.
A
“3 mm
a 11L.” cu}
‘ a.
dmlc

.
.7 4.2, u
a):

.2

ﬁmw.‘
I’ddﬁml :3 _

Hwy.» 5.
ﬁv‘l"
S.

i.“
1. .512
112.5

3

49.61;! . .,
u .3. x4.
zéawﬂﬂna a.
91!}; 5..
....,...:.\..
1..

{launvﬁ .

 

 

 

1.... x. . . Ta... 2:
5&3 .uﬁﬁxé...¥§ gag? fﬁgggwﬁﬂ . 4 V. 4

I :li...

 

 

III
I

 

THESIS

20:)!

This is to certify that the
dissertation entitled

HOW DO DISTRICT AND SCHOOL CONTEXT S RELATE TO
TEACHERS’ OPPORTUNITIES TO LEARN?
MULTIPLEVEL ANALYSIS OF THE 2000 SASS DATABASE

presented by

KYOUNG-OH SONG

has been accepted towards fulfillment
of the requirements for the

Ph.D degree in Education Policy

 

 

bow?) Wu

Major Professor’s Signature
angloe

 

 

Date

MSU is an Afﬁrmative Action/Equal Opportunity Ins titution

 

 

 

LIBRARY

Michigan State

University

 

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

M A°R61192261t9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRC/DateDue.indd-p.1

_._____

HOW DO DISTRICT AND SCHOOL CONTEXTS RELATE TO
TEACHERS’ OPPORTUNITITES TO LEARN?
MULTILEVEL ANALYSIS OF THE 2000 SASS DATABASE

By

KYOUNG-OH SONG

A DISSERTATION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements
For the degree of

DOCTOR OF PHILOSOPHY

Education Policy
College of Education

2006

ABSTRACT

HOW DO DISTRICT AND SCHOOL CONTEXTS RELATE TO
TEACHERS’ OPPORTUNITIES TO LEARN?
MULTILEVEL ANALYSIS OF THE 2000 SASS DATABASE

By

KYOUNG-OH SONG

“Teachers’ opportunities” to learn (TOL) has emerged as one of the most
important policy issues in the context of standards based reform. Consequently, policy
and school contexts that inﬂuence TOL are clearly the next critical issue on the agenda.
This dissertation investigates the relationship between district and school contexts, and
TOL, based on a conceptual frame that derives from the multi-perspective embedded
context model developed by McLaughlin and Talbert (2001). This conceptual approach
contributes to a better understanding about the relationship between multiple and
embedded “real” school contexts and TOL, through conceiving of “context” as a complex
and interactive construct.

Using a series of HLM/3L (3-level Hierarchical Linear Modeling) statistical
analyses of a large dataset that is nationally representative, the 1999-2000 Schools and
Stafﬁng Survey, this dissertation provides several noticeable ﬁndings regarding the
relationship between embedded district and school contexts and TOL. First, this study
found that a substantial proportion of the variation in TOL is attributed to two multi~level

district and school contexts. In particular, district contexts apparently contribute more to

the variability of TOL than school contexts. Furthermore, it may be that district
administrative contexts relate to the effects of professional learning communities on TOL.
Second, the ﬁndings of this study show that diverse social system contexts,
administrative and organizational contexts, and institutional contexts factors at the school
and district level have complex and interactive relationships with TOL. Finally, this study
found that institutional contexts identiﬁed in this study by three factors - community
circumstances, parents’ socio-economic conditions, and diversity of students’ race — are
signiﬁcantly related to TOL.

For policy researchers, this dissertation suggests that adopting the multi-
perspective embedded context model as a conceptual framework would provide a more
complete understanding of how multiple and embedded contexts shape TOL. This
dissertation also provides three implications for policy makers and educational leaders.
First, district units have to be given more attention in terms of the improvement of TOL.
In addition, the importance of institutional contexts must be recognized and discussed.
Finally, some speciﬁc contextual factors thought to inﬂuence TOL in the literature are
suspect. There is a need for cautiously verifying the effectiveness of these contextual

factors on TOL.

C0pyright by
Kyoung-oh Song
2006

This dissertation is dedicated to my parents,
Song, Nam-yea! and Kim, Hyungsim,

who, through their example, have shown their children the diligent life is valuable.

ACKNOWLEDGEMENT

My journey to completing this dissertation was possible with that help of many
people whom my Lord Jesus Christ sent, who have guided me, supported me, and prayed
for me.

First of all, I would like to express my deepest appreciation to Dr. Gary Sykes
who was my graduate advisor and dissertation chair. His steadfast belief in me kept me
working toward accomplishing this goal through the times that were tough. From my ﬁrst
year in graduate school, Gary constantly encouraged me and always reminded me that
learning is challenging work. My experience working with him through each step of this
dissertation — conception, framework, analysis, and write-up — as well as graduate studies
have been precious for my development as a novice scholar. I want Gary to know how
fortunate I feel to have been advised by him.

I am also grateﬁil for the commitment and support of my dissertation committee.
Dr. David Arsen guided me from the beginning of the dissertation in his research
practicum class, engaging my ideas and sharing the pleasures of crafting research. Dr.
Brian DeLany asked thoughtful and practical questions that helped ensure that my
dissertation communicated to educational practitioners and policy researchers. His
support for my personal and professional development has meant a lot. To Dr. Kim Maier,
I owe a special thanks for the time spent with me in providing statistical advice during
my data analysis and detailed editorial review. I am proud and honored to have had them
as my dissertation committee.

lam also appreciative to the Graduate School of Education at Michigan State

vi

University for a Dissertation Completion Fellowship that provided me with ﬁnancial
support. In addition, I owe thanks to Dr. David Plank, the former director of the
Educational Policy Center at Michigan State University that allowed me to use the data
for this study.

During the time of my graduate study, I had the support of numerous colleagues
and people in Michigan. Cohort members of the Educational Policy Program, Yongmei,
Tara, and Kamila, were not only academically but also psychologically helpful in getting
me through graduate school in the USA. We worked our way through courses,
comprehensive exams, and dissertations together. Through the various stages of my
writing, I have consulted with writing consultants in the MSU Writing Center: Kym
Buchanan, Michael Sherry, Ann Sherry, and Steve Sharra. I have beneﬁted from their
feedback. To Ann, I owe special thanks for sage advice and editorial suggestions. I also
appreciate encouragement and friendship provided by Korean colleagues at Michigan
State University. Especially, I express special thanks to Jinyoung Choi, J eonghee Noh,
Wangjun Kim and his wife Bun—Jung Lee, Won-Pyo Hong and Kwanghyun Lee. They
have been caring and generous collaborators. In addition, I thank the congregation of the
Lansing Korean Catholic Church, Father Back, and Father Kim. The spiritual community
kept me grounded and sustained my journey to the PhD.

