding in middle
schools. She reported that teacher support for achievement, caring behavior of non-
teaching staff, behavior patterns of students and students’ level of social comfort were
important contributors to middle school students’ bonding to school. Furthermore,
schools with larger populations had students with lower attachment scores.

Other researchers have developed models to examine the relationship between
school climate and school bonding. For example, McBride et al. (1995) investigated the
availability of social opportunities in the school environment as predictor of social
bonding among high school students by proposing a heuristic model. In their model they
tested the influence of the bonding environment on 9th grade students’ social bonding and
risk taking behaviors. Furthermore, aggregated levels of attachment, commitment and
involvement in school were used as indicators of the availability of bonding opportunities
in the school environment. Their results showed that pro-social bonding environments
were associated with a significant increase in bonding among ninth grade male and
female students and a reduced likelihood of their engagement in risk-taking behavior.
Kumpfer and Turner (1990-1991) also developed a model to explore the relationship
between school climate and school bonding. In their Social Ecology Model family and
school climate were depicted as two major ecological influences of school bonding.

Their findings confirmed that in a sample of high school students, school bonding was

20

significantly predicted by the latent variables of school and family climate. Students
with poor family or school climate did not appear to bond to their school. Furthermore,
they indicated that youth are more likely to choose pro-social peers if they are involved in
schools with positive climate.

Some researchers have compared students by race/ethnicity when looking at the
relationship between school climate and school bonding. Chase (2000) explored
differences in race/ethnicity in examining the relationship between school bonding and
school climate and noted that bonding to school may be impacted by individual’s
race/ethnicity. The author compared bonding of African American students to school
with that of the Caucasian students. It was found that African American students
exhibited distrustful feelings about individuals in their schools and indicated having a
negative perception of school climate, which contributed to their lack of positive bonding
to school. Kester (1994) conducted a study on a sample of 10 African American students
(five who had been with the same teacher for 2 years and five students who were new to
school). Interviewing these students, the author explored the effects of school structure
and peer group on students’ bonding to school. In particular, this study examined
whether students who have the same teacher for all three years of the middle school
develop a stronger connection to school or not. Results indicated that students in two
groups had different loyalty to their peer group and their homeroom teacher. Those who
had one teacher for a long period of time with whom they developed a close relationship
seemed to have greater bond to school than students who were new to school. Anson
(1995) compared African American, Hispanic, and Asian students with Caucasian

students and found that African American and Hispanic students felt less attached than

21

did their Caucasian and Asian peers. Furthermore, she indicated a link between higher
percentage of Afi'ican Americans in a school and a higher school bonding for African
American students.

Gender differences were also examined in the interactions between school climate
and school bonding (e.g., Kumpfer, & Turner, 1990-1991). The authors tested their
ecological model and found a better fit of the model for males than females. Anson
(1995) also examined gender differences when looking at school climate and school
bonding and found that females reported higher bonding to school.

In sum, school climate has been identified as one of the significant predictors of
students’ bonding to school. Studies have indicated that student-school bonding can be
highly increased by improving school climate. Research has shown that pro-social
bonding environments were associated with a significant increase in bonding. Also,
family and school climate were indicated as two major ecological influences that
predicted school bonding and their findings indicated that school bonding was
significantly predicted by the latent variable of school climate.

Previous literature examining the interactions between school climate and school
bonding has also explored differences among students by demographic variables of
race/ethnicity and gender. In examining differences on race/ethnicity it was indicated
that Afiican American students had lower bonding to school than Caucasian students due
to their negative perceptions of school climate and feelings of mistrust. Results on
gender differences showed that the model describing the relationship between school
climate and school bonding was significantly different for female students than for male

students.

22

Limitation of Previous Research

Despite the evidence revealing the importance of school climate on school
bonding, the number of studies focusing on the impact of school climate on students’
bonding to school is sparse and this area of research is not completely understood. A
literature search on the topic of school climate and school bonding since 1992 yielded
only 45 studies, of which, only a few closely related research were found addressing the
relationship between school climate and students’ bonding (e.g., Kumpfer, & Turner,
1991; Anson, 1995).

Moreover, pathways to school bonding are not well known and further research to
understand how school climate variables impact students’ bonding to school is needed.
In particular, how various aspects of school climate as perceived by students contribute to
school bonding has not received the attention it deserves. In addition, only a few studies
have focused on gender issues when considering the relationship between school climate
and school bonding. Only one had an ethnically diverse sample and the rest had included
samples with mostly Caucasian students. This limits the generalizability of the results to
other minority population. Also, differences among students by race/ethnicity
background or by school in relation to this topic are not well investigated. Only few
studies have addressed the race/minority differences and their samples have only
included African American and Caucasian population. Only one study made
comparisons with other minority students. Furthermore, only one study investigated the
differences by school and had indicated that schools with larger populations had students
with lower bonding to school. Comparison of schools is an important area to be

considered since schools with various demographic compositions and different school

23

climate create different bonding environment for students. This in turn impacts how

school climate affects students’ bonding to school.

Addressing Gaps in the Literature

More studies need to be conducted to explore the effects of gender, race/ethnicity
and school differences in examining school bonding and pathways of perceived school
climate to students’ bonding to school, particularly for schools in urban areas with high
percentage minority students. The reason for this claim is because evidence presented in
this document revealed a) the vulnerability of students during middle school years
especially the underserved population, b) the importance of school bonding in protecting
adolescents against risk factors, c) perceptions of school climate as one of the Significant
factors to be considered for improving students’ bonding to school, (I) that few studies
have investigated how school bonding varies by students’ gender, race/ethnicity, grade
level and school, and studies that have addressed these issues are limited in terms of their
sample or the comparisons they have made, e) that to date few studies have explored the
impact of school climate perceptions on school bonding, especially, among the middle
school population, in urban areas with high percentage of minority population, and t) that
only few studies have explored differences among students by gender, race/ethnicity and
school in examining the impact of school climate perceptions on school bonding.
Therefore, the present study attempts to close these existing gaps by shedding light on the

perceived school climate pathways to school bonding.

24

Research Hypotheses

The following research hypotheses will be addressed in this study:
Research Hypotheses Related to School Bonding

Hypothesis #1. There are significant differences between male students and
female students on their perceptions of school bonding.

The first hypothesis in this study will address gender differences on school
bonding. Previous research on gender differences was inconclusive. One study
(McDonald, & Wright, 2002) showed that female students who smoked had a lower
bonding to school than the non-smoker females and the male students in the sample.
Another study (Simons-Morton, et al., 1999) found a higher bonding to school for
females than the male students.

Hypothesis #2. There are significant differences between Caucasian students,
African American students and Hispanic/Latino students on their perceptions of school
bonding.

With respect to race/ethnicity, previous studies that had reported the racial
composition of their sample were conducted either on a sample of Latino or Hispanic
students (Storino, 2001; Robertson, Harding, & Morrison, 1998) or on a sample where
the majority of students were Caucasian (e.g., Mouton, et al., 1996). Only one study
(Anson, 1995) made a comparison between students from various racial backgrounds.
The following research hypothesis will add to previous knowledge by comparing school
bonding among a sample of middle school students from three various racial/ethnic
backgrounds: Black/Afi'ican American students, Mexican/Latino/Hispanic students and

White/Caucasian students.

25

Hypothesis #3. There are Significant differences between students in grades 6'“, 7th

and 8th on their perceptions of school bonding.

Previous research has found that perceptions of school bonding vary by grade
level (Mouton, et al., 1996; Simons-Morton, et al., 1999). However, only one of the two
studies was conducted on a sample of middle school students. Furthermore, this study
used an index of school bonding that measured only school commitment. They did not
include other dimensions of school bonding in their measurement. Therefore, the third
research hypotheses in this section will extend previous literature with regard to grade
level differences in middle school by considering measures of school bonding that
includes several constructs. Because this study will utilize multiple dimensions of school
bonding, and will be conducted on a sample of middle school students, it will contribute
to the research on school bonding by adding new information about how middle school
students in different grade levels vary on measures of school bonding.

Hypothesis # 4. There are significant differences between students in one of the
middle schools in the sample (school A) and students in the other middle school in the
sample (school B) on their perceptions of school bonding.

Gottfredson, (1986) indicated that student bonding to school is affected by the
culture, policies and practices of school. When these are positively defined, rules and
policies are fair, and opportunities for students to participate in school activities are
increased, students’ bonding to school can be improved. Previous research has also
shown that student-school bonding is impacted by school structure (Kester, 1994) and
school climate (McBride, et al., 1995; Kumpfer & Turner, 1990-1991). Although, these

studies have indicated the impact schools can have on students’ bonding to school, none

26

have made a comparison among schools on the measure of students’ bonding to school.
This hypothesis will extend the previous research by providing such a comparison.
Research Hypotheses Related to Perceived School Climate as Predictor of School
Bonding

Hypothesis # 5. There are significant pathways from perceived school climate to
middle school students’ bonding to school for the entire sample.

School climate has been identified as one of the significant predictors of students’
bonding to school (McBride, et. al., 1995; Anson, 1995; Kumpfer, & Turner, 1990-1991;
Pyper, Freiberg, Ginsburg, & Spuck, 1987). However, most of these studies had
considered a narrow aspect of school environment such as organization of classroom
(Kester, 1994) or bonding opportunities in the school environment (McBride et al., 1995)
in relation to school bonding in their studies. Two studies (Kumpfer & Turner, 1991;
Anson, 1995) examined several dimensions of school environment. Only one of the two
was done on a sample of middle school students. This research question will add to
previous studies by examining the impact of perceived school climate on middle school
students’ bonding to school by considering a broader range of school environment, some
of which are different from previous studies.

Hypothesis # 6. Pathways of perceived school climate to students’ bonding to
school are predicted to be different for male students than those for female students.

How students’ gender impacts students’ bonding to school is not clear. Previous
evidence concerning gender differences has been vague and unsettling. Kumpfer and
Turner (1990-1991) examined gender differences in relationship between school climate

and school bonding. They tested their ecological model and found a better fit of the

27

model for males than females. Anson (1995) also explored gender differences when
looking at school climate and school bonding and found that females reported higher
bonding to school than did male students. This hypothesis will provide further
information by examining the differences among male and female students in linking
school climate to school bonding.

Hypothesis # 7. Pathways of perceived school climate to students’ bonding to
school are predicted to be different for each subgroup of students as identified by
race/ethnicity: Black/African American students, Mexican/Latino/Hispanic students and
Caucasian students.

Only few studies have explored differences in race/ethnicity of students when
examining the relationship between perceived school climate and school bonding, and
comparisons were mostly made on differences between African American students and
Caucasian students. Of three studies on this topic, two (Chase, 2000; Kester, 1994)
compared the bonding of African American students to school with that of the Caucasian
students. Chase (2000) indicated that African American students lacked positive bonding
to school due to a negative perception of school climate and distrustful feelings towards
other individuals in their school. Kester (1994) used a very small sample (10 students)
and conducted an interview and a survey to explore the effects of school structure and
peer group on students’ bonding to school. It was found that African American students
had a different loyalty to their peer group and their homeroom teacher than the Caucasian
students. Only one study (Anson, 1995) included other minority groups (Hispanic and
Asian students) in addition to African American students in the sample. This study

Showed that African American and Hispanic students were less attached to their school.

28

The author used regression methodology to make the comparisons between different
groups. This hypothesis will add to previous research by employing a multivariate
technique (SEM analysis) and including a sample consist of three minority groups
(Black/African American students, Mexican/Latino/Hispanic students, and
White/Caucasian students) to examine if the pathways of perceived school climate to
school bonding remain the same or not for these three subgroups.

Hypothesis # 8. Pathways of perceived school climate to students’ bonding to
school are significantly different for each subgroup of students as identified by school
variable: School A, School B.

With regard to schools, each school has a different climate unique to itself as
determined by many factors that impact its climate (Rutter, 1979). This can be impacted
by measures of socio—economic background of students attending the school and other
student demographic variables as well as school level variables. These factors determine
the climate of each school, which in turn impact students’ perceptions of school climate
at the school level (Rutter, 1979) and student outcomes (Smey-Richman, 2001; Wang,
2000). Similarly, students’ bonding to school may vary depending on the school
composition such as percentage of minority students in school (Anson, 1995). Therefore,
it is expected that the two schools in this sample would be different with regard to

perceived school climate variables that link to students’ bonding to school.

29

CHAPTER 3

METHODOLOGY

Participant Recruitment

This study used secondary data (from wave 4 of 4), collected for a larger project,
the “Michigan Model Partnership for Character Education” (MMPCE), and recruited the
participants used in the analysis from two middle schools in the sample. In that larger
study, a convenience sample of three school districts was recruited for their participation
in a two-year study. One of the middle schools in this study was selected as a control
group for the larger project while member was considered for a full implementation of
the intervention To recruit the schools, the district superintendent of each school was
contacted and asked to select two fairly comparable schools within their district to
participate. One school (the experimental or intervention group) would receive
comprehensive progranuning and additional funds (school climate and Character
Education interventions) while the other school (the control group) would receive
Character Education (CE) programming alone and lesser funds. Each superintendent
assigned a convenience sample of two schools to participate in the study. The school
principals selected the classrooms and teachers that would participate in the CE
programs. Principals were asked to coordinate within school districts to provide
comparable CE programming. Attempts were made to coordinate the delivery of the CE
curriculum in terms of which grade level would be targeted for the CE curriculum and in

which class period it would be delivered.

30

Participants

The sample in this research includes students in two middle schools from an
urban district in Michigan with high level of poverty and a high percentage of minorities
(drawn from the larger study—MMPCE as stated above). Table 3.1 below shows the
profile of the community and middle school students in the district from which the two
schools in the sample were selected at the beginning of the project. A majority of
students in the district were eligible for free and reduced lunch. Also, a majority of
students scored below the passing rate on MEAP scores across subjects and their absence
rate was high.
Table 3.1

Characteristics of the Community and Middle School Students in the District’

 

Percent Reported
Community characteristics of school district

 

Percent of population residing in school district that 38
is non-white alone

Percent of population residing in school district in owner 60
occupied housing

Percent of adults in the community with low SES 75

Characteristics of students in middle schools in the district

Percent students qualified for Free/reduced lunch 76
Percent Absent more than 12 days per year 45
MEAP passing rate in Reading 38
MEAP passing rate in Math 34
MEAP passing rate in Writing 48
MEAP passing rate in Science 44
Percent Caucasian 34
Percent African American 44
Percent Hispanic 19
Percent Asian 2
Percent Native American 1

 

 

I NCES-CCD Based on 2000 Census * Character Education Report 2001

31

Table 3.2 below indicates the profile of the overall schools at the building level at
the beginning of the project. The schools from which the sample population was drawn
were similar on levels of economic advantage, academic achievement, gender
composition, and different on racial composition of students, and staffing ratio. Both
schools had ethnically diverse populations with a large percentage of students from both
African-American and Caucasian backgrounds. However school B had a considerable
larger population of African American students than the other.

Table 3.2

Characteristics of Participating Schools During 2000 to 20012

 

 

 

School A School B
Student demographics
Percent of students economically disadvantaged 73 77
Student Achievement Dat_a
MEAP passing rate 18 25
MEAP participation rate 60 68
Staffing Ratio
Student per teacher 30 14
Gender
Female 48 48
Male 52 52
Race/Ethnicity
Caucasian 45 22
African American 35 68
Hispanic 17
Native American 2 2
Asian Pacific Islander 1 2

 

 

2 Sources of data: NC ES-CC D & Standard and Poors 2001

32

Of approximately 1,272 students in grades 6th through 8th from the two middles
schools, 763 students completed the questionnaires for this study. The background
characteristics of students in the research sample are indicated in Table 3.3 below. Based
on the data from students who filled out the surveys, the proportion of male students is
slightly higher than the females in the sample. While there are more 7th graders in the
sample, equal proportion of students are in 6th and 8th grades. With regard to
race/ethnicity the sample is diverse and the majority of the students are African
American, Caucasian and Hispanic. Data also shows that the two schools are similar
with regard to the proportion of male and female students as well as the distribution of
students in each grade. However, with respect to race/ethnicity, majority of the students
in school B are African Americans while in school B the majority are Caucasians.

Table 3.3

Demographic characteristics of the participants for the entire sample and by school

 

 

Entire Sample School A School B
N 0/03 N % N %
Gender
Female 350 46 186 44 163 48
Male 408 54 233 56 174 52
Not reported 5 --- 2 --- 3 ---
Total 763 100 421 100 340 100
Grade
6‘“ 235 31 130 32 105 31
7th 283 38 163 40 120 36
8th 230 31 118 29 112 33
Not reported 15 --- 10 --- 3 ---
Total 763 100 421 100 340 100

 

 

3 The number of cases that did not report is not considered in calculation of percentage in this table.

33

Race/Ethnicity

Caucasian 213 30 162 40 51 17
African American 270 38 93 23 177 58
Hispanic 104 15 57 14 47 16
Asian 9 1 7 2 2 1
American Indian 14 2 8 2 6 2
or Alaska Native
Pacific Islander 0 0 0 0 0 0
Other 105 15 84 20 21 7
Not reported 47 --- 9 ---- 36 ---
Total 763 100 421 100 340 100

 

Data Collection Procedure

Students in both middle schools were given self-administered paper and pencil
surveys at the end of the 2003-2004 school year. This survey included items assessing
individual student’s perception of school climate, student’s perception of bonding to
school as well as their demographic data on gender, race/ethnicity, and grade level.

Students were presented with forty-three statements and were asked to rate the
degree to which each sentence described what was occurring in their school. This survey
was administered to students during their class period as decided by their teachers in
agreement with the school principal. It took approximately twenty minutes to complete
the survey. Students were given an option to either participate in completing the survey
or to refuse. Those who chose to participate were given instructions as to how to respond
to the items (usually read to them by their teacher). For confidentiality purposes students
were provided with envelopes to place their completed surveys, or empty surveys if they
refused to participate. Prior to implementation of the larger project (MMPCE), teachers

had received training, as to how to administer the survey to their students, during a one

34

day in-service at the beginning of the school year. In addition, instruction sheets
regarding the implementation of the survey were made available to them at the time of

training for future references.

Measurements

School Bonding

Latent construct of school bonding is the outcome or dependent variable in this
study. This construct will be assessed by eighteen items from the Survey of School
Climate (Anson, 1995) measuring four domains on a five point Likert scale ranging from
none to all (1: None or Not at All for some questions to 5= All or Very for some
questions). A description of the four domains, number of items, examples of the items,
and the reliability data reported in a previous study (Anson, 1995) are as follows (see
Appendix B for the full description of items):

1. Personal attachment to school. Refers to student’s enjoyment and seeing contact
with school, its inhabitants and its environment including 6 items with alpha
reliability of 0.87.

(Examples of items: If you had to stop going to school, how much would you
miss your teachers?; If you had to stop going to school, how much would you
miss the principal?; If you had to stop going to school, how much would you
miss the sense of school spirit?)

2. Pride in school. Refers to student’s report on feelings of pride for both themselves

and as they perceive it in others with 6 items and alpha reliability of 0.83.

35

(Examples of items: How proud do you feel of your school?; How many students
feel proud of this school?; How many teachers feel proud of this school?)

3. Fairness of school rules/punishment. Highlights the importance early adolescents
place on feeling that adults are being fair and that policies are carried out in a just
way. This is an important construct in measuring early adolescents’ positive
feelings towards school and includes 3 items with alpha reliability of 0.70.
(Examples of items: How clearly stated are the school rules?; How fair are the
school rules?; When students are punished at school, how fair is the punishment?)

4. Personal dislike of school. Contains items indicating each student’s feelings about
school or going to school. Reverse coded, a high value indicates positive feelings
towards school and it includes 3 items with alpha reliability of 0.63.

(Examples of items: When I wake up in the morning, I often don’t feel like going

to school; School is boring; sometimes I pretend to be sick so I don’t have to go to

school).
School Climate

In this study climate constructs are measured by the School as a Caring
Community Profile —11 instrument (SCCP-II; Center for 4th and 5th R’s, 2000), developed
to help schools assess themselves as a caring community, and measure students’
perceptions of climate of the school. This 45-item instrument includes five scales (See
Appendix A for the full description of items) that are arranged on a five point Likert scale
from “Almost Never” to “Almost Always” (l=almost never, 2=sometimes, 3=as often as

not 4=frequently, 5=almost always). The five scales are listed below:

36

l.

Perceptions of student respect. Measures respect for other students, staff and
school property and includes nine items with the reported reliability of 0.88.
(Examples of items: Students treat classmates with respect; Students respect
the personal property of others; Students refrain from put-downs).

Perceptions of student friendship and belonging. Measures student caring,
respect, and kindness for one another including 9 items with a reported
reliability of 0.87.

(Examples of items: Students exclude other students because they are
different; Students help each other, even if they are not friends; Students work
well together).

Perceptions of students’ shaping of their environment. Measures students’
attempts to influence the behavior and actions of others and the overall well-
being of the school which includes 7 items with the reported reliability of
0.90.

(Examples of items: Students help to improve the school; Students try to have

a positive influence on the behavior of other students; Students resolve

. conflicts without fighting, insults, or threats).

Perceptions of support and care by and for faculty/staff. Measures caring and
respect students, parents and other faculty, staff, have toward faculty and staff.
It includes 10 items with reported reliability of 0.82.

(Examples of items: Teachers go out of their way to help students who need

extra help; Students can talk to their teachers about problems that are

37

bothering them; Teachers go out of their way to help students who need extra
help).

5. Perceptions of support and care by and for parents. Measures the caring and
respect that the school, teachers, students, and other parents Show parents
including 7 items with reported reliability of 0.73.

(Examples of items: Parents Show that they care about their child’s education
and school behavior; Teachers treat parents with respect; this school cares

about the thoughts and feelings of parents).

Demographic Variables
Student’s grade, gender, and race/ethnicity, school as well as background
variables of educational aspiration, self-reported grades obtained in school, are included
in this study. Categorical variables of gender, race/ethnicity and school are coded as
follow: Gender, 1=female, 2=male; Race, 1=White/Caucasian, 2=Black/African
American, 3=Mexican/Latino/Hispanic, and; School, l=middle school A, 2=middle

school B.

38

CHAPTER 4

ANALYSIS

This section includes two sets Of analyses. First, the background analyses are
presented including descriptive analysis of both dependent and independent variables,
and correlation matrix for dependent and independent variables. Secondly, Multivariate
Analysis of Covariance (MANCOVA) and Structural Equation Modeling (SEM) analyses

are presented which include testing of all research hypotheses in this study.

Background Analyses
The analyses plans in this section include preliminary analyses. Their purpose is
to provide background information concerning the measures and variables that are

included in this study and are not performed for testing of the hypotheses.

Descriptive Analyses

Descriptive analyses are utilized to report the mean, standard deviation and the
number of students rating each measure of the dependent and independent variables in
this study. These results are indicated in Tables 4.1. Both school climate and school
bonding items were scored on a range from 1 to 5 with 5 indicating high scores.
School Climate Descriptive Analysis

Results for the school climate measures Showed that perception of support and
care by and for parents and those for faculty/staff had mean values slightly above 3,
revealing that all together, students perceived these two aspects of the climate in their

school as slightly above average. However, the values for measures of student respect,

39

student friendship and belonging, and student shaping of their environment were between
2.3 to 2.6 indicating that overall students perceived these aspects of school climate as
slightly below average (See Table 4.1).
School Bonding Descriptive Analysis

Similarly, for the school bonding variables the mean values for student attachment
to school, pride in school and personal dislike of school were around 2.6 showing that on
the average students reported these items as Slightly below average, while the mean
ratings for fairness of school rules/punishment was around 3.0 which shows students
rated the items for this measure as average (See Table 4.1).
Table 4.1

Distribution of School Climate and School Bondingfor the Sample

 

 

Variable Mean SD N
School Climate

Perceptions of Student Respect 2.61 0.64 715
Perceptions of student friendship and belonging 2.63 0.60 683
Perceptions of students’ shaping of their environment 2.31 0.74 731
Perceptions of support and care by and for faculty/staff 3.11 0.85 706
Perceptions of support and care by and for parents 3.33 0.86 685
School Bonding

Personal Attachment to School 2.61 0.96 697
Pride in School 2.68 0.83 708
Fairness of School Rules/Punishment 3.03 0.95 687
Personal Dislike of School 2.62 0.95 733

 

Correlation Matrix
Spearman rank correlations of all research variables are Shown in Table 4.2.
Means and standard deviations are presented in the first columns and reliabilities of the

measures are indicated on the diagonal.

4O

Overall, the independent variables representing school climate had moderate to
high correlation with all four constructs of the dependent variable and they were all
statistically significant (p <0.01). Two independent variables of perceptions of support
and care by and for faculty/staff and perceptions of support and care by and for parents
had a relatively moderate to high correlation with the four dependent variables.
Perception of student respect, perceptions of student friendship and belonging as well as
perceptions of students’ Shaping of their environment had moderate correlations with the
school bonding variables of personal attachment to school and pride in school, and low-
moderate correlations with fairness of school rules and punishment and personal dislike
of school. The highest correlation between dependent and independent variables was
found among fairness of school rules/punishment and perceptions of support and care by
and for faculty/staff, and the lowest correlation existed between fairness of school
rules/punishment and perceptions of student respect.

An examination of correlations among the school climate variables in Table 4.2
indicates that most of the school climate variables had moderate to high correlations with
one another, although a few had low correlations. All of these measures had positive
correlations with one another. The highest correlation was found between perceptions of
support and care by and for parents and perceptions of support and care by and for
faculty. Students’ perception of friendship and belonging was also highly correlated with
perceptions of student respect. High correlation also was found between perceptions of
students’ shaping of their environment and perception of student respect as well as
perception of student friendship and belonging. The lowest correlation existed between

perceptions of support and care by and for parents and perceptions of student respect.

41

Perception of student respect also had low correlation with perception of support and care
by and for faculty/staff. In addition, low correlation was found (a) between perceptions
of support and care by and for parents and (b) perception of students’ Shaping of their

environment.

42

.85me 2: :o :39? 2a moist? on: Co 82:23:: 98 mumofieea E @8538 8a :37. 29:5 _

 

 

 

:::. AmAc: Aaoc: Ahxe: Aooo: A::6: A:_A: Amcov Am:::
:3: :52: :am: New: :32 :33 :22 :3: 3.: S: 328 .8 8:25 3:88: .o
2.2: :3: is $3: $68 83: a2: zeoefieieoax
N50 .1;ch 3."?de singed aim—WC *nmvmd nimmmd aim—Nd mod MOM Bozom mo 3.0:;an .x
Amnc: Lowe: Aaeov Acme: A486: knee:
9.: 1:9: 23.4.: 3:4,: 1:3: 3.3:: 1:3: Q: :0: 623 E :25 s
Aomc: Ammo: Ammo: A:m:: Acco:
Nwd eucmmd aim—v.3 *svwmd enemmd uscomd cod _c.N Bozum Cu :5:.-_oa:< fission— .c
$2: $9,: Ammo: 84:: 2488.5: 25 .3 .
3.: 3:3,: 1:3: :32: 3:: _ : cm: 2.». 88 ea: :89; :o 25:86.5: m
5...: 5.3 an: :saéae 8: as E
3.: 2.8:: :8»: L8,: 5...: :a 88 :5 28:3 Co 82585: .4
:9: .32 565855 :2: :6
inc .1390 12m: 3.: 3%.». means: .35me no EEEOOSL .m
is wagon: :5.
S: :22 :c: a: 35.22:282338888: .N
cod wed EN Somme. Eon—cm? mcocaoocom ._
a. a A. o w v m u _ am :32

 

_83:t:> ganged >3: :0 55:2 55650
We :35

Reliability Analysis

Reliabilities of the measures for the sample in this study are calculated and
reported in Table 4.3 below. As indicated, reliability of the items for all constructs of
school climate and school bonding in this study are between 0.69 and 0.84. Among the
measures of school climate, perceptions of support and care by and for faculty/staff,
perceptions of student shaping of their environment, and perceptions of support and care
by and for parents had the highest reliabilities while for school bonding domains, pride in
School, personal attachment to school, and fairness of school rules and punishment had
the highest reliabilities. Similarly, the school bonding constructs had reliabilities
between 0.60 and 0.83 for the sample in this study. Pride in school had the highest
reliability while personal dislike of school had the lowest reliability among the school
bonding measures.
Table 4.3

Comparison of Scale Reliability for the School Climate and School Bonding Variables
Based on the Study Sample to Previously Reported Studies

 

 

Standardized
Scale Items No. of Item alpha Standardized
Items (Previously Item alpha
reported for this study)
School Climate
Perceptions of Student Respect 9 0.88 0.69
Perceptions of Student Friendship & Belonging 9 0.87 0.69
Perceptions of Student Shaping of their Environment 7 0.90 0.77
Perceptions of Support and Care by and for Faculty/Staff 10 0.82 0.84
Perceptions of Support and Care by and for Parents 7 0.73 0.76
School Bonding
Personal Attachment to School 6 0.87 0.82
Pride in School 6 0.83 0.83
Fairness of School Rules and Punishment 3 0.70 0.72
Personal Dislike of School 3 0.63 0.60

 

The reliabilities found in the study were comparable to those reported by the
measurement authors. Compared to the reliability of school climate measures reported by
the instrument developer (Center for 4‘h and 5th R’s, 2000), item reliabilities based on the
sample in the present study are higher for two measures (i.e., support and care by and for
faculty/staff; support and care by and for parents) and lower for three measures (i.e.,
student respect, student friendship and belonging, and student shaping of their
environment) than those reported previously.

Similarly, the measures of school bonding for the sample in this study were
comparable to those reported previously (Anson, 1995). Personal attachment to school
and personal dislike of school had slightly lower reliabilities than those reported
previously. Fairness of school rules and punishment had a slightly higher reliability than
the prior study while the reliability of pride for school was exactly the same for both
samples.

Testing Research Hypotheses

In this section two sets of analyses will be conducted to test the research
hypotheses. The first analyses examine potential group differences on the key research
outcome variable of school bonding. The second analyses examine the relationships
between the school climate predictor variables and the school bonding outcome variables
for different groups of students.

Analysis of Group Differences

To examine the research hypotheses (1 through 4) relevant to group differences

on indicators of school bonding, several Multivariate Analysis of Covariance

(MANCOVA) were conducted. These analyses Simultaneously examined the

45

relationship of a single student background characteristic on all the school bonding
outcomes, while controlling for the remaining student background characteristics. The
MANCOVA test is based on the assumption that the covariates have linear relationships
with the outcome variables. Therefore, preliminary analyses (not shown here) were
conducted to first determine whether each of the proposed covariates demonstrated a
linear relationship to the outcome variables. While existence of a linear relationship is
generally best demonstrated by scatter plots of the variables, here scatter plots did not
seem to be the best detenninant of linear relationships between covariates and the
outcome variables. This was due to the fact that both the covariates and the outcome
variables were ordinal variables ranging from values of I to 5. Therefore, Spearman rank
correlation coefficients were used to determine linearity between covariates and the
outcomes (personal attachment to school, pride in school, fairness of school
rules/punishment and personal dislike of school). Only those background variables that
had significant linear relationships with at least three of the four dependent variables.
Using this decision rule, all of the initially proposed student background variables were
used in MANCOVA analyses because all met the assumption of linearity. These
proposed background variables included students’ gender, grade level, self-reported
grade obtained in school, and educational aspirations. The MANCOVA results for

hypotheses 1 through 4 are reported in Tables 4.4 through 4.31.

46

Interpretation of MANC 0 VA Tables

The MANCOVA analysis involved four steps once the assumptions of linearity
were tested. (1) The omnibus test was conducted to determine if there were any
significant differences between groups of students (e.g. by race, gender, school and grade
level) on the entire set of school bonding scales once student background characteristics
were controlled. Background characteristics included race, gender, grade level, self-
reported grades, and educational aspirations. Wilks’ lambda was used for this omnibus
test since the assumptions appeared to be met. The Wilks’ lambda test has been described
as a good balance between power and assumptions when assumptions are met (Olson,
1973). (2) If the Wilks’ 1amda was significant, indicating group differences did occur,
then the second step was to conduct a univariate analysis to determine on which of the
school bonding scales significant group differences occurred. This analysis yielded an F
score which was again examined for significance. A significant F value for a particular
scale indicated that group differences existed on that scale. (3) For those scales in which
group differences were present, the mean values of the scales were examined for each
group to determine which group had a significantly higher value than the other group. (4)
When more than two groups of students were analyzed for mean values on their outcome
variables, additional post-hoe contrasts were needed to determine which of the student
groups differed from each other on a particular measure.
Presentation of MANCO VA Analyses

For each of the following four hypotheses one table will be presented. The results
of the multivariate analysis (described in step one above) will be presented in the text.