Many people in Korea helped me in diverse ways. My parents have always
encouraged me to pursue my dreams, instilled in me the belief that I could achieve my
dreams, and then helped me do so each and every step of the way in my life. Without
their love and support, I could not have made this journey. I am also appreciative to Dr.

Chong-Jae Lee who was my advisor in Korea. He showed strong belief in my personal

vii

and intellectual ability to complete this journey.

My friends outside academia cheered me on in my academic endeavors. I really
would like to express my thanks to Hyeryung Jung, Su-jung Lee, Minjo Kim, Jungeun
Lee, and Yunjung Hwang, for providing much needed support at important times.
Hyeryung has provided unfailing ﬁ-iendship as my long-distance cheering person. And
my soul mate Bo-Young was the central force from my very ﬁrst step in a doctoral

program to the ﬁnal moment. She always reminds me what is truly important in life.

Finally and most importantly, I dedicate this humble work to my Lord Jesus

Christ who has given me love and strength so genuinely and unselﬁshly throughout all

my life.

viii

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES xii
LIST OF FIGURES xiii
CHAPTER I
INTRODUCTION 1
The Problem Statement 4
Research Questions 5
Research Design Overview 6
Signiﬁcance of the Study 7
Overview of the Dissertation 8
CHAPTER II
LITERATURE REVIEW (PART I) 10
Overview of the Literature on Inﬂuences on TOL 10
Contemporary Approaches to TOL: Strengths and Weaknesses 17
“Policy” as Source of Inﬂuence on TOL 18
Approaching TOL through Professional Community 21
Another Conceptual Approach:
the Multi-perspective Embedded Context Model 24
Theory and Method must be ‘Integrated’ 24
Multiple Theoretical Perspectives 25
Nested Contexts vs. Embedded Contexts 28
CHAPTER III
LITERATURE REVIEW (PART II) 32
TOL Deﬁnitions and Indicators 32
Deﬁning TOL 32
Literature on Effective TOL Indicators 39
Hypotheses about the Relationship between
District and School Contexts and TOL 48
The Rationale for Hypothesis 1 51
The Rationale for Hypothesis 2 52
The Rationale for Hypothesis 3 58
The Rationale for Hypothesis 4 63
The Rationale for Hypothesis 5 65

 

ix

CHAPTER IV

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

RESEARCH METHODS 67
Data Sources 68
Measures 69
Measures of TOL 70
School Level Variables 73
District Level Variables 78
Teacher Background Variables 83
Statistical Modeling 84
HLM Analysis for Testing Hypothesis 1 85
HLM Analysis for Testing Hypothesis 2 and 3 87
HLM Analysis for Testing Hypothesis 4 89
Analytical Model for the Study 92
CHAPTER V
RESULTS AND DISCUSSION 95
Descriptive Statistics 98
Correlations 101
Correlations among Outcome Variables 101
Correlations of Teacher Background to Three TOL Variables 102
Correlations of School Level Variables to Three TOL Variables _____________________ 103
Correlations of District Level Variables to Three TOL Variables ____________________ 104
HLM Model Results 105
Results 1: Results of the First Hypothesis Test 106
Results 2: Results of the Second Hypothesis Test 109
Results 3: Results of the Third Hypothesis Test 121
Results 4: Results of the Fourth Hypothesis Test 125
Results 5: Results of the Fifth Hypothesis Test 129
Summary Remarks 132
Discussion 133
Discussion 1: Relationship between Multilevel School Contexts and TOL__,133
Discussion 2: Relationship between Contextual Factors and TOL ................. 136
Discussion 3: Variability across Three TOL Indicators 149
CHAPTER VI
CONCLUSIONS 151
Strengths and Limitations 151
Implications of the Study 154
Implications for Policy Research 154
Implications for Policy and Practice 157

 

APPENDICES

 

Appendix A. 1999- 2000 SASS Variables and Items used in Data Analysis ................ 161
Appendix B. 1999- 2000 SASS Bivariate Correlation Matrix for

items presented in HLM Analysis 167
REFERENCES 170

 

xi

Table 4-1

Table 4-2

Table 4-3

Table 4-4

Table 5-1

Table 5-2

Table 5-3

Table 5-4

Table 5-4

Table 5-6

Table 5-7

Table 5-8

LIST OF TABLES

 

 

 

 

 

Dependent Variables 73
School Level Variables 78
District Level Variables 82
Teacher Background Variables 83
Descriptive Statistics 99
Unconditional HLM Analysis for Three TOL Variables 108

 

Intercepts as Outcomes HLM Analysis (I): All teacher, School, and
District Level Variables and Content-focus
as the Dependent Variables 118

Intercepts as Outcomes HLM Analysis (H): All teacher, School, and
District Level Variables and Active learning
as the Dependent Variables 119

 

Intercepts as Outcomes HLM Analysis (III):A11 teacher, School, and
District Level Variables and Total hours
as the Dependent Variables 120

Cross-level Interactive Effects between District-level Accountability
related to Performance and School-Level Professional Leaming
Community on Active Leaming 127

Cross-level Interactive Effects between District-level Performance
Indicators and School-level Professional Learning Community
on Active Learning 128

Different Relationships with Contextual Factors
across three TOL variables 131

 

xii

LIST OF FIGURES

Figure 3-1 TOL Deﬁnition used for this Study 38

 

Figure 3-2 Main Notions of the Model and Five Hypotheses 49

 

Figure 4-1 A Model used to Analyze the Relationship between District and
School Contexts and TOL 94

Figure 5-1 Roadmap representing the Structure of How the Chapters proceed .......... 95

Figure 5-2 Contextual Factors Proposed in This Study and Six Themes ____________________ 137

xiii

CHAPTER I
INTRODUCTION

Over the past two decades, some educational scholars have focused on “teachers’
opportunities” to learn (TOL) as a cornerstone of successful implementation of standards
based reform. They argue that educational reform that requires teachers to carry out the
demands of high standards in the classroom would succeed if and when teaching quality
increases through TOL (AFT, 1995; Ball & Cohen, 1999; Corcoran, 1995; Darling-
Hammond & Sykes, 1999; Thompson & Zeuli, 1999; Wilson & Beme, 1999). For
example,

Without professional development, school reform will not happen... The nation can
adopt rigorous standards, set forth a visionary scenario, compile the best research
about how students learn, change the nature of textbooks and assessment, promote
teaching strategies that have been successful with a wide range of students, and
change all the other elements involved in systemic reform. But, unless the classroom
teacher understands and is committed to the plan and knows how to make it happen,
the dream will come to naught (AFT, 1995, pp. 1-2).