The key information for the results of the above steps two, three and four (if applicable)

47

will be presented in one table. The full details of each step in the analysis will be
presented in the appendices.
Hypothesis #1

There are significant differences between male students and female students on
their perceptions of school bonding.

For this hypothesis 3 between-subjects MANCOVA was performed on four
dependent variables: (1) personal attachment to school, (2) pride in school, (3) fairness of
school rules/ punishment, and (4) personal dislike of school. The results were controlled
for students’ (1) grade level, (2) self-reported grade obtained in school, and (3)
educational aspirations which had all been previously tested and found to have met the
assumption of linearity. That is, the model tested the hypothesis that males and females
did not differ on the four variables of school bonding after accounting for differential
effects of the covariates.

Multivariate tests revealed a significant main effect for gender (Wilks’s lambda =
0.984, E = 2.74 (4, 676), p = .028). The univariate tests (see Appendix C for full details
of this analysis) showed that the differences between males and females were significant
for only one of the four measures of school bonding: fairness of school rules/punishment
(E = 10.87, p= 0.00). No significant differences were observed between male and female
students for measures of personal attachment to school, pride in school or personal dislike
of school. Furthermore, the female students had higher average than their male
counterparts on all four measures of school bonding, although only one of the differences
was significant. Differences between males and females on measures of school bonding

are indicated in Table 4.4.

48

Table 4.4

Comparison of School Bonding Measures (Scales) Mean Values by Gender

 

 

 

Descriptive Statistics Univariate
test name
Variable Group Mean SD n E
(cl_f=l,679)
Personal Attachment Male 2.54 0.98 375 1.90
to school
Female 2.68 0.88 309
Pride in School Male 2.62 0.83 375 2.46
Female 2.75 0.79 309
Fairness of Male 2.92 0.93 375 10.87““
Rules/Punishment
Female 3.17 0.85 309
Personal Dislike of Male 2.57 0.95 375 0.83
School
Female 2.68 0.89 309

 

* p<0.05 ** p<0.01
Hypothesis #2

There are significant differences between Caucasian students, African American
students and Hispanic/Latino students on their perceptions of school bonding.

For hypotheses 2, Multivariate one-way Analysis of Covariance (MANCOVA)
was conducted to detect group differences on indicators of school bonding by students’
race/ethnicity (African American and Caucasian, Hispanic), controlling for several
students’ background characteristics. The between-subjects MANCOVA was performed
on four dependent variables: (1) personal attachment to school, (2) pride in school, (3)

fairness of school rules/ punishment, and (4) personal dislike of school. The results were

49

controlled for students’ (I) grade level, (2) gender (3) self-reported grade obtained in
school, and (3) educational aspirations which had all been previously tested and found to
have met the assumption of linearity. The model tests the hypothesis that African
American, Caucasian, and Hispanic students did not differ on the four variables of school
bonding after accounting for differential effects of the covariates. That is, students’ grade
level, gender, self-reported grade obtained in school, and aspirations were entered as

covariates in the model. Summary results for this hypothesis are shown in Table 4.5.

Table 4.5
Comparison of School Bonding Measures (Scales) Mean Values by Race

 

 

 

Descriptive Statistics Univariate Significant
test name Post hoc
Variable ‘ Group Mean SD 6 £12, 557) ”was”
Personal 1 African 2.70 0.96 255
Attachment to American 3.74: 1.3“
school 2 Hispanic/ 2.55 0.97 99
Latino
3 Caucasian 2.47 0.92 210
Pride in 1 African 2.68 0.77 255
School American 1 .88
2 Hispanic/ 2.75 0.89 99
Latino
3 Caucasian 2.61 0.81 210
Fairness of 1 African 3.02 0.85 255
Rules/Punish American 0.03
ment 2 Hispanic/ 3.05 0.88 99
Latino
3 Caucasian 3.06 0.96 210
Personal 1 African 2.73 0.88 255
Dislike of American
School 2 Hispanic/ 2.66 0.90 99 6.22“ :3:
Latino '3

3 Caucasian 2.47 0.93 210

 

* p<0.05 ** p_<0.01 ***p<0.001

Results of the multivariate tests indicated that overall significant differences

existed among the sub-groups by race/ethnicity on measures of school bonding when

50

student background characteristics were controlled (Wilks’ lambda =0.96,

E=2.91(8,1 108), p=0.003). The univariate tests (see Appendix C for full details) revealed
that the sub-groups were significantly different on two of the four measures of school
bonding. These were personal attachment to school (E=3.74, p=0.024), and personal of
dislike of school (£=6.22, p=0.002). There were no significant differences between
students by race/ethnicity on variables of pride in school or fairness of school
rules/punishment.

Post-hoe contrasts among the three sub-groups, indicated that when African
American students were compared to Caucasian students significant differences were
found on measures of personal attachment to school (p= 0.006) and personal dislike of
school (p=0.001) (See Appendix C for full details of this analysis). Afiican American
students had a higher average on both of these measures than did the Caucasian students
as indicated in Table 4.5. No significant differences were found between these two sub-
samples on measures of pride in school or fairness of school rules/punishment.
Differences between Hispanic/Latino students and their Caucasian counterparts were
significant only on measure of personal dislike of school (p=0.026) and the average was
higher for Hispanic/Latino students. Finally, Hispanic/Latino students and Afiican
Americans did not Significantly differ on any of the measures of school bonding.
Hypothesis #3

There are significant differences between students in grades 6‘”, 7th and 8th on
their perceptions of school bonding.

A Multivariate One-Way Analysis of Covariance (MANCOVA) was employed to

examine group differences on indicators of school bonding by students’ grade level,

51

while controlling for students’ background characteristics as covariates in the model.
The between-subjects MANCOVA was performed on four dependent variables: (1)
personal attachment to school, (2) pride in school, (3) fairness of school rules/
punishment, and (4) personal dislike of school. The results were controlled for students’
(1) gender (2) self-reported grade obtained in school, and (3) educational aspirations
which had all been previously tested and found to have met the assumption of linearity.
The model tests the hypothesis that African American, Caucasian, and Hispanic students
did not differ on the four variables of school bonding after accounting for differential
effects of the covariates. Results for this hypothesis are shown in the table below (Table

4.6)

52

Table 4.6

Comparison of School Bonding Measures (Scales) Mean Values by Grade Level

 

Descriptive Statistics Univariate Significant
test name Post hoc

 

 

Variable Group Mean SD n _F_(2, 678) contrasts
Personal 1 6th grade 2.80 0.94 212
Attachment to 1-2**
th *al:
school 2 7 grade 2.56 0.96 262 7.67 1-3***
3 8h grade 2.45 0.86 210
Pride in I 6th grade 2.95 0.78 212
School m l7.08*** 1-2***
2 7 grade 2.60 0.85 262
1_3***
3 8h grade 2.52 0.74 210
Fairness of 1 6‘h grade 3.21 0.92 212
Rules/Punish [h 5.81" 1-2“
ment 2 7 grade 2.96 0.89 262
1_3**
3 8h grade 2.94 0.88 210
Personal 1 6h grade 2.79 0.91 212
Dislikeof 4.73** 12”
School 2 7‘h grade 2.55 0.92 262 l 3"

3 8‘“grade 2.54 0.93 210

 

* p<0.05 ** p<0.01 *** p<0.001

Results of the multivariate tests indicated that overall significant differences
existed among the sub-groups by grade level on measures of school bonding when
controlling for students’ background characteristics Results of the multivariate tests
indicated that overall significant differences existed among the sub-groups by
race/ethnicity on measures of school bonding (Wilks’ lambda =0.96, E=2.91(8,1108),

9:0.003).

53

Results of the univariate tests (see Appendix C for full details) showed that
students’ grade level were significant for all four measures of school bonding. The post-
hoc contrasts (see Appendix C for full details) comparing students by grade level showed
that those in sixth grade were significantly different on all measures of bonding to school
than their counterparts in seventh grade (p=0.007, 0.000, 0.004 and 0.008 for variables of
personal attachment to school, pride in school, fairness of school rules/punishment and
personal dislike of school respectively). Also significant differences were found on all
measures of bonding between students in sixth grade and those in eighth grade (p=0.000,
0.000, 0.003, and 0.007 for variables of personal attachment to school, pride in school,
fairness of school rules/punishment and personal dislike of school respectively).
However, seventh and eighth graders were not significantly different on any of the
measures of bonding to school.

Comparison of the mean value for measures of school bonding (Table 4.6)
indicated that sixth graders had a higher score on all variables of school bonding than the
seventh or eighth graders.

Hypothesis # 4

There are significant differences between students in one of the middle schools in
the sample (school A) and students in the other middle school in the sample (school B) on
their perceptions of school bonding.

For this hypothesis, a two group between-subjects MANCOVA was performed on
four dependent variables, controlling for students’ grade level, gender, self-reported
grade obtained in school, and aspirations as covariates. The analysis was performed on

four dependent variables: (1) personal attachment to school, (2) pride in school, (3)

54

fairness of school rules/ punishment, and (4) personal dislike of school. The results were
controlled for students’ (I) grade level, (2) gender (3) self-reported grade obtained in
school, and (4) educational aspirations which had all been previously tested and found to
have met the assumption of linearity. The analysis tests the hypothesis that students in
school A are not different from their counterparts in school B on the four measures of
school bonding. Results are presented below in Table 4.7.

Table 4.7

Comparison of School Bonding Measures (Scales) Mean Values by School

 

 

 

 

 

 

Descriptive Statistics Univariate test
name
Variable Group Mean SD n E (I , 678)
Personal School A 2.57 0.94 398 1.45
Attachment to
““01 SChOOI B 2.64 0.93 286
Pride in School School A 2.77 0.81 398
10**
School B 2.55 0.80 286
Fairness of School A 2.99 0.92 398
Rules/Punishment 233
School B 3.08 0.88 286
Personal Dislike of School A 2.56 0.92 398
School 591“
School B 2.70 0.93 286

 

* p<0.05 ** p_<0.01 *** p<0.001

Results of the multivariate test indicated that significant differences existed
between students in school A and those in school B on measures of school bonding
(Wilks’s lambda = 0.938, E(4,675) = 11.15, 9 =0.000). Results of the univariate tests

(see Appendix C for full details) Showed that students in school A were significantly

55

different than those in school B on pride in school and personal dislike of school (See
Table 4.7). The two schools were not significantly different on personal attachment to
school or fairness of school rules/punishment.

Comparison of the mean values for measures of school bonding between the two
schools indicated that the average for students in school A was higher on measure of
pride in school than the average for students in school B. However, for personal dislike
of school, students in school B had a higher mean value than their counterparts in school
A. These differences are Shown in Table 4.7.

Research hypotheses (5 through 8) are related to the impact of school climate on
students’ bonding to school. Each hypothesis tests the same model, but on different
samples. The model hypothesizes that school climate variables predict school bonding
variables. This model was tested for the overall sample and again separately for

subgroups of students based on grade level, gender, and race.

Structural Equation Modeling (SEM) Analysis

Structural Equation Modeling (SEM) was conducted to test each hypothesis. The
best fitting model was determined in several steps: (I) The initial hypothesized model
was tested. Chi-square value and goodness of fit statistics were examined to determine if
revisions to the model were needed. (2) The magnitude of path coefficients and their
significance levels in the initial hypothesized model were examined. Paths with the
lowest coefficients were removed from the model in a step-wise fashion and the revised
model was analyzed. (3) Modification indices were also simultaneously examined to

determine if the chi-square value of the model could be improved with the addition of

56

pathways between variables not already linked. (4) At each step in the addition or
subtraction of pathways, the chi-square value and goodness of fit statistics were
examined. The initial model was revised until the chi-square value and goodness of fit
statistics could not be improved.
Interpretation of the SEM Analyses

Chi-square values and goodness of fit statistics were used to interpret the SEM
results. A significant chi-square indicated that the hypothesized model was significantly
different than the actual relationships present in the data. The goodness of fit statistics
examined including the Bentler-Bonnett Normed Fit Index (NFI; Bentler & Bonett,
1980), CF] and RMSEA. While the scale of the fit for NFI indices are not easily
interpretable and experience is required to establish values of indices that are related to
various degrees of meaningfulness of results, authors of NF I indices suggest that models
with overall fit indices of less than 0.9 can usually be improved substantially (Bentler &
Bonett, 1980). A value of 0.05 or less for RMSEA indicates a close fit of the model. A
value of 0.1 or more indicates a lack of fit of the model.
Presentation of SEM Analyses

The SEM analyses are presented as follows: (1) The graphic of the initial
hypothesized model for each hypothesis or sub-group are presented with coefficients
represented on the pathways. (2) The statistics associated with the hypothesized model,
the chi-square value of the overall model, the goodness of fit statistics for the overall
model, the statistics related to path coefficients, and significance levels of the path
coefficients are presented. (3) The hypothesized model is modified based on the

Significance levels of the path coefficients and the modification indices, and results are

57

presented. For hypotheses related to sub-groups (hypotheses 6-8), the best-fitting model
for one sub-group is examined against the other sub-group (s) in order to ascertain
whether group differences exist or not. If findings support that sub-groups are different,
the best-fitting model for the other sub-group (s) are determined and results are presented.
(4) The graphic of the final revised model are presented with coefficients represented on
the pathways. (5) The chi-square value of the overall model and the goodness of fit
statistics for the final model are presented and compared to the hypothesized model. The
statistics associated with the final revised model, including path coefficients and the
significance levels of the path coefficients are presented. (6) Finally, the sub-groups are
compared based on their chi-square values and their goodness of fit statistics and results
are indicated.
Initial Hypothesized Model for all Analyses

The initial model (Figure 1) shows the hypothesized pathways that link the
indicators of students’ perceptions of school climate to the indicators of students’
perceptions of school bonding. The levels of five school climate scales (Student Respect,
“RES”; Friendship and belonging, “FB”; Shaping of the environment, “Env”; Support
and Care for and by faculty and staff, “SF”; Support and Care for and by parents,” SP”)
are expected to contribute to the students’ levels of four dimensions of school bonding
(Personal dislike of school, “Dis”; F aimess of school rules/punishment, “Rul”; Pride in
school, “Fri”; and Personal attachment to school, “Att”). In each of these hypothesized
models each indicator of the observed exogenous variable (school climate) has a direct

path to each indicator of the observed endogenous variable (school bonding). Also, each

58

of the five indicators of school climate is assumed to be correlated with one another as
indicated by two headed arrows in the model.

These models will be analyzed using the full sample of students and then again
separately for subgroups of the sample by race, gender, and school. The subgroup
analyses are based on the hypothesis that students’ race, gender, and the middle school
they attended would affect the pathways of school climate that leads to school bonding
outcomes.

Figure 4.1

Initial Hypothesized SEM Model for All Analyses

_ é ENV \"’

 

Key for Abbreviated Terms in Figure 4.1

Perceived School Climate Measuras:

ENV=Perceptions of student shaping of their environment
RES=Perceptions of student respect

FB=Perceptions of student friendship and belonging

SF =Perceptions of support and care for and by faculty/staff
SP=Perceptions of support and care for and by parents

School Bonding Measures:
ATT=Personal attachment to school
PRI=Pride in school

RUL=Fairness of rules/punishment
D1S=Personal dislike of school

59

Hypothesis #5

There are significant pathways from school climate to middle school students’
bonding to school for the entire sample.

Step I .' Examination of the Hypothesized Model 's Fit for Total Sample. Figure
4.2 below illustrates the Hypothesized SEM model that was tested for the entire sample
to find pathways of school climate that lead to school bonding (hypothesis #5). The fit
indices in Table 4.8 revealed that the fit for the hypothesized model was not satisfactory
(chi-square = 406.83, df =6 and p<.05). The NFI value of 0.84 indicated that the model
did not fit well and could be improved substantially. Furthermore, the RMSEA value
0.31 confirmed what the other indices had revealed. That is, the model needed to be
improved. The regression estimates and significance levels for these pathways are
provided in Table 4.9. Of the twenty pathways hypothesized in the initial model, eight

were not Significant.

Figure 4.2
Results of Hypothesized SEM Model for the Entire Sample3

.22
MT l‘ - (915

,7

 

.28

 

3 The numbers (estimates) illustrated on pathways are also presented in Table 4.9.

60

Table 4.8

F it Indices of the Hypothesized Models for the Entire Middle School Sample

 

 

Model N x2 df NFI4 CF15 RMSEA6
tilt

Hypothesized 706 406.83 6 0.84 0.84 0.308

***p<0.001

Table 4.9

Significant Standardized (in Italics) and Un-standardized Regression Coefficients of the
Hypothesized Model for the Entire Sample

 

Regression Path B S.E. C.R.7 __
Environment to Attachment 0.23 (0.18) *H 0.06 3.81
Environment to Pride 0.12 (0.11) I 0.05 2.49
Respect to Attachment 0.16 (0.11) * 0.06 2.68
Respect to Pride 0.23 (0.17) 0.05 4.60
Respect to Rule 0.16 (0.11) H 0.06 2.87
Respect to Dislike 0.18 (0.12) .. 0.06 2.94
Friendship to Dislike 0.28 (0.17) 0.08 3.64
Support for Staff to Attachment 0.32 (0.28) m 0.06 5.02
Support for Staff to Pride 0.22 (0.22) 0.05 4.12
Support for Staff to Rule 0.41 (0.38) 0.06 6.87
Support for Staff to Dislike 0.32 (0.28) 0.07 4.86
Support for Parents to Pride 0.18 (0.17) 0.05 3.40

 

* p<0.05 ** p<0.01 ***p_<0.001

 

’ Note: NFI= The Bentler-Bonett (Bentler & Bonett, 1980) Normed F it Index. While the scale of the fit for NFI
indices are not easily interpretable and experience is required to establish values of indices that are related to various
degrees of meaningfulness of results, authors of NFI indices suggest that models with overall fit indices of less than 0.9
can usually be improved substantially (Bentler & Bonett, 1980).

5 CF I=Comparative Ft Index

6 RMSEA=Root Mean Square Error of Approximation. Value of 0.05 or less for RMSEA indicates a close fit of the
model—value of 0.1 or more indicates a lack of fit of the model.

7 C .R. =Critical Ratio and is calculated by dividing the estimated value of B coefficient by its Standard Error (SB).

61

Step 2: Modification of the Hypothesized Model for Total Sample. The model was
modified several times by removing each non-Significant path from the model one at a
time, and comparing the value of chi-square to chi-square value prior to removal of the
path. If the change was large, the path was placed back into the model. In other word,
non-significant paths were removed from the model when the change in chi-square value
was not substantial. Furthermore, based on the results of the modification indices from
AMOS output, errors for three of the measures of school bonding were correlated. These
modifications reduced the chi-square value substantially from the hypothesized model
and allowed for construction of the most parsimonious model. Figure 4.3 shows the
final model with the best degree of fit of the data and the significant paths linking
indicators of school climate to those of school bonding. Table 4.10 reveals the fit indices
of both the hypothesized and modified models and Table 4.11 indicates significant

regression paths for the final model.

62

Figure 4.3
Final Modified Model for the Entire Sample

 

G}.

    

 

 

 

 

 

.‘1

fl

9

 

 

 

Table 4.10

Fit Indices of the Final Model in Comparison to the Hypothesized Model for the Entire
Middle School Sample

 

 

 

Model N x2 df NFI CFI RMSEA
Hypothesized 706 406.83". 6 0.84 0.84 0.31
Final 706 74.95m 1 1 0.97 0.98 0.091
***p<0.001

Comparing the fit indices of final model and those of the hypothesized model
(Table 4.10) it is clear that the model greatly improved. The chi-square value was
reduced from the value of 406.83 to 74.95, and values of NFI (from 0.84 to 0.97) and
RMSEA (from 0.31 to 0.091) also confirmed that the final model had a better fit than the

hypothesized model.

63

The regression paths from students’ perceptions of respect as well as their
perceptions of support and care for staff to all measures of school bonding were
significant in the final model. Furthermore, paths from students’ shaping of their
environment to personal attachment and pride in school were Significant. The path from
students’ friendship and belonging was significant for personal dislike of school but not
for pride in school, although the latter was kept in the model due to its impact on the
overall model (i.e., Significant change in chi-square value was observed when this path
was removed). Similarly, the only Significant path from students’ perceptions of respect
and care for parents was the one to pride in school. The path from students’ respect and
care for parents to fairness of school rules/punishment was insignificant but was kept in
the final model due to its contribution to the overall model. These are reflected in Table
4.11.

Table 4.1 l

Standardized (in Italics) and Uri-standardized Regression Coeflicients of the Final Model
for the Entire Sample

 

 

Regression Path B “i S.E. C.R.
Environment to Attachment 0.25 (0.20) WT 0.05 5.14
Environment to Pride 0.16 (0.14) 0.04 4.12
Respect to Attachment 0.18 (0.12) H 0.06 3.08
Respect to Pride 0.25 (0.19) 0.05 5.32
Respect to Rule 0.18 (0.12) m 0.05 3.62
Respect to Dislike 0.17 (0.11) ” 0.06 2.87
Friendship to Dislike 0.25 (0.16) 0.07 3.81
Support for Staff to Attachment 0.34 (0.30) at“ 0.04 8.25
Support for Staff to Pride 0.25 (0.25) 0.05 5.28
Support for Staff to Rule 0.51 (0.46) 1'" 0.04 13.70
Support for Staffto Dislike 0.30 (0.26) 0.04 7.17
Support for Parents to Pride 0.14 (0.14) it” 0.04 3.38

 

* p<0.05 ** p<0.01 ***p<0.001
Hypothesis # 6. Pathways of school climate to students’ bonding to school are

predicted to be different for female students than those for male students.

Step I .' Examination of the F it of the Hypothesized Model for Sub-Groups. To
explore the hypothesis that the pathways from school climate to school bonding are
different for males and females, SEM model was tested once for females and once for
males in the sample to see if the same constructs and/or indicators remained in the model.

Figure 4.4 and 4.5 indicates the hypothesized models for the two sub-groups by gender.

Figure 4.4

Hypothesized SEM Model for Sub-Group of Female Students

7 .27
ATT 1.191;)“ \

 

65

Figure 4.5

Hypothesized SEM Model for Sub-Group of Male Students

 

,_ a erw \
as if
.69 l /
-46 is l
.31 "3° _ I,
l
-38 K
\ \~ - '
\4 SP
Table 4.12

Fit Indices of the Hypothesized Models for Sub-Groups by Gender

 

 

 

Sub-Group N x2 df NFI CFI RMSEA
Males 383 258.03m 6 0.84 0.84 0.33
Females 323 147.95" 6 0.86 0.86 0.27
***p<0.001

As the fit indices in Table 4.12 reveals the fit for the hypothesized models were
not adequate for either of the sub-groups although it seemed to have a better fit for female
sub-group (chi-square = 147.95, df =6 and p__<.__05_) than for male sub-group (chi-square =
258.03, df = 6 and p<.05). Both the NFI values (0.84 and 0.86) and RMSEA values (0.33
and 0.27) Showed that the model could be greatly improved for both groups. For the

male sample, half of the regression estimates were significant (10 of the 20). Similarly

66

for the female sub-group 9 of the 20 paths from school climate to school bonding were
Significant in the hypothesized model (See Tables 4.13 and 4.14).
Table 4.13

Significant Standardized (in Italics) and Un-standardized Regression C oefficients of the
Hypothesized Model for Females

 

 

Regression Path B S.E. C.R.
Environment to Attachment 0.31 (0.25) ii: 0.08 4.01
Environment to Prfi? 0.16 (0.15) * 0.07 2.41
Respect to Pride 0.26 (0.19) an 0.07 3.64
Friendship to Dislike 0.27 (0.16) * 0.11 2.53
Support for Staff to Attachment 0.37 (0.36) m. 0.08 4.54
Support for Staff to Pride 0.23 (0.24) ” 0.07 3.15
Support for Staff to Rule 0.49 (0.47) 0.08 6.08
Support for Staffto Dislike 0.30 (0.28) 0.09 3.33
Support for Parents to Pride 0.16 (0.16) H 0.08 2.12

 

* p<0.05 ** p<0.01 ***p<0.001

67

Table 4.14

Significant Standardized (in Italics) and Un-standardized Regression C ocj]icients of the
Hypothesized Model for Males

 

 

Regression Path B S.E. C.R.
Respect to Attachment 0.27 (0.18) " 0.09 3.13
Respect to Pride 0.21 (0.16) " 0.07 3.09
Respect to Rule 0.28 (0.19) 0.08 3.52
Respect to Dislike 0.27 (0. 18) 0.09 3.13
Friendship to Dislike 0.30 (0.19) ” 0.11 2.67
Support for Staffto Attachment 0.24 (0.20) I 0.10 2.49
Support for Staff to Pride 0.19 (0.19) I 0.08 2.49
Support for Staff to Rule 0.32 (0.28) 0.09 3.55
Support for Staffto Dislike 0.31 (0.26) " 0.10 3.23
Support for Parents to Pride 0.18 (0.18) I 0.07 2.57

 

* p<0.05 ** p<0.01 ***p<0.001

Step 2: Examination of Group Differences. To examine whether the groups were
different with regard to pathways that linked school climate to school bonding, first the
model was modified for female sub-sample, following the same procedure as those
explained previously for the entire sample. The best model for females (see Table 4.15)
was then tested for the male sample. Results showed that the best model for female sub-
group did not fit well for the male subgroup (chi-square = 92.47, df=l4, p<.05; NFI=
0.94. CFI= 0.95. and RMSEA=0. 1211. indicating that there was indeed a Significant
difference between the two groups.

Step 3: Modification of the Hypothesized Model for Sub-Group. Further analysis
was conducted by modifying the hypothesized model for male sample in order to find the

best fitting model for the males. Figures 4.6 and 4.7 Show the final model with the best

68

degree of fit of the data and the significant paths linking indicators of school climate to

those of school bonding for each sub-group.

Table 4.15 reveals the fit indices of the final modified models for females and

males. The regression paths of the final model for each sample are indicated in Tables

4.16 and 4.17.
Figure 4.6

Final SEM Model for Sub-Group of Female Students

 

 

 

 

 

 

 

 

 

 

 

 

 

.21
,4; ENV ._ . '16 , g ATT <~~ 51K
. ._.20 -- ~ I
{p.56 ‘ . 14 30
“/69 r: RES .37.: _ 19 '
~ ‘ , _./.::’- . .~7», 7" '3 PR' <‘ ,- 32,4,
. «I 23 ‘26 / 7 " «
,.46 [fat-73$: __ N _A <// , ‘25
.31 '33 _ 3’ FB “I" "2,, : RUL < ife3)'”~
'- “.4” =11; )6
.213 . g.‘\.' / / .18 ,-
l 438 [/15 S: I“ - ~15— PEI- 4:2 DIS « e4
\.\ l _ /
\ng SP
Figure 4.7
Final SEM Model for Sub-Group of Male Students
.25
.25 ATT e1
.14
.28
.17
PR1 32
.37
.27
-6 RUL e3
.52
.15

 

'32 DIS

 

 

 

 

 

 

69

, .33

.26

Table 4.15

Fit Indices of the Final Models in Comparison to the Hypothesized Models for Sub-
Groups by Gender

 

 

Sub-Group N x2 df NFI CF] RMSEA
Hypothesized Model

Females 323 147.95‘" 6 0.86 0.86 0.27

Males 383 258.03‘" 6 0.84 0.84 0.33

Final Model

Females 323 25.15‘ 14 0.98 0.99 0.05

Males 383 55.56'” 10 0.97 0.97 0.1 1

 

* p<0.05 ***p<0.00l

As is shown in Table 4.15, the modified models for both samples were greatly
improved compared to the hypothesized models. For the female sub-group, the chi-
square value was reduced to 25.15 from the value of 147.95 and for the male sample it
was reduced from 258.03 to 55.56. Values of NF 1 and RMSEA also indicated that the fit
of the final models were improved significantly from the hypothesized models. In
addition, comparing the final models for females and males suggested that the model for
females had a better fit than that for the male sample as the chi-square value of the final
model for females was lower than that for males. Comparison of the path diagrams of the
final models for the two sub-samples (Figures 4.6 and 4.7) showed that the two groups
differed slightly. For males, the final model included a path from perceptions of support
and care for parents to fairness of school rules/punishment and it was significant (See

Tables 4.16), whereas the final model for females did not include this path.

70

Table 4.16

Standardized (in Italics) and Un—standardized Regression Coefficients of the Final Model
for females

 

 

Regression Path B S.E. C.R.
Environment to Attachment 0.31 (0.25) I" 0.06 5.19
Environment to Pride 0.16 (0.14) " 0.05 3.03
Respect to Pride 0.24 (0.17) 0.06 3.98
Friendship to Dislike 0.28 (0.17) 0.09 3.32
Support for Staff to Attachment 0.39 (0.37) "I 0.05 7.40
Support for Staff to Pride 0.25 (0.26) ”I 0.07 3.74
Support for Staffto Rule 0.54 (0.52) 0.05 10.95
Support for Staffto Dislike 0.34 (0.32) 0.06 6.12
Support for Parents to Pride 0.15 (0.15) I 0.06 2.28

 

* p<0.05 ** p<0.01 ***p<0.001

Table 4.17

Standardized (in Italics) and Un-standardized Regression Coeflicients of the Final Model
for Males

 

 

Regression Path B S.E. C.R.

Environment to Attachment 0.20 (0.16) 0.07 2.82
it

Environment to Pride 0.16 (0.15) 0.06 2.88
***

Respect to Attachment 0.30 (0.20) 0.08 3.61
***

Respect to Pride 0.25 (0.19) 0.07 3.75

Respect to Rule 0.28 (0.20) 0.07 4.25
**

Respect to Dislike 0.25 (0.17) 0.08 3.03
it

Friendship to Dislike 0.26 (0.16) 0.10 2.64
*i*

Support for Staff to Attachment 0.28 (0.23) 0.06 4.48
**

Support for Staff to Pride 0.22 (0.21) 0.07 3.14
**#

Support for Staff to Rule 0.32 (0.28) 0.08 3.98
tilt

Support for Staff to Dislike 0.25 (0.21) 0.06 3.91
**

Support for Parents to Pride 0.18 (0.17) 0.06 2.98
t

Support for Parents to Rule 0.16 (0.15) 0.08 2.16

 

* p<0.05 ** p<0.01 ***p_<0.00l

71

Hypothesis # 7. Pathways of school climate to students’ bonding to school are
predicted to be different for each subgroup of students as identified by race/ethnicity:
Black/African American students, Mexican/Latino/Hispanic students and Caucasian
students.

Step 1: Examination of the Hypothesized Model 's Fit for Sub-Groups. Similarly,
to examine the hypothesis that the model is different for White/Caucasian students than
for their Black/African American and Mexican/Hispanic/Latino counterparts, SEM
model was employed once for each subgroup to see if different indicators remained in the
model. It is imperative to note that due to the small degrees of freedom for the three sub-
groups by race, comparisons between the sub-groups in hypothesis #7 were done as
exploratory analyses.

Figure 4.8

Hypothesized SEM Model for Sub-Group of African American Students

 

72

Figure 4.9

Hypothesized SEM Model for Sub-Group of Hispanic/Latino Students

e2 ":1
/ e3‘

yeti“)

 

Figure 4.10

Hypothesized SEM Model for Sub-Group of Caucasian Students

’92,]?

(ed '>

 

73

Figures 4.8 to 4.10, above illustrate the Hypothesized SEM models that were
tested for the sub-samples by race to find pathways of school climate that lead to school
bonding (hypothesis #7).

Table 4.18

Fit Indices of the Hypothesized Models for Sub-Groups by Race

 

 

 

Sub-Group N x2 (if NFI CFI RMSEA
African American 267 109.43"* 6 0.86 0.86 0.26
Hispanic/Latino 102 50.58 6 0.87 0.88 0.27
Caucasian 212 222.57 6 0.80 0.80 0.41
***p<0.001

The fit indices in Table 4.18 revealed the fit for the hypothesized models were not
satisfactory (chi-square =109.43, 50.58 and 222.57 for African Americans, Hispanics and
Caucasians respectively, df =6 and p<.05 for all three sub-groups). The NF I values for
each of the three subgroups indicated that the models did not fit well and could be
improved substantially. Furthermore, the RMSEA values confirmed what the other
indices had revealed. That is, the models needed improvement. The regression estimates

and significance levels for these pathways are provided in Tables 4.19 to 4.21.