A common logic of these arguments is that when teachers acquire a deep
understanding of the content knowledge and skills they teach through effective and
continuous TOL, teaching practices and student learning will be improved. That is,
successful instructional reform relies on teachers’ knowledge and skills, and their
knowledge and skills rely on effective and ongoing TOL (Cohen & Hill, 2001).

Policy makers also have increasingly recognized the importance of TOL in the
successful implementation of reform efforts. The recent federal education legislation, the

No Child Left Behind (NCLB) Act requires highly-qualiﬁed teachers who have in-depth

content knowledge in the ﬁelds that they teach, reﬂecting the belief that teachers are

critical to any effort for improving student achievement. A logical extension of this
assumption is that TOL is a primary means for upgrading teachers’ practices and for
promoting student learning. Therefore, the need for an increased focus on TOL for
educational reform is prompted not only by educational researchers, but also by policy
makers.

The emphasis on TOL as a prominent instrument for successful reform has led to
educational researchers’ interest in contexts that inﬂuence the effectiveness of TOL. They
have begun to realize the potential of contexts as critical inﬂuences for implementing
effective TOL. Smylie (1995) notes, “If we wish to improve schools as places for
teachers to learn, we need to be able to identify those workplace conditions that promote
or constrain learning” (p. 94). A large literature on inﬂuential contexts on TOL can be
divided into two distinctively different approaches to effective TOL.

The ﬁrst approach to TOL focuses on relatively local teacher community as a
critical inﬂuence on TOL. It argues that particular norms and practices of such a local
teacher community shape learning conditions for TOL, assuming that each school
constructs its own teacher community (Lieberman 1994; Little 1982, 1990a, 1999b, 2001;
Lord, 1994; Perry, 1996). For example, Little (1982) shows that norms of collegiality and
experimentation in a teacher community have a signiﬁcant effect on teacher learning on
the job. Teachers are more likely to participate in TOL when they work collaboratively on
matters of curriculum and instruction and when professional risk taking is encouraged. In
another paper, Little (1990a) helps us think about how teacher community conditions
inﬂuence TOL, using her distinction of "strong ties" and "weak ties" among teachers in

the same school. In communities of "strong ties," teachers work collaboratively on

matters of instruction and develop shared understanding. In such communities, teachers
are motivated to pursue professional learning. By contrast, in communities of "weak
ties," teachers experience strong individual autonomy and isolate from peers. In such
communities, teachers’ work and learning are discouraged.

Another approach to TOL focuses on more distal policy as a sources of inﬂuence
on TOL. It argues that coherent and effective policy strategies at the district or state level
can lead to high-quality TOL, expecting signiﬁcant effects of the policy mechanisms on
local practices (Cohen & Hill, 2001; Corcoran, Passantino, & Gerry, 2000; Desirnone,
Porter, Birrnan, Garet, & Yoon, 2002; Firestone, Mangin, Martinez, & Polovsky, 2005;
Knapp, 2003; Wilson, Darling-Hammond, & Berry, 2001). One representative study
supporting this approach is the Eisenhower study that demonstrates statistically
signiﬁcant relationships between self-reports of TOL and district policy strategies, such
as alignment, multiple funding sources, continuous efforts for instructional improvement,
and district and teachers’ participations in TOL (Desimone et al., 2002). In this study,
districts’ coherent visions and effective management of TOL led to reported high quality
TOL, such as participation in active learning and content-focused learning. Wilson and
colleagues (2001) also show that a state’s consistent long-term investment to enhance the
quality of the teaching force have a signiﬁcant effect on teachers’ engagement in learning
opportunities and changes in teaching practices. Knapp (2003) concludes that through
effective policy instruments, such as mandates, capacity building, inducements, and
system changing, districts or states can play an important role in stimulating, guiding, and

supporting TOL.

The Problem Statement

Even when taken together, these two approaches to TOL (teacher community
inﬂuence and policy inﬂuence) do not provide a sufﬁcient description of the contexts that
inﬂuence TOL. These two approaches still assume that context as inﬂuence on TOL is
one-dimensional; and is most often associated with a particular condition at a single level
such as policy strategies at the district level, teachers’ collegial relationship at the school
level, or students’ socioeconomic background.

If we can deﬁne context as "the whole situation, background or environment
relevant to some happening," (Grossman & Stodolsky, 1994, p.181), then the contexts in
which teachers work and learn are multiple and embedded (McLaughlin & Talbert 1990,
2001; Scribner, 1999; Sykes, 2002; Talbert & McLaughlin, 1999). That is, teachers are
embedded within the most proximal zone, the school setting, while schools are embedded
within a broader administrative zone such as districts and states, and formal educational
systems are embedded within larger institutional environments (McLaughlin & Talbert,
2001). In these multiple embedded school contexts, contextual inﬂuences may be diverse
and overlapping (McLaughlin & Talbert 1990, 2001; Talbert & McLaughlin, 1999).

Each of these multiple and embedded contexts can directly affect teachers’
beliefs and thinking about their own learning, and interact with one another to shape
teachers’ work. For example, social relationships with colleagues in the schools may
affect TOL. TOL may be also constructed by the educational policy system and school
sector; and by larger institutional and community culture; by higher education,
professional associations, and networks. Subject departments may shape TOL as well.

Further, TOL is constructed by the complex interplay of contexts that mutually reinforce

and/or compete with another (McLaughlin & Talbert, 2001). Therefore, contextual
inﬂuences on TOL should be conceived as highly complex and diverse.

Although contemporary approaches to TOL have made signiﬁcant contributions
to our knowledge about inﬂuences of the single level contexts (i.e., district or school),
these approaches afford little opportunity to account for these diverse and interactive
inﬂuences on TOL. A more complete understanding of the relationship between contexts
and TOL requires an approach that is capable of capturing the multifaceted and
interactive inﬂuences of these embedded contexts.

The current study proposes a more promising conceptual approach framework to
explore the relationship between embedded district and school level contexts and TOL,
drawn from important assumptions of McLaughlin and Talbert’s (2001) multi-
perspective embedded context model. This model considers the complexity of the forces
inﬂuencing the life of a teacher - not as a simple set of one-to-one connections, but
through several different lenses. It is also able to demonstrate complex interactions
among layers of contexts and diverse inﬂuences of each context. Therefore, the multiple
perspective embedded model is a particularly useful framework for studying school
environments and their inﬂuences on TOL. This study will claim that a more systematic
and comprehensive approach capable of capturing the diverse and interactive contextual
inﬂuences might provide a more complete insight into the relationship between TOL and

contexts.

Research Questions

The research interest of this study is how embedded district and school contexts

.i ll

relate to teachers’ opportunities to learn (TOL). The speciﬁc research questions are as
follows:

1. What district level contextual factors inﬂuence TOL? How do they do so?

2. What school level contextual factors inﬂuence TOL? How do they do so?

3. How do these district and school level factors relate to one another and how do

these relationships affect TOL?

This study proposes that these research questions could be well answered using
the multi-perspective context embedded model. Drawn from the main notions (multiple
theoretical perspectives and embedded contexts) of this model, the current study develops
and tests the hypotheses regarding the relationships between district and school contexts

and TOL.

Research Design Overview

This study involves a cross-sectional examination of the relationship between
embedded district and school contexts and TOL, using nationally representative data
from the 1999-2000 Schools and Stafﬁng Survey (SASS), a database developed and
managed by the National Center for Educational Statistics (N CES). Data from the SASS
2000 is used to address the hypotheses relevant to the issue of what and how district and
school level factors inﬂuence TOL.

A series of HLM/3L analyses (three-Level Hierarchical Linear Modeling) is
performed to investigate variations in TOL that occur within school, between schools,
and between districts. It also allows the examination of cross-level interactive effect

between district and school levels.

Significance of the Study

First, this study provides policy makers and educational leaders with more
comprehensive information about the relationship between district and school contexts
and TOL. By simultaneously taking into account the effects of these two contexts on
TOL, this study explores roles of school level contextual factors advocated by traditional
social system theorists while it points out conditions under which the role of district
policy may predominate. It goes beyond prior research that tends to treat district and
school contextual inﬂuences as mutually exclusive. A more comprehensive understanding
of the relationship between district and school contexts and TOL would help policy
makers to consider effective and eﬂicient policy strategies to create resources and
conditions for TOL. Furthermore, it would enable educational leaders to guide and
support teachers’ engagement in learning opportunities, by modifying social and
organizational conditions as well as incentive structures for TOL.

Second, this study provides a powerful case to explain school contexts as highly
interactive and quite sophisticated. While the data used for this study do not permit
complete modeling of the embedded context formulation, the study is illustrative of the
general approach that conceives of “context” as a complex construct.

Finally, this study complements prior quantitative research on contexts for TOL,
by employing the statistical technique of hierarchical linear modeling (HLM). The prior
research on school contexts on TOL that used OLS regression methodologies fails to
simultaneously estimate each district-and school-level effect on TOL. This study moves
beyond these previous methodological difficulties of analyzing these multilevel

phenomena. Hierarchical linear modeling permits a more accurate and direct estimation

of the effects of the embedded district and school contexts on TOL simultaneously.
However, reliance on cross-sectional data creates problems of inﬂuence that will be

discussed in Chapter 6.

Overview of the Dissertation
Chapter 1 presents a brief statement of the background and problem for the study,
together with the research questions, their signiﬁcance, and an overview of the research
design.

Chapter 2 reviews literature about inﬂuences on TOL and investigates the
strengths and limitations of the contemporary approaches to TOL. To address the
shortcomings of these approaches, this chapter presents another approach to TOL, drawn
from McLaughlin and Talbert’s multi-perspective embedded context model.

Chapter 3 discusses the deﬁnitions of TOL and indicators of effective TOL. Then,
based on the main assumptions of the multi-perspective embedded context model, I
develop ﬁve hypotheses to examine the relationship between district and school contexts
and TOL contexts for this study.

Chapter 4 discusses the methodology employed in this study, including a brief
description of data sources for this study, the dependent variables and independent
variables; and how these variables are selected and constructed for this study. I also
provide analytical techniques used in this study.

Chapter 5 presents ﬁndings from the HLM analyses employed to examine the
variations in TOL within-school, between-schools, and between-districts. This chapter

represents the results of hypotheses tests regarding district and school factors related to

TOL and the results of hypotheses tests regarding interactive relationships between
district and school factors. Also included is a discussion of study limitations.

Chapter 6 discusses the strengths and limitations of this study. Then, I provide
some implications for policy research, and policy and practices. I conclude my study with

ﬁnal thoughts about effective approaches to TOL.

CHAPTER II
LITERATURE REVIEW (PART I)

The literature review of this study is divided into two chapters. The ﬁrst chapter
(Chapter H) initiates a review of literature on inﬂuences on TOL, together with a new
model for exploring inﬂuences on TOL. Then, the next chapter (Chapter IH) provides a
review of literature on TOL deﬁnition and indicators, working out speciﬁc hypotheses
regarding the relationship between district and school contexts and TOL.

This chapter (Chapter H) consists of three sections. The ﬁrst section presents a
general overview of inﬂuences on TOL. The second section explores the two
contemporary approaches to TOL: the approach to policy as source of inﬂuence on TOL,
and the approach to TOL through professional community. It also addresses conceptual
and methodological limitations of these approaches. The third section introduces another
conceptual approach, the multi-perspective embedded context model that speciﬁes the

relation between contexts and TOL.