74

Table 4.19

Significant Standardized (in Italics) and Un-standardized Regression Coeflicients of the
Hypothesized Model for African Americans

 

 

Regression Path B SE. C .R.
Environment to Attachment 0.35 (0. 25) I“ 0.10 3.58
Environment to Pride 0.20 (0.1 7) n 0.07 2.67
Respect to Pride 0.23 (0.1 7) II 0.08 2.98
Friendship to Dislike . 0.31 (0.18)‘ 0.12 2.52
Support for Staff to Attachment 0.35 (0.29) 0.10 3.45
Support for Staff to Pride 0.25 (0. 25) II 0.08 3.24
Support for Staff to Rule 0.40 (0.38) I” 0.09 4.47
Support for Staff to Dislike 0.33 (0.29) II 0.10 3.25
Support for Parents to Pride 0.17 (0.17) i 0.08 2.22

 

* p<0.05 ** p<0.01 ***p_<0.001

Table 4.20

Significant Standardized (in Italics) and Un-standardized Regression Coefficients of the
Hypothesized Model for Hispanic/Latino Sub- Group

 

 

Regression Path B S.E. C.R.

Friendship to Pride 0.35 (0.22) 0.17 2.05

Support for Staff to Rule 0.37 (0.35) 0.16 2.29
*

Support for Parents to Pride 0.31 (0.31) 0.15 2.13

 

* p<0.05 ** p<0.01 ***p<0.001

75

Table 4.21

Significant Standardized (in Italics) and Un-standardized Regression Coefficients of the
Hypothesized Model for Caucasians

 

 

Regression Path B S.E. C.R.
Environment to Rule -025 (-020) I 0.12 -210
Respect to Pride 0.42 (0.38) 0.09 4.76
Respect to Rule 0.32 (0.24) " 0.10 3.03
Respect to Dislike 0.37 (0.29) 0.11 3.40
Support for Staff to Attachment 0.45 (0.4 I ) I" 0.12 3.90
Support for Staff to Pride 0.30 (0.31) " 0.10 3.02
Support for Staff to Rule 0.53 (0.47) 0.12 4.61
Support for Staff to Dislike 0.34 (0.31) n 0.12 2.82

 

* p<0.05 ** p<0.01 ***p<0.001

Step 2: Examination of Group Differences. To examine whether the three sub-
groups were different with regard to pathways that linked school climate to school
bonding, first the model was modified for African American sub-group, following the
same procedure as those explained previously for the entire sample. The best model for
African American sub-group (See Table 4.22) was then tested for the Hispanic/Latino
sub- sample and for Caucasians. Results showed that the best model for African
American sub-group did not fit well for the other two subgroups (chi-square =25.21,
68.77, for Hispanic/Latino sample and Caucasians respectively, df=l4, p<.05 for each of
the two sub-groups; NFI= 0.94, 0.94, CFI= 0.97, 0.95 and RMSEA=0.09, 0.14 for
Hispanic/Latino and Caucasian sub-groups respectively). These data revealed that
significant differences existed between the three sub-groups by race.

Step 3: Modification of the Hypothesized Model for Sub-Groups. Further analyses

were conducted by modifying the hypothesized model in order to find the best fitting

76

models for the Hispanic/Latino and the Caucasian samples. For each sub sample, the
model was modified following the same procedure as those explained previously for the
entire sample. Figures 4.11 through 4.13 show the final models with the best degree of
fit of the data and the significant paths linking indicators of school climate to those of
school bonding. Tables 4.22 reveals the fit indices of the final models compared to the
hypothesized models for African Americans, Hispanic/Latino sample and Caucasians.
Figure 4.11

Final SEM Model for Sub-Sample of African American Students

 

, . ENV tet ‘
, \
[If ' - \ \-
// .K '1'] '~ .
. ’ RES ‘~ ..
1‘ 34 l ,7 ’33 l .38 ~
I I“,// \ ~'\ ,. . I
24 I08 ~/. FB * 1. e3 1‘
i I-\\ ~ ‘7" ,-/
( j.2'4 ’1
\ ~\ ’
\22 l , .64
,\ .\\
\2-5

77

Figure 4.12

F inal SEM Modelfor Sub-Group of Hispanic/Latino Students

‘ ENV
- .25
. .39
"~/ ' \ A RES ~ .16
65 / y _
.28
34 .32 .
. \ 26
.29 328 _ (.9 FB
:34 ~ .48
.20 \
. ‘h . .43
.26 ' SF
, .39
\84
\ .
\\\ g -
't SP
Figure 4.13

f .18
g ATT 4
.31
. .31
‘ RUL ‘
7 v
.15
DIS ‘

Final SEM Model for Sub-Group of Caucasian Students

. ‘ ENV
. , .18
/ .65 . 7 , '
' ,75 /; RES » 4:15
“5‘ .64 .- _
‘\ ' ' .28 ‘
' .33 .
.39 '37 .. F8 .42
36 .43. ‘
. 1 \ 4 ‘ .
fi , 25
\ /' V '
‘5 / SF
\7."
‘\.\~. ., - »
~\
\1 SP

.26
_g ATT -‘
.31
*-. PRI p
p \ 7 3.32
“ *1 RUL ~-
V
.53
. .20
‘ DIS I<

78

92,6

. e19 V.

.37

. .35

93'“

.Ie4

‘1 e1 \
. .63
«4.92746 : ~45

.. .50
’ ‘x /_',/'

./

Table 4.22

Fit Indices for the Final Models in Comparison to the Hypothesized Models for Sub-
Groups by Race

 

 

Sub- Groups N )6- df NFI CF] RMSEA
Hypothesized Models

African American 267 109.43 I 6 0.86 0.86 0.26

Hispanic/Latino 102 50.58 6 0.87 0.88 0.27

Caucasian 212 222.57 6 0.80 0.80 0.41

Final Models

African American 267 19.46 14 0.97 0.99 0.04

Hispanic/Latino 102 31.12 17 0.92 0.96 0.09

Caucasian 212 52.67 14 0.95 0.96 0.1 1

 

* p<0.05 ***p<0.001

Comparison of the final models with the hypothesized models for each of the
three sub-groups showed significant improvement for the final models. The value of chi-
square was reduced significantly for each sub-group (from 109.46 to 19.46 for African
Americans, from 50.58 to 31.12 for Hispanic/Latino sample and from 222.57 to 52.67
for Caucasians). In addition, values of NFI and RMSEA indicated a significant
improvement of the fit for the final models from the hypothesized models for all three
sub-groups. Furthermore, comparing chi-square values of the final models for the three
sub-groups seemed to suggest that the model fitted the best for African Americans since
the chi-square value of the final model for this sub-group was statistically non-significant
indicating a good fit of the model. Also, comparing the chi-square values of the final
models and the critical values of goodness of fit for the sub-samples, the model seemed to

have a better fit for Hispanic/Latino sub-group than for Caucasian.

79

The regression paths of the final model for each group are indicated in Tables
4.23 through 4.25
Table 4.23

Standardized (in Italics) and Un-standardized Regression Coefficients ofthe Final Model
for African Americans

 

 

Regression Path B S.E. C.R.
Environment to Attachment 0.37 (0.27) 0.08 469
Environment to Pride 0.21 (0.19) 0.06 3.44
Respect to Pride 0.19 (0.14) " 0.07 2.91
Friendship to Dislike 0.3 3(0.I8) " 0.11 3.06
Support for Staff to Attachment 0.33(0.27)m 0.07 4.67
Support for Staff to Pride 0.2 6(0.27)m 0.07 3.65
Support for Staff to Rule 0.48 (0. 45) 0.06 8.27
Support for Staffto Dislike 0.25 (0.22) 0.07 3.75
Support for Parents to Pride 0.17 (0.1 7)" 0.06 2.63

 

* p<0.05 ** p<0.01 ***p<0.001

Table 4.24

Standardized (in Italics) and Un-standardized Regression Coefficients of the Final Model
for Hispanic/Latino Sub-Group

 

 

Regression Path B 8.13. C.R.
Environment to Rule 0.21 (0.17 ) I 0.11 1.98
Respect to Attachment 0.43 (0. 25 ) II 0.15 2.90
Friendship to Pride 0.41 (0.26) 0.12 3.48
Support for Staffto Attachment 0.33 (0. 28) II 0.11 3.05
support for Staff to Rule 0.50 (0.48) I" 0.09 5.57
Support for Parents to Pride 0.41 (0.43) 0.08 5.34
Support for Parents to Dislike 0. 40 (0.39) 0.10 4.27

 

* p<0.05 ** p<0.01 ***p<0.001

80

Table 4.25

Standardized (in Italics) and [in-standardized Regression Coefficients of the Final Model
for Caucasians

 

 

Regression Path B S.E. C.R.
Environment to Rule -024 (-021) II 0.09 -292
Respect to Attachment 0.22 (0.18) .4 0.08 2.75
Respect to Pride 0.45 (0.40) 0.08 5.82
Respect to Rule 0.36 (0.27) .4... 0.09 3.81
Respect to Dislike 0.36 (0. 28) 1... 0.09 4.25
Support for Staff to Attachment 0.46 (0.42) I" 0.07 6.61
Support for Staff to Pride 0.31 (0.31) 0.08 3.81
Support for Staffto Rule 0.61 (0.54) 0.08 8.14

 

* p<0.05 ** p<0.01 ***p<0.001

Hypothesis #8

Pathways of school climate to students’ bonding to school are Significantly
different for each subgroup of students as identified by school variable: School A, School
B.

Step 1: Examination of the Fit of the Hypothesized Model for Sub-Groups. To
test the school effect, the hypothesized model was tested once for the sample in middle
school A and once for students in middle school B. These analyses were done to
investigate whether or not the indicators in the model were different for students from

each school.

81

Figure 4.14

Hypothesized SEM Model for Sub- Group of Students in School A

in 2,, , ,u ,. ”.25
/‘.:»£ ENV k- — “.18.“ _ —f ATT i‘ * -/§j

/

1’

,,
1,,
/ r’/
/
,/./ ’
i I
.

-.13 \x.

 

   

<sz .

- /' /-_ \. , .
I / ‘s I

35 " ' ‘ " —' " “ i 83 ~
‘ » \ I ' - , I

l v ‘. I , I a

‘ .’ 1» / ’V’

I \ I; ~ I , .
\ ~ ..r/,/ ' rm 7

Figure 4.15

Hypothesized SEM Model for Sub- Group of Students in School B

.21

 

82

Figures 4.14 and 4.15 above, illustrate the Hypothesized SEM model that was
tested for the sub- samples by school to find pathways of school climate that leads to
school bonding (hypothesis #8).

Table 4.26

F it Indices for the Hypothesized Models for Sub-Groups by School

 

 

 

Sub- Group N )8 df NFI CFl RMSEA
tint

SchoolA 406 261.40 6 0.85 0.85 0.32

SchoolB 300 172.11 6 0.81 0.81 0.30

***p<0.001

The fit indices in Table 4.26 revealed the fit for the hypothesized models were not
satisfactory (chi-square = 261.40, an 172.1 I for school A and School B respectively, and
df =6 and p<.05 for each school). The NF 1 and RMSEA values also indicated that the
model did not fit well and could be improved substantially. The regression estimates

and significance levels for these pathways are provided in Tables 4.27 to 4.28.

83

Table 4.27

Significant Standardized (in Italics) and Un-standardized Regression Coeflicients of the
Hypothesized Model for Students in School A

 

 

Regression Path B S.E. C.R.
Environment to Attachment 0.22 (0.18) H 0.08 2.79
Respect to Attachment 0.18 (0.12)’ 0.08 2.34
Respect to Pride 0.25 (0.19) 0.06 3.88
Respect to Rule 0.27 (0.19) 0.07 3.68
Respect to Dislike 0.32 (0. 22) 0.08 4.05
Friendship to Dislike 0.35 (0. 23) 0.10 3.62
Support for Staffto Attachment 0.26 (0. 23) II 0.09 2.93
Support for Staff to Pride 0.28 (0.29) I" 0.07 3.84
Support for Staffto Rule 0.39 (0.35) 0.08 4.60
Support for Staff to Dislike 0.28 (0. 25) H 0.09 3.18

 

* p<0.05 ** p<0.01 ***p<0.001

Table 4.28

Significant Standardized (in Italics) and Un-standardized Regression Coefficients of the
Hypothesized Model for Students in School B

 

 

Regression Path B S.E. C.R.
Environment to Attachment 0.26 (0.19) " 0.09 2.91
Environment to Pride 0.20 (0.18) " 0.07 2.77
Respect to Pride 0.17 (0.12) I 0.08 2.20
Support for Staff to Attachment 0.37 (0.33) I" 0.09 4.05
Support for Staffto Pride 0.17 (0.18) i 0.08 2.26
support for Staff to Rule 0.43 (0. 40) 0.09 5.00
Support for Staff to Dislike 0.34 (0.30) 0.10 3.57
Support for Parents to Pride 0.20 (0.20) I 0.08 2.64

 

* p<0.05 ** p<0.01 ***p<0.001

84

Step 2: Examination of Group Differences. To examine whether the schools were
different with regard to pathways that linked school climate to school bonding, first the
model was modified for school B as the hypothesized model seemed to have a better fit
for school B. The modification was done following the same procedure as those
explained previously for the entire sample. The best model for school B (see Table 4.29)
was then tested for school A. Results showed that the best model for school B did not fit
well for school A (chi-square = 80.84, df=l3, p<.05; NFI: 0.95. CFl= 0.96. and
RMSEA=0.11). indicating that there was indeed a significant difference between the two
groups.

Step 3: Modification of the Hypothesized Model for Sub-Groups. Further analysis
was conducted by modifying the hypothesized model for school A in order to find the
best fitting model for the subgroup of school A. Figures 4.16 and 4.17 show the final
model with the best degree of fit of the data and the significant paths linking indicators of
school climate to those of school bonding for each sub-group by school. Table 4.129
revealed the fit indices of the final modified models for each school. The regression

paths of the final model for each sample are indicated in Tables 4.30 and 4.31.

85

Figure 4.16

Final SEM Model for Sub-Group of Students in School A

A ENV ]

/, ".51 (

 

Figure 4.17

g, .24
‘ ATT 4

/.
- , .31

Final SEM Model for Sub-Group of Students in School B

 

.25 _ p

.20
ATT 4

.25

:\ .10 ;//‘~\\\ I
\‘T '2 .’ PRI 4

‘1/
/.19 ,2/8/ .22
’ ‘ RUL <

’l ' i .43 “'1

,'_‘/~'
2' 2. _,15 77.14

.30

86

7:} DIS ~— ——

Keg}: / p .32
.39

. e4

e1\
.43.

752:": '29

ll ~~\
I .34}

Table 4.29

F it Indices for the Final Models Compared to Hypothesized Models for Sub-Groups by
School

 

 

 

Sub-Group N x2 df NFI CFI RMSEA
Hypothesized Model
SchoolA 406 261.40m 6 0.85 0.85 0.32
it!!!

SchoolB 300 172.11 6 0.81 0.81 0.30
Final Model

SchoolA 406 55.83 13 0.97 0.98 0.09
SchoolB 300 36.49 13 0.96 0.97 0.08
***p<0.001

As it is shown in Table 4.29 above, the modified models for both samples were
improved compared to the hypothesized model. For school A, the chi-square value was
reduced from the value of 261.40 to 53.83 and for school B it was reduced from 172.11
to 36.49. Values of NFI and RMSEA also indicated that the fit of the final models were
improved significantly from the hypothesized models.

In addition, comparing the final models of the two schools suggested that the
model had a better fit for school B than for school A as indicated by chi-square value of
the final models as well as the goodness of fit indices of the two sub-samples.

Comparison of the path diagrams for the two sub-samples (Figures 4.16 and 4.17)
Showed that the two groups differed with regard to some of the pathways in the final

model. These results are shown in Tables 4.30 and 4.31 below.