Overview of the Literature on Inﬂuences on TOL
During the past two decades, several scholars have explored the contexts that
inﬂuence TOL. They have articulated a number of inﬂuential factors or conditions for
TOL, assuming that TOL does not take place in a vacuum.
Studies from general social organizational theories (e.g., Little, 1982) have
identiﬁed school level social or organizational factors affecting TOL. One facilitating
condition mentioned in this line is collegial relationship among teachers and professional

norms in a particular school. For example, Little (1982) found that norms of collegiality

10

and experimentation have a signiﬁcant effect on teachers’ learning on the job. Simply put,
teachers are more likely to participate in TOL when they feel the support of colleagues
and when professional risk taking is encouraged. Collaboration among teachers also has
been regarded as an inﬂuential factor for teacher learning. Smylie (1995) indicates, “One
of the most salient conditions [that may promote learning in the workplace] is
opportunities for individuals to work with and learn from others on an ongoing basis” (p.
103). The level and type of teacher collaboration are signiﬁcantly associated with the
effectiveness of TOL (Little, 1990b; Rosenholtz, 1989; Scribner, 1999; Smylie, 1995).
For example, Rosenholtz (1989) found that sharing, help-giving and help-receiving, and
joint work among teachers related to instruction and students’ work, yield statistically
signiﬁcant effects on teachers’ reports of opportunities to learn in their schools. The more
teachers have opportunities to work together, the more they tend to be engaged in
learning opportunities (Darling-Hammond & McLaughlin, 1995).

More recent research in this line focuses on teacher communities as critical
inﬂuences on TOL. Studies using this concept assume that each school constructs its own
teacher community that encourages or inhibits teacher learning. Each community
functions in somewhat distinctive structural and cultural contexts, takes varied forms, and
makes a range of contributions to teachers’ learning. Some teacher communities enhance
teachers’ motivation and commitment to teachers or student learning while other
communities discourage professional learning (Little, 1999b; McLaughlin, 1994).
McLaughlin and Talbert (2001) identify three kinds of teacher communities: weak
teacher community, strong traditional teacher community, and strong learning teacher

community. Weak communities value individualism and low expectations about teachers

11

and student learning. In these communities, teachers’ work and learning are isolated and
discouraged. Not all strong communities encourage teachers’ learning. Some strong
communities value conservatism about learning and instruction. In such communities,
teachers’ learning Opportunities are often incoherent and not directly associated with
reform goals. In contrast, strong learning communities value collegiality norms, mutual
respect, and faculty responsibility for the accomplishment of learning. In these learning
communities, teachers share the sources of motivation and support for their professional
learning and engage in new practices and learning consistent with the reform goals.

Some of the literature on school level factors for TOL focuses on principal
leadership as a critical inﬂuence for TOL. One area in which principals have particular
inﬂuences on TOL relates to the degree to which the principal supports the development
of shared goals that focus on teacher learning and student performance and directs
resources toward attaining those goals (Sykes, 2002; Youngs & King, 2002). Sykes
(2002) identiﬁes this leadership as “learning-centered principal leadership”, arguing that
these principals place a high value on teacher learning, seek to connect professional
development to the common mission, and consider their role to be the “learner in chief”
responsible for creating conditions that encourage continuous professional growth in
teachers. Youngs and King (2002) add another area of principal leadership that creates a
school culture and structure cultivating teachers’ learning. In the research on principal
leadership for TOL in urban elementary schools, Youngs and King (2002) found that an
effective principal can enhance school’s capacity to promote TOL by creating structures
or organizing resources for teachers to collaborate and reﬂect on their practices,

establishing shared commitment to school goals, and connecting teachers to external

12

expertise. While much research points to principals as being key leaders in TOL, it is also
true that others can take on facilitative leadership and support roles. Cox (1993) notes
that while principals seem to play an important role in clarifying expectations and goals
about teacher learning and instructional practices, principals who often know more about
a speciﬁc practice and facilitate collaborative school cultures can effectively support
teachers in their attempts to participate in TOL and change their instructional practices.
Such a “facilitative” leadership provides teacher feedback, promotes a positive learning
climate, and encourages a sense of community in the school (Clement & Vandenberghe,
2000; Scribner, 1999).

Some studies on school level inﬂuences on TOL also underscore teachers’
autonomy as another important school organizational factor facilitating TOL (Clement &
Vandenberghe, 2000; Poulson & Avramidis, 2003; Smylie, 1988). In particular,
involving teachers in making decisions that affect professional development and
instructional policy can provide an opportunity for them to examine their current
assumptions and practices and increase their responsibilities for professional
enhancement (Smylie, 1988). Clement and Vandenberghe’s research (2000) shows that
working in a school that recognizes autonomous dimensions of TOL makes teachers want
to engage in their learning opportunities and get better at what they do. Furthermore,
teachers’ autonomy is directly linked to collegiality, so that the combination of autonomy
and collegiality is more likely to lead to organizational incentives for teacher learning
opportunities (Clement & Vandenberghe, 2000; Poulson & Avramidis, 2003).

Although school level social or organizational conditions have long been

recognized as signiﬁcantly inﬂuential factors in TOL, increasing attention is being paid

13

to the broader context of districts, states, and federal level. In particular, recent literature
that considers TOL as a key instrument for instructional reform underscores the critical
role of district and state policies stimulating and supporting TOL.

Some studies in this line have investigated whether the features of systemic
reform that emphasize clear and ambitious standards, aligned assessment systems, and a
strong accountability framework have a signiﬁcant effect on teachers’ engagement in
learning opportunities and changes in teaching practices. For example, David,
McDiarmid, and Corcoran (2000) studied Kentucky's reform efforts to support TOL as
part of its systematic reform. In the early stage of the reform, educational reformers in
Kentucky expected that strong guidance from the state’s standards and accountability
systems have an effect on TOL. State policy makers assumed that if performance goals
are clear, and accountability for student learning is available, teachers would demand
high-quality forms of TOL in order to achieve the goals. However, David et a1. (2000)
found that Kentucky teachers have adhered to familiar forms of professional development
and have been reluctant to embrace promising approaches, such as mentoring and
coaching.

Berry and colleagues’ (2003) study shows similar results with Kentucky’s case.
Employing teacher survey and case studies in six states, the researchers found that despite
strong state accountability systems, teachers did not change in their learning and practices
since they did not fully understand the accountability systems in which they worked. That
is, teachers did not necessarily believe that all students could not reach the academic
standards posed by the accountability tests, nor did they regard current professional

development as better than what they had experienced in the past. The researchers

l4

conclude that teachers’ professional learning would be improved when they sufﬁciently
understand the features of accountability systems and what they need to know and
implement to inﬂuence the results in a serious way.

Other research on Connecticut’s reforms found that the state’s long-term
investment to enhance the quality of the teaching force has a signiﬁcant effect on
teachers’ engagement in learning opportunities and changes in teaching practices (Wilson
et al., 2001). To create greater capacity for the teaching force, Connecticut built new
infrastructure to guide and support teachers, such as mentorship programs for beginning
teachers, and the Beginning Educator Support and Training system, while it required a
certiﬁcation system and licensing standards for teachers. The districts were also required
to create linked plans for incentives, professional development, and evaluation of
teachers’ work. The effects of these state policy actions were powerful. Almost half of
Connecticut's teachers had participated in the program as beginning teachers, mentors,
and assessors and three quarters found that it helped them to reﬂect on their teaching
(Fisk, 1997, cited in Knapp, 2003).

Some of the literature in this line underscores the effects of district policy
mechanisms on high quality TOL. That is, some scholars argue that district policy
strategies that combine strong district mandates with creation of local capacity, and that
increase districts’ overall investment in professional learning tend to enhance TOL
(Elmore & Burney, 1999; Knapp, 2003; Togrreri & Anderson, 2003). For example, New
York City Community School District 2 mandated the direction of TOL, and the system
of accountability for performance in relation to what students were learning. The district

also reallocated substantial resources to support these mandates and sent consistent and

15

strong signals to all about what teachers needed to know and be able to do. At the same
time, the district has strengthened capacity at all levels in the educational system,
including school principals and school-based staff developers and created new structures
for supporting teaching, such as the Professional Development Laboratory, and inter-
visitation among classrooms. The district has also spent a speciﬁc proportion of the
budget as an expression of the priority it places on teacher learning. Consequently,
teachers in this district had numerous and varied opportunities to encounter ideas about
improved teaching, and student achievement have showed sustained improvements
(David & Shields, 2001; Elmore & Burney, 1999; Fink & Resnick, 2001).

In addition to the effects of educational policy mechanisms, some scholars argue
that districts’ organizational conditions can have a strong effect on TOL as well. That is,
district climate has been considered as a key organizational factor. For example, Spillane
and Thompson (1997) found that trust among teachers and between teachers and
administrators in a district are associated with TOL. In their research, administrators did
not trust teachers to initiate change. Teachers did not trust administrators to support
change. Teachers did not trust each other to implement meaningful change. In such a
district, distrust diminished opportunities for meaningﬁrl and substantive discussion about
instruction with their colleagues and collaboration among educators.

District leadership also makes an important contribution to learning for teachers
(Elmore & Burney, 1999; McLaughlin, 1994; Rosenholtz, 1989). Rosenholtz’s study
shows that a learning-minded and instruction-oriented district leadership can stimulate
teachers’ learning opportunities. In this study, the “moving superintendents,” who had

strong district goals for improvement, conducted active performance monitoring, and

16

carefully selected school principals and teachers were more likely to encourage and
create plentiful learning opportunities for principals and teachers. In these districts,
teachers’ engagement in their learning opportunities increased. By contrast, the “stuck
superintendents,” who directed their districts via uncertain and wavering goals of school
improvement did not make efforts to advance neither their own knowledge, nor that of
principals and teachers. In these districts, teachers’ engagement in learning opportunities
was discouraged.

In sum, the literature on TOL provides numerous lists of social and
organizational factors at the school level, state or district policy strategies, district
administrative conditions that shape TOL. Although these studies have made signiﬁcant
contributions to our knowledge about inﬂuences of a particular condition (such as
schools’ ﬁscal resources or districts’ leadership), these studies, which underscore single
level approach to TOL do not account for complex and interactive contextual inﬂuences
on TOL. In the next section, this study examines strengths and weaknesses of these

contemporary approaches to TOL.

Contemporary Approaches to TOL: Strengths and Weaknesses

Studies of inﬂuences on TOL have typically adopted two different approaches.
One is to consider policy as a source of inﬂuence on TOL. This approach focuses on
district or state policy effects on TOL, assuming that coherent policy strategies can
stimulate and support high quality TOL. Another is to approach TOL through
professional community. This approach underscores inﬂuences of local teacher

community, assuming that particular norms and practices of the teacher community

17

construct learning environments for TOL. Each approach to TOL has particular strengths
and limitations in its conception and methods. The next section will highlight advantages
and disadvantages of contemporary approaches to TOL, through the review of two

empirical studies.

“Policy” as Source of Inﬂuence on TOL

A representative study supporting policy effects on TOL is The Eisenhower
Evaluation Study (EES) conducted by the American Institutes for Research (AIR). The
purpose of the EES was to examine the effectiveness of district policies on TOL in
districts that received federal funding from the Eisenhower Professional Development
Program. The Eisenhower study was conducted in two parts. The ﬁrst part was to identify
features of high quality TOL and to explore how the features of high quality TOL relate
to one another and to teacher outcomes (teachers’ knowledge and skills, and teaching
practices). The second part was to examine how districts build a vision for the high -
quality TOL, and plan and evaluate Eisenhower-assisted activities supporting high
quality TOL. The current study here will mainly discuss the approach to TOL with the
second part of the Eisenhower study.

The Eisenhower study shows that district policy mechanisms are related to high
quality TOL (Desimone et al., 2002). The researchers surveyed a nationally
representative sample of 400 professional development coordinators in districts funded
from the federal government. Using OLS regression, they found that the quality of TOL
was more likely to be improved through particular district policy strategies, such as

alignment of TOL with standards, multiple funding sources for TOL, continuous

18

improvement efforts, and teacher involvement in planning for TOL.

Using a path model, the researchers indicate a ﬂow of district policy inﬂuences
from coherent and effective district vision and implementation to high quality TOL.
Speciﬁcally, districts which build coherent visions of TOL through the strategies of
alignment and co-funding, tend to monitor and manage these visions of TOL using the
continuous improvement efforts such as strategies of accountability and teachers’
involvement in planning for TOL. Then, these policy strategies result in high quality
TOL (such as participation in reform types of TOL, active leaming, and the longer
duration of activities) (Desimone et al., 2002). In conclusion, the researchers note,
“Alignment and coordination activities require districts to be thoughtful about planning
and continuous improvement strategies, such as use of data, needs assessment, and
evaluation, and that all of these activities can affect the quality of the professional
development” (Desimone et al., 2002; p. 1287).