87

Table 4.30

Standardized (in Italics) and Un-standardized Regression Coeflicients of the Final Model
for Students in School A

 

 

 

Regression Path B S.E. C.R.
Environment to Attachment 0.23 (0.18) 0.06 4.03
Respect to Attachment 0.22 (0.15) " 0.07 3.07
Respect to Pride 0.31 (0.24) 0.06 5.52
Respect to Rule 0.26 (0.18) 0.06 3.95
Respect to Dislike 0.28 (0.20) I" 0.08 3.71
Friendship to Dislike 0.25 (0.17) " 0.08 3.03
Support for Staffto Attachment 0.32 (0. 28) 0.06 5.78
Support for Staff to Pride 0.42 (0.43) 0.04 9.77
Support for Staff to Rule 0.50 (0.45) 0.08 5.39
Support for Staff to Dislike 0.25 (0. 22) I" 0.06 4.48

 

* p<0.05 ** p<0.01 ***p<0.001
Table 4.31

Standardized (in Italics) and Un-standardized Regression Coefficients of the Final Model
for Students in School B

 

 

Regression Path B S.E. C.R.
Environment to Attachment 0.33 (0.25) 0.07 4.59
Environment to Pride 0.27 (0.17) 0.06 4.23
Environment to Rule 0.14 (0. 23) I 0.07 2.05
Respect to Pride 0.13 (0.10) I 0.06 2.11
Friendship to Dislike 0.31 (0.16) II 0.10 2.99
Support for Staff to Attachment 0.36 (0.31) 0.06 5.87
Support for Staffto Pride 0.19 (0.19)" 0.07 2.71
support for Staffto Rule 0.47 (0.43) 0.06 8.23
Support for Staff to Dislike 0.34 (0.30) 0.06 5.43
Support for Parents to Pride 0.19 (0.19) " 0.06 3.02

 

* p<0.05 ** p<0.01 ***p<0.001

88

Power Analysis for Tests of Model Fit

Further investigation of the power of the tests for model fit (MacCullum, Brown
& Sugawara, 1996) revealed additional information regarding the SEM analyses
presented in the above. The approach permits for direct estimation of power where effect
size is defined in terms of a null and alternative value of the Root Mean Square Error of
Approximation (RMSEA) fit index proposed by Steiger and Lind (1980). In case of
close fit power analysis addresses the question that if the fit of the model in actuality is
mediocre and the hypothesis that the fit is close is tested, what is the likelihood of
rejecting the null hypothesis? Similarly, power analysis for test of not-close fit examines
the question that if model fit is actually extremely good, and the hypothesis that fit is not
close is tested, what is the probability of rejecting the null hypothesis? Finally, in the
case of exact fit power analysis investigate the test of exact fit when true fit is close.
Although, all three alternative tests are reported in Table 4.32, the power analysis
addressing test of exact fit is preferred for the purpose of this study.

Using the sample size and degrees of freedom from final SEM analysis for each
subgroup (by gender, race and school) the power of tests for fit of the models were
examined. These results are reported in Table 4.32. When sample sizes were in-between
two values identified by MacCullum et al., (1996), a range of values for power were
considered instead of only one value. For example, for sub-group of males the sample
size was 383 (Table 4.32), but the power for model fit was calculated and reported for
sample sizes of 300 and 400 (McCullum, et al., 1996, p. 142). Since sample size of 383

falls in between these two values, a range was identified indicating that the power for

89

tests of fit of the model for males was in between the values calculated for sample sizes

of 300 and 400.

Table 4.32

Power Analysis of Tests of Model Fit

 

Range of Power By Degree of Fit of the Model

 

 

 

Sub-Group N d_f Fit N Close Fit Not Close Exact Fit

fit
Total Sample 706 1 l 700 0.612 0.555 0.661
Males 383 10 300-400 0.413-0.520 0.304-0.429 0.406-0.541
Females 323 14 300-400 0.413-0.520 0.304-0.429 0.406-0.541
African 267 14 200-300 0294-0413 0191-0304 0266-0406
Americans
Hispanics 102 17 100 0.206 0.127 0.167
Caucasians 212 14 200-300 0294-04 1 3 0191-0304 0266-0406
School A 406 13 400 0.520 0.429 0.541
School B 300 13 300 0.413 0.304 0.406

 

As indicated, the power of the fit of tests for “Exact Fit” were reasonable and was
medium for most all sub-groups except for those by race. The sample sizes for these sub-
group were lower than the other sub-groups, contributing to a lower power for testing the
fit of the model. For total sample the power was the highest due to the size of the sample.
In sum, examination of the power of tests indicated that the SEM results for fit of the
models can be trusted as the power of the tests were reasonable. The only exception was

the results for the sub-groups by race, especially the Hispanic sample. It is plausible to

90

think that a low power of the tests of fit contributed to the findings for Hispanic students

showing a smaller number of significant pathways than the other two sub-groups by race.

91

CHAPTER 5

DISCUSSION

Group Differences on Measures of School Bonding
School Bonding by Gender

The first hypothesis in this study stated that “there are significant differences
between male students and female students on their perceptions of school bonding”. This
hypothesis was partially supported.

Results indicated that afier adjusting for Students’ grade level, self-reported grade
obtained in school and educational aspirations, males and females were significantly
different on only one of the four indicators of school bonding. Females reported a
significantly higher level of perceived fairness in how school rules and punishments were
determined and enforced. The two groups did not Significantly vary on any other
measures of school bonding. Males and females were similar in how much pride they had
in themselves and their school, and how much they enjoyed school, including the people
and the setting. Overall, gender did not have a large influence on school bonding. These
results somewhat mirror previous research in which gender has not been systematically
related to school bonding in a consistent manner. Previous research on gender
differences for school bonding has been inconclusive. One study reported a higher level
of school bonding for males than for females (McDonald, & Wright, 2002) while other
studies (Simons-Morton, et al., 1999; Anson, 1995) found just the opposite. One
explanation could be that school bonding is not affected by gender since findings have

fluctuated. Another explanation could be that the findings are dependent on the

92

indicators used to measure school bonding. The present study result applied to fairness
of school rules/punishment as one of the measures of school bonding.

Findings of this study showed that female students had more positive feelings
towards school than male students, believed more than males that adults’ behaviors in
school were fair and school policies and rules were just. These results follow previous
findings (Nichols, and Good, 1998) that females perceived school rules and punishment
as more fair than did the male students. These authors conducted two surveys which
included seventh and eighth grade students in one middle school and ninth through
twelve graders in a high school. Results from both studies showed that female students
consistently rated their schools and classes as more fair than males did. Furthermore,
they reported that in both schools females especially were different than males in matters
of personal concerns.

School Bonding by Race/Ethnicity

Hypothesis 2 in this study indicated that “there are significant differences between
Caucasian students, African American students and Hispanic/Latino students on their
perceptions of school bonding”. Findings partially supported this hypothesis.

Afier controlling for students’ background characteristics of grade level, self-
reported grade obtained in school and educational aspirations, significant differences
between sub-groups by race were detected on two of the four measures of school
bonding. These were personal attachment to school and personal dislike of school where
personal dislike of school assessed students’ feelings about going to school. A high value
indicated less negative feelings towards school, while a high value revealed students’

feelings of aversion towards school. Compared to Caucasian and Hispanic/Latino

93

students, African American students reported higher levels of enjoyment and seeking
contact with school, its inhabitants and its environment. Questions had asked students
how much they would miss the school, their peers and teachers if they were to stop going
to school. African American students seemed to have more affective attachment to
school than their Caucasians or Hispanic/Latino counterparts. But, Hispanic/Latino
students and Caucasians were similar in how much they felt bonded to school.

Both African American students and Hispanic/Latino students had less feelings of
aversion towards school than their Caucasian peers. When comparing African Americans
to Hispanic/Latino students, African Americans reported having less negative feelings
towards school and more bonded to school. To explain these differences, one may
consider the sample sizes for each group of student (Tables 1 and 2). It is possible that
the African Americans and Hispanic students felt more bonded to school and less
aversion towards school, simply because the majority of students in the sample were from
diverse backgrounds (38% African Americans compared to 30% Caucasians, and 15%
Hispanics) and this being part of the majority group in school, in turn, had contributed to
their positive feelings towards school. In light of the findings of a previous research
(Anson, 1995) which associated a higher percentage of African Americans in school to a
higher level of school bonding for African American students, this explanation could
particularly apply to school B where most of its student population were African
Americans.

Furthermore, the fact that most of the population in school B had racial
background other than Caucasians (i.e., African American, Hispanic/Latino and other

minority groups) could have created a climate in school where minority students felt

94

more comfortable thus more positive towards school. This might explain the findings of
higher bonding and less aversion towards school for both African Americans and
Hispanic/Latino students in comparison to their Caucasian counterparts.

Results of the present study, however, seem contradictory to other studies (Anson,
1995; Chase, 2002) that showed a lower bonding to school for African American and
Hispanic students compared to Caucasians. Anson’s study was done on a middle school
sample using the same measures of school bonding as those in the present study and one
would expect Similarity in the findings. The reasons for these different results are not
clear. One possible explanation could be that unlike the present study, previous studies
did not control for background characteristics as covariates in conducting their analyses.
School Bonding by Grade Level

Hypothesis 3 stated ”there are significant differences between students in grades
6‘“, 7m and 8th on their perceptions of school bonding”. This hypothesis was fully
supported.

A comparison of students’ bonding to school by grade level showed that after
controlling for students’ background characteristics of gender, educational aspirations
and self-reported grade obtained in school, students’ grade level were significant for all
four measures of school bonding. Sixth grade students were significantly different than
their counterparts in seventh and eighth grade on all measures of school bonding. A
trend was found for bonding to school that indicated steady decrease from sixth to
seventh and to eighth grades, with the most significant declines occurring early in 6th and
7‘h grades. That is, 6th grade students a) enjoyed the people in school and the environment

of school more than those in 7‘h and 8th grades, b) felt proud of their school and perceived

95

 

others in school to be proud of the school more than their peers in upper grades, c) felt
that adults and school rules were fair and policies were conducted in just ways more than
7m and 8th graders, and (I) felt more positive about going to school than students in higher
grades in middle school. One reason for these findings could be that as students move up
in grades from 6th to 8‘“, they experience more pressure from school both academically
and socially which makes them less bonded to school. In addition, students in 7th and 8h
grades may be closer in age to adolescence and perhaps dealing with stresses of puberty
makes them less bonded and less positive towards school.

Results of this study confirmed the findings of a previous research (Simons-
Morton, et al., 1999) where bonding of 6‘h graders to school was found to be significantly
higher than the 7th and 8h graders. Similarly, 7th graders had higher bonding than their
peers in 8th grade. In addition, students in higher grades liked school less, and felt less
closely affiliated with school. Similar findings were also reported by another study
(Boggiano & Katz, 1991) that showed a decrease in student satisfaction with school
climate during middle school years. Authors attributed this finding to a lack of person-
environment fit

These findings however, are somewhat counter intuitive since one might assume
that because 6th grade students are new to the middle school environment, their lack of
familiarity with school would result in a weaker bond to school. In comparison, 7th and
8th graders would have a better command of the rules in middle school, and would have
established social groups, which would consequently create a stronger school bonding

for these students. Results of the present study as well as those of previous research

96

 

however suggested that bonding to school decreased steadily during the middle school
years.
School Bonding by School Eflect

Hypothesis 4 stated “there are significant differences between students in one of
the middle schools in the sample (school A) and students in the other middle school in the
sample (school B) on their perceptions of school bonding”. This hypothesis was partially
supported.

Results indicated that after controlling for students’ background characteristics of
grade level, gender, self-reported grade obtained in school, and educational aspirations as
covariates, significant differences existed between students in school A and those in
school B on two measures of school bonding: pride in school and personal dislike of
school.

Students in school A reported having higher feelings of pride for themselves, their
school, and others at school than students in school B. The findings that students in
school A had significantly higher pride in their school than their counterparts in school B
may be a reflection of characteristics of the school, style of leadership and organizational
climate of school. Some observational data from the middle schools in the sample
revealed that the leadership style of the principals in the two schools were different. In
school A the principal was more democratic in leadership style, teachers seemed to be
more involved in decision making, organizational climate of school seemed to be more
intact and as a whole there seemed to be more sense of pride in school. However, a very
different leadership style seemed to be operative in school B, where the principal seemed

to be making most of the decisions regarding completion of the surveys, training time,

97

and teaching the curriculum without teachers’ input. School B was characterized by
higher levels of disorganization and had lower rates of compliance with the training,
intervention and evaluation components of the study. It is possible that these variables
(style of leadership, organizational climate of school) had influenced the sense of school
spirit, thus affected students’ feelings of pride in school and their perceptions of how
proud other students and teachers were in school.

In school B students seemed to have less feelings of aversion towards school than
those in school A. To explain this difference, the same reasoning as those explained for
the discussion of findings in hypothesis 2 applies. It (See discussion of differences
among students by race/ethnicity for hypothesis 2, above), Perhaps a higher percentage of
students with racial backgrounds other than Caucasians in school B could have created a
social climate that was more comfortable and stimulating for students of color which
contributed to a lower aversion towards school for students in school B compared to
those in school A. Furthermore, because a smaller number of Caucasian students
attended school B, it is possible that their response to how they felt about school were
overshadowed by the response of students with other racial background, hence the
findings of significant differences between the two schools. In other words, the findings
that students in school B felt less aversion towards school may be reflecting the feelings

of students from racial backgrounds other than Caucasian.

98

Structural Equation Modeling (SEM) Analysis
Pathways of School Climate to School Bonding for the Entire Sample

Hypothesis 5 asserted, “There are significant pathways from perceived school
climate to middle school students’ bonding to school for the entire sample”. This
hypothesis was partially supported.

There were significant pathways from perceived school climate to school bonding
when entire middle school sample was considered. This confirmed the findings of
previous literature that students’ perception of school climate is a significant predictor of
their bonding to school (McBride, et. al., 1995; Anson, 1995; Kumpfer, & Turner, 1990-
1991; Pyper, Freiberg, Ginsburg, & Spuck, 1987). Findings of this study revealed
additional information on how school climate might predict bonding to school through
identifying specific paths that linked the indicators of school climate to those of school
bonding.

The final model identified two indicators of school climate that weighted heavily
on students’ bonding to school. These were students’ perceptions of respect (measuring
perceptions of respect for other students, staff and school property) and their perceptions
of support and care for staff (measuring students’ perception of caring and respect
students, parents and other faculty and staff have toward faculty and staff). Perhaps these
two attributes of school climate are indicative of a warm and friendly climate which
consequently increases students’ bonding to school. Furthermore, it appeared that except
for friendship and belonging, all other indicators of school climate significantly

contributed to students’ perceptions of pride in school.

99

Students’ involvement in the shaping of their environment (measuring students’
attempts to influence the behavior and actions of others and the overall well-being of the
school) was also a significant, although a medium contributor to school bonding. This
finding confirmed the results of previous research that showed students’ involvement in
school was a Significant variable that affected students’ bonding to school (McBride et
al., 1995). The present study added to pervious literature by revealing that involvement
in school influenced school bonding by affecting students’ attachment to school and their
feelings of pride in school.

Finally, support and care that school provided for parents also influenced
students’ bonding to school. This finding was supported by previous research that
identified parent-school connections (Pryor, 1994; Sunah, 2000) as a key variable that
affected students’ bonding to school.

Pathways for Sub—Groups by Gender

Hypothesis 6 postulated that “pathways of school climate to students’ bonding to
school are predicted to be different for male students than those for female students”.
This hypothesis was partially supported by the results indicating that the pathways in the
final models were different for males than those for female students. Support and care
that school provided for teachers and staff significantly contributed to both males’ and
females’ school bonding.

In addition, student’s involvement in shaping of their environment and their
perceptions of friendship and belonging in school affected the bonding of males and
females to school in similar ways. However, the impact of these two variables on

students of both genders was mild. Degree to which students’ perceived respect in school

100

 

heavily affected male students’ bonding to school, but did not have a substantial
influence on female students’ bonding.

Moreover, the final model for males included a pathway from perceptions of
support and care for parents to fairness of school rules/punishment. The final model for
female students did not include this pathway at all. This finding suggest that male
students’ perceptions that school (including teachers, students, and other parents)
provides support and care for parents somehow was related to their perceptions that
adults in school were fair and that school policies were carried in just ways. A closer
look at the survey items for measure of perceptions of support and care for parents may
shed some light about these results.

Some of the items for this construct were: parents show that they care about their
child’s education and school behavior; teachers treat parents with respect; this school
cares about the thoughts and feelings of parents; in their interactions with children,
parents display the character qualities the school is trying to teach; and students try to get
other students to follow school rules. It is possible that students’ perceptions that school
and teachers treated students’ parents fairly and with respect had instilled a sense of
belief particularly in male students, that adults in school were fair and that rules and
punishments also were just. In addition, one of the items measured students’ perceptions
about parents’ displaying the character traits that are taught in school. Perhaps male
students were more affected by parents’ authority than girls, especially during the
adolescence, and if they perceived that the character of their parents was similar to the

values of school, it would impact their opinions that school rules were fair.