The Eisenhower study focusing on the policy effects as an approach to TOL
provides a useful insight into how district policy mechanisms affect high quality TOL.
Yet, it has little to tell us about the effects and reactions ﬁ'om inside schools on TOL.
School conditions and factors, such as local professional communities can construct
directly teachers’ work, and furthermore interpret and mediate the effects of the broader
contexts (Darling-Hammond et al., 2003; Little, 1999a, 2001; McLaughlin & Talbert,
2001; Rosenholtz, 1989). For example, Darling-Hammond and her colleagues (2002)
show that teachers in two different teacher communities within the same district have
distinctively different professional experiences and learning opportunities. In addition,

they respond differently to professional resources and support at the district level. For

19

instance, teachers in a teacher community that values individualism may undermine
opportunities to assess these professional resources at the district level. However, the
policy inﬂuence approach tends to ignore these effects of school contexts. By only
stressing policy effects in the macro-level, this approach to TOL affords little opportunity
to capture the direct and mediated effects of school conditions such as teacher
communities.

Another limitation of the research that focuses on the policy inﬂuence approach
to TOL is founded in its’ methodology. These studies often adopt OLS regression
techniques with large-scale survey data in order to estimate policy effects. Yet, such
techniques are not appropriate for analyzing the inherently hierarchical structures that
encompass a variety of hierarchical relationships, such as educational systems
(Raudenbush & Bryk, 2002). In the multilevel data, variables exist at more than one level
of analysis (e.g., a lower level teacher or school and both lower level and higher level
district or state predictors). In this case, studies using OLS regression (including the
Eisenhower evaluation study) aggregate teacher or school data to the district or state level.
However, this approach discards potentially meaningful lower level variance by
aggregating lower level data. For example, the Eisenhower study on the district policy
effects on TOL did not care about lower level (teacher or school level) variance.
Furthermore, the use of aggregating data tends to increase any relationship found
between variables. In the Eisenhower study, the estimation of relationship between
average teachers’ participation in decision-making of a district and school mean of
teachers’ response to participation in learning opportunities would be greater than the

estimation between teachers level responses to decision-making and to TOL. Such an

20

 

aggregation bias reduces random error. Due to these problems, researchers suggest the
technique of multi-level modeling to analyze separately the effects of different level

variables.

Approaching TOL through Professional Community

In contrast to the approach focused on policy effects on TOL, other studies
approach TOL through professional community. One representative study supporting this
approach is The School Restructuring Study conducted by Little and colleagues. The
purpose of this study was to evaluate the progress of reform in schools funded by
California's state-sponsored School Restructuring Demonstration Program. The study
examined the degree of ﬁt between high school reform and T 0L, and the effectiveness of
TOL on a school’s capacity.

In the school restructuring study, the researchers show that the effectiveness of
reform efforts on instructional practices and TOL varies depending on the school
conditions (Little, 1999b). Using an in-depth case study of three high schools which
participated in the California Restructuring Program, the researchers found that school
conditions, in particular collegial norms and practices of local teacher communities
interpret and mediate the effects of broader reform efforts. Further, these norms and
practices are important sources that encourage teachers to engage in their learning
opportunities. For example, strong teacher communities indicate that teachers view their
colleagues as open to innovation, respectful of one another, and generally both able and
inclined to offer support and good advice. In such communities, teachers are encouraged

to share important resources for instruction, to design together new instructional practices,

21

and to pursue professional learning conceming new reform ideas (Little, 1999b). By
contrast, weak teacher communities value comparatively more conservatism, and
individualism, and have low expectations about student learning. In these communities,
teachers’ work and learning are isolated and discouraged. The contents of learning
opportunities are often irrelevant to reform goals.

These different features of teacher communities are presented not only in school
units, but also across subject departments. The school restructuring study shows different
TOL patterns within and across six departments in the three schools. In this study, not all
strong teacher commmrities derive their strength from a commitment of learning. A
teacher community of a subject department that preserves traditional beliefs and values
related to instruction and learning, inhibit teachers to challenge teaching routines and to
approach to learning opportunities about new reform ideas. In the conclusion, Little
(1999b) argues that TOL for the successful implementation of ambitious instructional
reforms would be enforced within a professional community that sustains professional
norms and collegial relationships.

Although the professional community approach to TOL has made signiﬁcant
contributions to our knowledge of dynamic inﬂuences of professional communities, this
approach has little to tell us about the interactive effects between professional
communities and broader educational environments. In other words, professional
communities alone may have variable effects on teachers’ beliefs and activity, but the
professional communities also affect TOL by incorporating with certain broader
contextual conditions such as district policies or institutional environments (Knapp, 2003;

McLaughlin & Talbert, 2001). For example, Knapp (2003) asserts that professional

22

communities that encourage teachers’ learning can be established or strengthened through
districts’ consistent messages about professionalism and expectations on instruction.

However, the approach to focusing on professional community as an inﬂuence on TOL
does not capture these interactive relationships. By restricting its scope to the school level,
the professional community approach to TOL affords little opportunity to account for the
interactive relationships between proximal school conditions and distal contexts.

Further, most intensive case studies of professional community inﬂuences on
TOL, such as the school restructuring study have been conducted with small samples of
teachers and schools. Even though these intensive case studies are able to provide rich
and detailed information about the relationships between TOL and professional
communities, they provide little diversity across schools (McLaughlin & Talbert, 2001).
For example, the result of the school restructuring study that professional communities
are signiﬁcant variables in TOL is on the basis of only three schools. Consequently, it is
difﬁcult for these small sample studies to generalize the results of the study to broader
groups of teachers.

In sum, it seems clear from the review of these illustrative studies that
contemporary approaches to TOL are insufﬁcient in providing a complete understanding
of how multiple and embedded contexts affect TOL. These approaches fail conceptually
and methodologically to capture the features of embedded contexts due to their single
perspective on contexts and to certain methodological limitations. These limitations call
for a new conceptual approach taking a deep and broad look at the multifaceted

inﬂuences on teachers.

23

Another Conceptual Approach: The Multi-perspective Embedded Context Model
To address the limitations of the contemporary approaches to TOL, this study

uses a conceptual framework that considers the multiple and complex inﬂuences on TOL,
based on McLaughlin and Talbert’s multi-perspective embedded context model. This
model begins with the assumption that school settings where teachers work are inherently
multi-embedded: teachers are embedded within the most proximal zone, the school
setting; schools are embedded within a broader administrative zone inﬂuenced by district
and state factors; and formal educational systems are embedded within institutional
environments (McLaughlin & Talbert, 2001). Each of these contexts inﬂuences teachers’
work. Furthermore, these diverse contexts interact to shape teachers’ professional roles
and dispositions. In this section, this study discusses several main tenets and advantages

of this model for the current study.