10]

Findings of this study also indicated that the final model for females had a better
fit than the final model for males. The results in the present study seems to contradict the
findings of previous research concerning gender differences (Kumpfer & Turner, 1990-
1991) where a better fit of the model was reported for males than for females.

Pathways for Sub- Groups by Race

Hypothesis 7 suggested that “pathways of school climate to students’ bonding to
school are predicted to be different for each subgroup of students as identified by
race/ethnicity: Black/African American students, Mexican/Latino/Hispanic students and
Caucasian students”. This hypothesis was partially supported.

Results indicated that the paths linking perceptions of support and care for faculty
to school bonding influenced bonding to school for all three sub-groups, especially the
African Americans and Caucasians. That is perceptions of respect for others and for the
school property as well as caring and respect students, parents and other faculty, staff,
showed toward faculty and staff were major contributors of students’ bonding to school
for African Americans and Caucasians. The effect of this indicator on Hispanic/Latino
students was less pronounced.

Moreover, findings confirmed the hypothesis that the pathways of school climate
to students’ bonding to school were different for African Americans,
Mexican/Latino/Hispanic students and Caucasians. Perceptions of student respect
seemed to have heavily influenced Caucasian students’ bonding to school compare to the
other two groups.

Furthermore, friendship and belonging and support and care for parents did not

have any affect on Caucasian students’ bonding while these two indicators of school

102

 

climate somewhat influenced the bonding of African Americans and Hispanic/Latino
students to school.
Pathways for Sub- Groups by School

Hypothesis 8 claimed that “pathways of school climate to students’ bonding to
school are significantly different for each subgroup of students as identified by school
variable: School A, School B”. This hypothesis was also partially supported.

Comparing the pathways of perceived school climate to school bonding for the
two schools in the sample, the fit of the final models seemed to be better for school B
than school A. For both schools, the indicators of perceptionsvof support and care for
faculty were major contributors to school bonding. Although perceptions of students’
involvement in shaping of their environment affected both bonding in both school, it
seemed to have a greater influence on students’ bonding to school in school B than in
school A. Direct pathways from this indicator of school climate affected students’
attachment to school, pride in school and fairness of school rules/punishment. Similarly,
perceptions of student respect had a greater effect on school bonding in school A, than in
school B.

While support and care for parents affected students’ bonding to school by a
direct path to feelings of pride in school, it had no effect in school A. Based on the
observational data gathered during the course of this study, schools seemed to be quite
different with regard to their social and organizational climate. Moreover, demographic
data indicated that school A was mostly composed of Caucasian students while School B
included mostly Afiican Americans, Hispanic/Latino students and students with other

minority backgrounds. The findings here could be partially due to the differences in the

103

 

culture and climate of the two schools. It is also possible that because the majority of
students in school B were African Americans and students with race/ethnicity
backgrounds other than Caucasian. Further qualitative data about school variables and
climate of the schools are warranted to shed more light on these finding.

The findings that showed a link from perceptions of support and care for parents
to feelings of pride in school for school B suggests that for students in school A the
perceptions that school respected and cared about the parents had affected their feelings
of belonging to school and being proud of their school. This could also be discussed in
light of the observed differences in organizational and social climate of the two schools.
In school A the principal demonstrated more democratic style of leadership and there
seemed to be more harmony among the school faculty, staff and students. Furthermore,
the school in general seemed to have friendlier climate than school B.

Among the items that measured support and care for parents were ‘Parents show
that they care about their child’s education and school behavior’; ‘teachers treat parents
with respect’; and ‘this school cares about the thoughts and feelings of parents’. It is
possible that due to the differences in climate of the two schools, these variables were
more pronounced in school A than in school B. Also, the parent-school connection could
be stronger in school A than it was in school B, affecting students’ perceptions and
having a more positive attitude towards school rules and policies. Further qualitative data
on various aspects of the climate of the two schools are needed to confirm these

speculations.

104

Summary of Findings and Conclusions

Summary of Findings for the Entire Sample

Considering the entire sample, perceptions of students’ respect and perceived
support and care for faculty seemed to be major contributors to middle school students’
bonding to school. . Perceptions of student respect assessed respect for other students,
staff and school property while perceptions of support and care for faculty measured
caring and respect students, parents and other faculty, staff, had toward faculty and staff.

These findings suggested that student respect and positive affect for teachers and
staff are key components related to school bonding. Because the school climate and
school bonding data were collected at the same point in time, it is difficult to know
whether or not respect students perceive in school, and positive regard for faculty are
precursors of school bonding or rather they are indicators of relationships characterized
by the presence of bonding.
Summary of Findings by Grade Level

Findings indicated a linear but negative relationship in middle school students’
bonding to school by grade level. That is, students in lower grades had stronger
bonding to school than those in higher grades. These findings were in line with previous
literature showing that as students moved up from sixth to eighth grades their bonding to
school decreased (Simons-Morton, et al., 1999), and their satisfaction with school
environment was reduced due to a lack of person-environment fit (Boggiano & Katz,

1991; Eccles, Midgley, Wigfield, Buchanan, Reuman, Flanagan, & Maclver 1993).

105

Summary of Findings by Gender

The present study findings by students’ gender revealed significant differences
between males and females on one of the four indicators of school bonding. That is,
fairness of school rules and punishment. Examinations of the pathways of perceived
school climate to school bonding indicated that perceptions of respect in school
environment had a much stronger effect on school bonding for males than for females.
Support and care for teachers and staff in school influenced both male and female
students’ bonding to school.

There seemed to be a lack of congruency in the findings of this study for gender
differences (pervious study results were also inconsistent with regard to gender
differences in school bonding). When gender differences for school bonding alone were
examined, a stronger bonding to school for females was detected than for males.
However, when the pathways from school climate to school bonding were investigated
male students seemed to have a better fit of the model than females did. This suggests
that additional variables may be important for fully understand and explain how males
and females may bond differentially to school.

Summary of Findings by Race

Results indicated that students from different racial groups had different levels of
bonding to school. African American students had a stronger bond to school than both
Hispanic/Latino students and Caucasians on some measures. African American students
reported being more personally attached to school and had less aversion towards school

than Hispanic/Latino students or Caucasians. While Hispanic/Latino students and

106

Caucasians were not different on their attachment to school, they had lower aversion to
school than Caucasian students did.

Examination of the pathways of perceived school climate to school bonding
showed that perceptions of support and care for faculty were major contributors to school
bonding for African Americans and Caucasians. This variable had a mild effect on
Hispanic/Latino students’ bonding to school. Furthermore, Caucasian students’ bonding
was heavily influenced by their perceptions of respect in the school environment.
Summary of Findings by School

The two schools in the sample were significantly different on two measures of
school bonding. Students in school A had a higher level of pride in their school than their
counterparts in school B. Students in school B, however, had less aversion towards
school than their peers in school A. Examination of the pathways of perceived climate to
school bonding showed that the indicators of perceptions support and care for staff had a
significant impact on students’ bonding in both schools. While, perceptions of student
respect had a stronger effect on school bonding in school A, than in school B, indicators
of school environment had a larger effect in students’ bonding in school B than for those
in school A.

Conclusions

In sum, the following conclusions can be made based on the results that were

shown and discussed in this study:

0 Perceived positive affect for teachers and staff seemed to be the most
significant contributor of students’ bonding to school.

0 Perceived respect for others in school and for school property was the
second significant contributor of school bonding. In particular,

107

indicators of student respect affected school bonding of males,
Caucasian sub-sample, and students in school A.

Degree of students’ involvement in the overall well being of their
school seemed to be the third significant contributor of students’
bonding to school The impact of this variable was most evident in
bonding of students in school B.

As students moved up from grade 6 to grade 8 their bonding to school
was shown to decrease steadily.

Females were more likely to report that rules and punishments at
school were fairly developed, and enforced than males.

Male students’ bonding to school was more likely affected by their
perceptions that students in school were treated with respect by adults
and by other students, than females’ bonding.

African American students had stronger bond to school than both
Hispanic/Latino students and Caucasians. They were more attached to
school and had less aversion towards school.

Hispanic/Latino students had stronger bonding to school than
Caucasians as indicated by their lower aversion towards school.

Respect for others and for school property affected Caucasian
students’ positive bonding to school more than it did for African
Americans and Hispanic/Latino students.

Positive affect towards teachers and staff in school contributed to
bonding to school for African Americans and Caucasians more than it
did for Hispanic/Latino students.

Schools were significantly different on two measures of school
bonding. Students in school A had a higher sense of pride in
themselves and in their school than those in school B. In school B
students had less aversion towards school than their peers in school A.

Support and positive affect towards teachers and staff influenced
bonding of students in both schools.

108

Methodological Problems/Limitations

One limitation of this study is that the sample was selected non-randomly. That is
schools were conveniently sampled. This may cause difficulty in the generalizing of the
results to other middle schools. Also, the small number of schools in the sample could
limit any school level inferences in the study. Furthermore, a relatively small sample size
for sub-samples by race, particularly the Mexican/Hispanic/Latino group may have
caused a smaller number of significant values for the estimates. In addition, there were
limited amount of qualitative information in this study, hindering explanations of some of

the findings, especially for discussing the results for sub-groups.

Implications and Future Directions

This study provided specific information about perceived school climate variables
that contributed to middle school students’ bonding to school. From the psycho-social
perspective, this seems to be an interesting phenomenon in that it shows how students’
perception of other students’ behaviors and attitudes as well as their own behaviors and
attitudes in school could affect individual’s bonding to school.

One major finding showed that perceived positive affect for teachers and staff in
school contributed to students’ bonding to school. It is possible that, the perceptions that
teachers were being respected and appreciated by adults in school; students were treated
fairly by teachers; teachers were involved in school decision making; and adults cared for
students’ well being and their safety could be an indication of a supportive and warm
school environment in which strong relationships between students and teachers and staff

were developed. In turn, these strong relationships affected students’ bonding to school

109

in positive ways. In addition, perhaps the style of leadership in such an environment was
more democratic to allow for teacher involvement in school decisions.

Another finding of this study revealed that students’ bonding to school were
affected by their perceived respect for one another, for teachers and staff, and for school
property. It is possible that these findings reflected school climate in which respect for
individuals and for school property were encouraged and practiced, thus providing more
opportunities for students to build trustworthy relationships and fiiendships These
relationships, in turn, promoted and encouraged a stronger bond to school.

These findings seem to point and encourage school characteristics that provide a
healthy context in which students can develop healthy attachment to adults, students and
to school. These findings along with those pertaining to students’ gender, race/ethnicity,
grade level, and schools provided information that has implications for both research and
practice.

Research in the area of adolescence development as well as those related to
school climate and school bonding could benefit from the knowledge provided in this
study. The literature search in this document revealed that the number of studies on
school bonding and particularly those linking students’ bonding and perceived school
climate variables were sparse. Furthermore, it was indicated that to date few research had
addressed differences among sub-groups on bonding to school as well as on pathways of
perceived school climate to school bonding While findings of this study provided
answers to some of the research questions that were posed here. They were constrained

by the limitations of the study and need to be addressed by future investigations.

110

Furthermore, the perspective of schools’ and practitioners’ findings from this
research may be very important, providing them with the knowledge base from which
programs could be developed to improve students’ bonding to school. If students’
respect for others and the property of school significantly contributes to their bonding to
school, as the findings of this study have indicated, then schools could place more
emphasis in cultivation of an environment in which students are respected and valued.
Similarly, as this study indicated if treating teachers and staff with respect and
appreciation, and teachers’ care for the safety and well being of students increases
students’ bonding to school, then the information from this study could be used to inform
practices that allows for creation of such an environment that contributes and promotes
positive bonding of students to school.

In addition, findings of this study informed and further confirmed what experts
and developmental psychologists, such as Bronfenbrenner (1979) and Garbarino (1995),
have been advocating for years. That is that creating a positive developmental context in
schools promotes positive attachment and bonding to school, thus contributes to students’
healthy development in all areas.

Future research should try to address the methodological limitations that were
addressed by this study. Those included selecting random sample for the study as well as
selecting a larger sample size for groups by race/ethnicity. Selection of a stratified
random sample would seem to be a prudent way to resolve the small sample sizes for
sub—group. Furthermore, future research should try to include a larger sample of schools
that permits greater generalizability of results to other schools. In addition, such studies

should try to include a richer qualitative data from schools and students in middle schools

1]]

 

in order to allow a greater explanation for some of the findings, especially those
pertaining to sub-groups.

Finally, with regard to findings of the present study as well as those of previous
research suggesting that school bonding decreased steadily during the middle school
years, future research Should conduct additional analysis to include grade by gender sub-
sample. This is to investigate whether a different trend than those that were shown here

would emerge.

112

APPENDICES

113

 

APPENDIX A

SURVEY ITEMS FOR STUDENTS’ DEMOGRAPHIC CHARACTEIRSTICS
&
PERCEIVED SCHOOL CLIMATE

Source: ‘School as a Caring Community Profile —II instrument (SCCP-ll)’ (Center for 4th
and 5‘h R’s, 2000),

114

Student School Climate Survey - Spring 2003
Middle Schools

 

 

What is this survey for? This voluntary survey is for all students and
adults at this school so that we can determine how to make this school a
better place for everyone.

We will ask you to fill this same survey out again later this year and again
next year so we can see how much progress we are making. 50 we need your
name to see how you think things have changed or stayed the same at your
schooL

All your answers will be kept completely confidential (a secret) and your
name will never appear in any reports. If you have any questions, please ask

your teacher or principal.

m Remember, no one of school will see your answers.

 

1. What is your school name?

 

2 . Your teacher's name:

 

3. Class period:

 

4. Your name:

 

5. Are you a male or a female?
1 Male
2 Female

6. What grade are you in? (circle one)
6 7 8
7. How old are you?

8. How many years have you been at this school?

 

115

 

8 hich group describes you best? ( you my circle more than one)

. W
1 American Indian or Alaska Native
2 Asian

3 Black/African American

4 Mexican/Hispanic/Latino

5 Pacific Islander

6 White—Not Hispanic

7 Other (specify)

9. What grades do you usually get?
Mostly A's

Mostly A's and 8's

Mostly B's

Mostly B's and Cs

Mostly C's

Mostly Cs and D's

Mostly D's

Mostly D's and F's

CD‘IO‘Ul-hWNH

10. How far will you go in school?

I Won't finish high school

2 Finish high school and stop
After high school «-

3 Go to trade or vocational school
4 Go to college for less than 4 years
5 Graduate from a 4-year college

11. What is the highest level your mother (or female guardian)
completed in school?

8’h grade or less

Some high school

Graduated from high school

Some college

Graduated from a four year college

I don't know

9014:th—

ll6

   

 

Please circle one number for each As often Frequently Almost
sentence that describes how often you _
think these things occur at your 3’
school:

1 Students treat classmates with

    
 
 
 
 

 

 

 

dents help each other, even if
are not friends

 
 
   

  

Students show respect for school
property (such as, desks, walls,
bathrooms, busses, buildings, and
- rounds).

    
 
 
 

 
 

Students try to get other students
to followschool rules. _

 
 
 
    

‘ Stduent work well to ether

  

Students are disrespectful toward
thei teachers.

    

  
  
   

Students try to have a positive
influence on the behavior of other 1 2 3 4 5

    

Students are willing to forgive each
oth_e_r.

 

117

Students are patient with each
other.

Students are disrespectful toward
their schoolmates.

When students see another
student being picked on, they try to
Sta - it.

Students share what they have
with others.

Students can talk to their teachers
about problems that are bothering
them.

In their interactions with students,
all school staff (the principal, other
administrators, counselors,
coaches, secretaries, aides,
custodians, bus drivers, etc.) act in
ways that demonstrate the
character qualities the school is

t i - to teach.

Teachers 90 out of their way to
hel students who need extra hel

 

118

      
  

 

Please circle one number for each Almost Some- As often Frequently Almost
sentence that describes how never times as not always
often you think these things

occur at - school:

33 In this school you can count on

adults to try to make sure that 1 2 3 4 5
students are safe.

 

 
   
 

 

  
 

35 Faculty and staff treat each other
with respect (are caring, 1 2 3 4 5
su- - -rtive, etc.).

       
 

37 Teachers are unfair in their 1 2 3 4 5
treatment of students. , ‘

39 Faculty and staff are involved in
hel - in- to make school decisions.

  
  

 
 
 
 

Parents show res ct for teachers.

 

43 This school shows appreciation for
the efforts of faculty aid staff.

 

 

119

APPENDIX B
SURVEY ITEMS FOR SCHOOL BONDING

Source: ‘Survey of School Climate’ (Anson, 1995)

120

In this section we will ask you questions about how you feel

about JOUI‘ school

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Please tell us how much you agree Strongly Agree Neither Disagree Strongly
with each comments by circling Agree 09"” "0" 05509?“
ONLY one answer for each d'509'ee
sentences below:
1. School is boring 1 2 3 4 5
2. When I wake up in the 1 2 3 4 5
morning, I often don't feel
like going to school
3. Sometimes I pretend to be 1 2 3 4 5
sick so I don't have to go to
school
If you had to stop going to this Not at Not very Some Pretty A lot
school, how much would you °' "‘“Cl‘ "WC"
miss: (Please circle ONLY one
answer for each question)
4. Your fellow students? 1 2 3 4 5
5. Your teachers? 1 2 3 4 5
6. The principal? I 2 3 4 5
7. Adults in the school who are 1 2 3 4 5
not teachers or the principal?
8. The sense of school Spirit? 1 2 3 5
9.The way this school treats 1 2 3 5
students?
Please Circle only one number for Not at Not very Kind of Proud Extreme
the following question:
all proud proud proud -ly proud
10. How proud do you feel of 1 2 3 4 5
your school?

 

121

 

 

 

 

 

 

 

 

 

 

 

Please circle one number for None A few About Most All
each sentence that describes half
what you think about each
question:
11. How many students feel proud 1 2 3 4 5
of this school?
12. How many students are willing 1 2 3 4 5
to defend the school when
bad things are said about it?
13. How many students care 1 2 3 4 5
About keeping the school
clean and in good condition?
14. How many parents Of 1 2 3 4 5
students feel proud of this
school?
Please ccontinue to circle one Not at Not Kind of Clear Extreme
number for this question: all Clear very clear ly Clear
clear
16. How clearly stated are the 1 2 3 4 5
school rules?
17. How fair are the school 1 2 3 4 5
rules?
18. When students are punished 1 2 3 4 5
at school, how fair is the
punishment?

 

 

 

 

 

 

122

 

APPENDIX C

Results of the Multivariate Analysis of Variance

123

Results of the Multivariate Analysis of Variance

Table 1

Diflerences Among Students by Gender on Measures of School Bonding

 

Test Value F Hyp df Error df Sig of F

 

Wilks 0.98 2.74 4.00 676.00 0.03
Univerate [Notez Univariate? Do search and replace thru text] F-Tests with (l, 679) (if

 

 

 

Variable MS Hyp MS Error F Sig of F
Personal Attachment to school 1.67 0.84 1.90 0.16
Pride in School 1.53 0.62 2.46 0.12
Fairness of Rules/Punishment 8.44 0.78 10.87 0.00
Personal Dislike of School 0.69 0.83 0.83 0.36
Table 2

Diflerences Among Students by Race/Ethnicity on Measures of School Bonding

 

Test Value F Hyp df Error df Sig of F

 

Wilks 0.96 2.91 8.00 1 108.00 0.003
(Univariate?) F-Tests with (2, 557) df

 

 

Variable MS Hyp MS Error F Sig of F
Personal Attachment to school 3.24 0.87 3.74 0.024
Pride in School 1.15 0.61 1.88 0.154
Fairness of Rules/Punishment 0.02 0.78 0.03 0.971
Personal Dislike of School 4.95 0.80 6.22 0.002

124

Table 3

Diflerences Among Students by Grade Level on Measures of School Bonding

 

 

 

 

 

Test Value F Hyp df Error df Sig of F
Wilks 0.95 4.49 8.00 1350.00 0.000
Univariate F-Tests with (2, 678 ) df

Variable MS Hyp MS Error F Sig of F
Personal Attachment to school 6.46 0.84 7.67 0.001
Pride in School 10.55 0.62 17.08 0.000
Fairness of Rules/Punishment 4.50 0.77 5.81 0.003
Personal Dislike of School 3.91 0.83 4.73 0.009
Table 4

Diflerences Among Students by School on Measures of School Bonding

 

 

 

Test Value F Hyp df Error df Sig of F
Wilks 0.94 1 1.15 4.00 675.00 0.00
Univariate F-Tests with (1, 678) df
Variable MS Hyp MS Error F Sig of F
Personal Attachment to school 1.22 0.84 1.45 0.23
Pride in School 6.12 0.61 10.00 0.00
Fairness of Rules/Punishment 1.84 0.77 2.38 0.12
Personal Dislike of School 4.85 0.82 5.91 0.02

 

125

REFERENCES

Ainsworth, M. (1989). Attachments beyond infancy. American Psychologist. 44.
709-716.

Anson, A. R. (1995). Risk and protection during the middle school transition: The role of
school climate in early adolescent development. Dissertation Abstracts
International, Section B: The Sciences and Engineering, 56 (3-B), 1716.

Bentler, P.M., & Bonett, D. G. (1980). Significance tests and goodness of fit in the
analysis of covariance structures. Psychological Bulletin, 88, 588-606.

Boggiano, A. K., & Katz, P. (1991). Maladaptive achievement patterns in students: The
role of teachers' controlling strategies. Journal of Social Issues 47, 35-51.

 

Bowlby, J. (1979). The making and breaking of affectionate bonds. New York:
Tavistock.

Bowlby, J. (1969). flachment and loss. Volume 1: Attachment. New York: Basic
Books.

Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA:
Harvard University Press.

Bryant, A. L., Schulenberg, 1., Bachman, J. G., O’Malley, P. M., & Johnston, L. D.
(2000). Acting out and lighting up: Understanding the links among school
misbehavior, academic achievement and cigarette use. Monitoring the Future
[On-line], 46. Available: www.edrs.com/members/s_p.cfm?An=ED443065.

Center for the 4th and 5th Rs. (2000). School as Caring Community Profile (SCCP).
Cortland, NY: State University of New York.

Chase, M. A. (2002). An examination of cultural mistrust in a school-based context for
African American adolescents. Dissertation Abstracts International, Section B:
The Sciences and Engineering, 62 (7-B), 3418.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests.
Psychometrical, 16, 297-334.

Dworkin, A. (1987). T_e_acher Burnout in Public Schools. NY: State University of New
York Press.

Eccles, J. 8., Lord, 8., & Buchanan, C. M. (1996). School transitions in early

adolescence: What are we doing to our young people? In J. A. Graber, J. Brooks-
Gunn, & A. C. Petersen (Eds), mnsitions Through Adolescence: Interpersonal

126

Domains and Context. (pp. 251-284). NJ: Lawrence Erlbaum Associates,
Publishers.

Eccles, J. S., Midgley, C., Wigfield, A., Buchanan, C. M., Reuman, D., Flanagan, C., &
Maclver, D. (1993). Development during adolescence: The impact of
stage-environment fit on young adolescents’ experiences in schools and families.
American Psychologist. 48. 90-101.

ETR Associates. (2000). ETR Character Edtfltion Survey PO. Box 1830, Santa Cruz,
CA 95061-1830. Phone: (831)438-4060. Fax: (831)438-4080. www.ctrorg

Fraser, B. J ., (1990). Individuflzed Classroom Environment Questionnaire. Revised and
Reprinted. The Australian Council for Educational Research Ltd., Sidney, AUS.

Garbarino, J (1985). Adolescent Development: An ecological Perspective. OH: Merrill
Publishing Co.

Garbarino, J. (1995). Raising Children in fiocially Toxic Environment. San Francisco:
Jossey-Bass.

 

Gonder, P. O., & Hymes, D. E. (1994). Improving school climate & culture. AASA,
critical issues report, 27.

Goodenow, C. (1991). The sense of belonging and its relationship to academic
motivation among pre and early adolescent students. Paper presented at the
Annual Conference of American Educational Research Association, Chicago,
Illinois.

Gottfredson, D. C. (1986). An empirical test of school-based environmental and
individual interventions to reduce the risk of delinquent behavior. Criminology,
2_4 (4), 705-731.

Hallfors, D., & Van-Dom, R. A. (2002). Strengthening the role of two key institutions in

the prevention of adolescent substance abuse. Journal of Adolescent Healthg30
(1), 17-28.

Hawkins, J. D., & Catalano, R. F. (1990). Broadening the vision of education: schools
as health promoting environments. Journal of School Health. 60 (4), 178-181.

Hawkins, J. D., Graham, J. W., Maguin, E., Abbott, R., Hill, K. G., & Catalano, R. F.
(1997). Exploring the effects of age alcohol use initiation and psychosocial risk

factors on subsequent alcohol misuse. Journal of Studies on Alcohol. 58 (3), 280-
290.

Hawkins, J. D., & Weis, J. G. (1985). The social development model: an integrated
approach to delinquency prevention. Journal of PrimagPrevention. 6, 73-79.

127

Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective factors for
alcohol and other drug problems in adolescence and early adulthood: Implications
for substance abuse prevention. Psychological Bulletin, 112 (1), 64-105.

Hirschi, T. (1969). Quses of delinquency. Berkeley: University of California Press.

Hoppe, M. J ., Wells, E. A., Haggerty, K. P., Simpson, E., Gainey, R. R., & Catalano, R.
F. (1998). Bonding in a high-risk and a general sample of children: Comparison
of measures of attachment and their relationship to smoking and drinking.
Joumal:of Youth and Adolescence. 27 (1), 59-81.

Howard, E., Howell, B. Brainard, E. (1987). Handbook for conducting school climate
improvement projects. Phi Delta Kappa Education Foundation: Bloomington, IN

Huber, M. Q., & Baraily, S. B. (2002). Comparative tafl of school climate assessment

instruments for middle-school and high school students. Unpublished
Manuscript.

Jessor, R. (1994). Adolescent development and behavioral health. In J. D. Matarazzo, S.
Weiss, J. A. Herd, N. E. Miller & S. M. Weiss (Eds), Behavioral Beam A
Handbook of Health Enhancement and Disease Prevention (pp. 69-90). New
York: Pergamon Press.

 

Johnson, (1998-1999). Learning Environment Research, 1(3), 353-367

Kester, V. M. (1994). Factors that affect African-American students’ bonding to middle
school. The Elementary School Journal. 95 (1), 63-73.

Kirby, D. (2002). The impact of schools and school programs upon adolescent sexual
behavior. Journal of Sex Research. 39 (1), 27-33.

Kumpfer, K. L., & Turner, C. W. (1990-1991). The social ecology model of adolescent
substance abuse: Implications for prevention. International Journal of the
Addictions 25 (4-A), 435-463.

 

Lanni, F. A. J ., & Orr, M. T. (1996). Dropping out. In J. A. Graber, J. Brooks-Gum, &
A. C. Petersen (Eds), Transitions Through Adolescence: Interpersonal Domaig
and Context (pp. 285-321). NJ: Lawrence Erlbaum Associates, Publishers.

 

Lerner, R. M., Lemer, J. V., von Eye, A., Ostrom, C. W., Nitz, K., Talwar-Soni, R., &
Tubman, J ., G. (1996). Continuity and discontinuity across the transition of early
adolescence: A developmental contextual perspective. In J. A. Graber, J. Brooks-

Gunn, & A. C. Petersen (Eds), Transitions Through Adolescence: Interpersonal
Domains and Context (pp. 3-22). NJ: Lawrence Erlbaum Associates, Publishers.

128

MacCallum, R. C., Browne, M, W., & Sugawara, H, M. (1996). Power
analysis and determination of sample size for covariance structure modeling.
Psychological Methods. 1 (2), 130-149.

McBride, C. M., Curry, S. J., Cheadle, A., Anderman, C., Wagner, E. H., Diehr, P., &
Psaty, B. (1995). School-Level Application of a Social Bonding Model to
Adolescent Risk-Taking Behavior. Joumafil of School Health, 65 (2), 63—68.

McBrien, J. L., & Brandt, R. S. (1997). The language of learning: A gpide to education
terms. Alexandria, VA: Association for Supervision and Curriculum

Development.

McDonald, M., & Wright, N. E. (2002). Cigarette smoking and the disenfranchisement
of adolescent girls: A discourse of resistance? Health Care for Women
Intemtional. 23 (3), 281-305.

Mouton, S. G., Hawkins, J ., McPherson, R. H., & Copley, J. (1996). School attachment:
Perspectives of low-attached high school students. Educational Psychology, 16
(3), 297-304.

Najaka, S. B. (2001). A meta-analytic inquiry into the relationship between risk factors
and problem behavior. Dissertation Abstgcts InternationzLSection A:
Humanities and Social Sciences, 62 (l-A), 340-349.

Partnership for Academic and Career Education (1992). Tech prep implementation: Self
Assessment Inventory. http://www.edrs.com/member§/sp.cfm?AN=ED362636

Pryor, C. B. (1994). F amily-school bonding and student success. Paper presented at the
Annual Meeting of the American Educational Research Association, New
Orleans, LA, April 4-8.

Pyper, J. R., Freiberg, H. J., Ginsburg, M., & Spuck, D. W. (1987). Instrument to
measure school climate. In L. W. Barber (Ed.), School Climate (pp. 87-96).
Bloomington Center on Education, Phi Delta Kappa.

Robertson, L. M., Harding, M. S., & Morrison, G. M. (1998). A comparison of risk and
resilience indicators among Latino/a students: Differences between students
identified as at-risk, learning disabled, speech impaired and not at-risk. Education
and Treatment of Childrem 21 (3), 333-353.

Rutter, M. (1979). Protective factors in children’s responses to stress and disadvantage.
In W. Kent & J. E. Rolf (Eds), Primary prevention of psychopathology (Vol. 3):
Social competence in children (pp. 49-74). Hanover, NH: University Press of
New England.

129

Schaps, E., & Battistich, V. (1991). Promoting health development through school-
based prevention: New approaches. In E. Goplerude (Ed.), Preventing Adolescent
Dug Use: From Theory to Practice (pp. 127-181). Washington, DC: US.
Department of Health and Human Services.

Simons-Morton, B. G., Crump, A. D., Haynie, D. L., & Saylor, K. E. (1999). Student-
school bonding and adolescent problem behavior. Health-Education-Researcha14
(1), 99-107.

Smentana, J. G. (1995). Conflict and coordination in adolescent-parent relationships. In
S. Shulman (Ed.), Close Relationships and Socio-emotional Development (pp.
155-184). NJ: Ablex.

Smey-Richman, B. (2001). School Climate and Restructuring for Low-Achieving
Students. Elementary School Joumarla95 (4), 427-452.

Squires, J., Bricker, D., Heo, K., & Twombly, E. (2001). Identification of social-
emotional problems in young children using a parent-completed screening
measure. Early Childhood Research Quarterly. 16. 405-419.

Storino, M. H. (2001 ). The relationships among problem behavior, resiliency,
and enculturation in a sample of Hispanic youth. Dissfltion Abstracts
Internaaiorall. Section A: Humanities and Social Sciences. 62 (6-A): 2010-2015.

Sunah, K. (2000). The effects of parent bonding, school bonding, belief on the
structure of problem behaviors in elementary school-age children. Dissertation

Abstrafi International. Section A: Humanities and SociaLSciences. 61 (6-A):
2467.

Tagiuri, R. (1968). Organizational Climate: Explorations of a Concept. Boston, MA:
Division of Research, Graduate School of Business Administration, Harvard
University.

Trickett, E.J. and Moos, R H., _C_lassroom Environment Scale Manual. 3rd Edition.
Consulting Psychologists Press, Palo Alto, CA. 1995.

Valverde, S. (1987). A comparative study of Hispanic high school dropouts and
graduates: why do some leave school early and some finish? Education and
Urban Society. 19. 320-329.

Wang, M. C. (2000). Research on School Effects in Urban Schools. Sociology of
Education. 82 (23), 463-468.

Welsh, W. N., Jenkins, P. H., & Greene, J. R. (1998). School culture, climate, and
violence: safety in middle schools of the Philadelphia public school system.

130

Williams, S. (1987). A comparative study of Black dropouts and high school
graduates: To go urban public system. Education and Urban Society. 19,
311-319.

131