Theory and Method must be ‘Integrated’

The multi-perspective embedded context model underscores the ‘integration’ of
theoretical perspectives as well as the ‘integration’ of research methods to examine
school contexts for teachers’ work. McLaughlin and Talbert (2001) argue that theory and
method must be integrated because a single theoretical framework and method tend to
miss the embedded character of school contexts. Integration among the theoretical
perspectives enables us to see how situated realities of school settings are constructed
from diverse contextual conditions and processes. For example, social systems theories
can provide insight into how professional norms and communities of practice shape TOL.

Organization systems theories may enable us to see how school organization and

24

governance affect TOL. An institutional perspective may explain the ways in which
social class cultures frame TOL. Therefore, integration of theoretical perspectives can
provide comprehensive insights into how multiple contexts shape TOL

The multi-perspective embedded context model also calls for an integration of
quantitative and qualitative research methods in order to capture teachers’ experiences
within the multiple embedded contexts. That is, using quantitative measurement
techniques enables the detection of overall tendencies of the effects on teachers’
schooling experiences at each level of multiple contexts while using qualitative
techniques helps account for interdependency or interaction among complex contexts
(McLaughlin & Talbert, 2001). Through such a strategic integration of the two research

methods, the model offers a contextually rich understanding of teachers’ work.

Multiple Theoretical Perspectives

The multi-perspective embedded contexts model suggests taking three multiple
theoretical perspectives on school contexts for teachers’ work: social system,
administrative and organizational, and institutional perspectives. These multiple
theoretical perspectives help provide a complete understanding of school contexts for

teachers’ work (McLaughlin & Talbert, 2001).

Social system perspective
This perspective focuses on school-level interactions among teachers and
administrators, as critical contexts of teachers’ work. It assumes that professional norms

and collegial relations within social systems account for differences in teaching and

25

learning between schools (Talbert & McLaughlin, 1999). Teacher communities are
considered primary social systems of teachers’ work. Particular norms and values of
these communities frame teachers’ practices. For instance, in the teacher community that
encourages professional experimentation and supports collaboration among colleagues,

teachers are more likely to engage in TOL (Little, 1982).

Administrative and organizational perspective

This perspective underscores administrative contexts (such as subject area
departments, school organizational conditions, and state or district policies) as critical
inﬂuences on teachers’ work. It assumes that educational policies, resource levels and
allocation patterns, and programs within administrative contexts affect teachers’ work. In
McLaughlin and Talbert’s model, administrative contexts are divided into intra- and
extra-school aspects. Intra-school administrative contexts encompass school
organizational resources, governance, and leadership that affect teachers’ work. For
example, involving teachers in making decisions that affect their learning opportunities
and instruction can increase their responsibilities for professional enhancement, and then
increase teachers’ engagement in TOL (Smylie, 198 8). Extra-school administrative
contexts encompass state or district policies and organizational contexts at the district
level. State and district policies nurture or discourage teachers’ work. For example,
standards-based reform strategies that aim to increase coherence in frameworks for
curriculum, accountability, and professional development may stimulate and support
improvement in teachers’ work. Apart from educational policies, differences in the

organizational contexts at the district level may matter for teaching and learning. Spillane

26

(2004) captures this point nicely in his study of how district ofﬁcials mediate state policy
through “sense-making” or interpretation of policy. Hence, not only do resource decisions
at the district level affect policy implementation but also the meanings that district

officials glean from external policies in processes of interpretation and understanding.

Institutional perspective

Multiple perspective embedded contexts model also suggest that teachers’ work
can be shaped by institutional environments outside of school systems. In other words,
institutional contexts (such as norms of higher education institutions, subject areas’
cultures, local labor markets’ conditions, and community and parents’ culture or
demographics) of K-12 schooling frame the works of teachers in ways quite independent
of educational policies and professional communities’ cultures (Talbert & McLaughlin,
1999)

For example, subject disciplines’ cultures within the same school constitute
signiﬁcantly different teachers’ beliefs about instruction. For instance, social studies
teachers tend to view their ﬁelds as changing continually, and generating new knowledge,
while mathematics teachers tend to see their subject as relatively constant, assuming the
same content and pedagogy regardless of student or organizational contexts (Grossman &
Stodolsky, 1994). Parents’ social economic conditions and local communities’
circumstances are also critical institutional contexts that shaped the quality of teaches’
work life (McLaughlin & Talbert, 1999). Teachers in suburban schools, serving students
from middle-class and well-educated families may tend to have high expectations to

student performance while teachers in urban schools, serving students from widely

27

diverse social class, race, and language contexts may tend to have low expectations.
Consequently, the multi-perspective embedded context model suggests that
through taking these three multiple theoretical perspectives (social system perspective,
administrative and organizational perspective, and institutional perspective), we can
understand more completely about diverse inﬂuences of each school context on teachers’

work.

Nested Contexts vs. Embedded Contexts

Another tenet of this model is the emphasis on ‘embeddedness’ of contexts.
McLaughlin and Talbert (2001) distinguish embedded from nested context. Nested
context, which is commonly used in social studies, presume linear and hierarchical
relationships among contexts, with inﬂuences transmitted from higher-level to lower-
level settings. The effects of nested contexts add up to affect teachers’ work. For example,
states’ systematic reform strategies such as professional accountability would result in
efforts by districts to give high quality learning opportunities for teachers, and
accordingly lead to high rates of teachers’ participation in learning opportunities and
improved teaching practices.

By contrast, the view of embedded contexts does not accept the assumption of
simply additive effects among contexts, because it assumes interactive and transactive
relationships among multiple contexts. Rather than the effects of contexts necessarily
transmitted in a linear way from higher level to local practices, the inﬂuence arrow might
point not only from policy to practice, but also from practice to policy; or from

organizational features to policy.

28

For a more complete understanding of the notion of embedded contexts, it is
necessary to note three assumptions. First, broader contexts of schooling, such as state or
district administrative contexts or institutional cultures are experienced directly in local
practices. That is, higher-level educational policies directly permeate micro-school
settings and teachers’ work. Cultures or norms of local communities also frame directly
teachers’ daily school life in ways quite independent of ofﬁcial school policies and social
relations in schools. For instance, a state policy on mathematics instruction can directly
affect teachers’ use of particular texts and assessments, regardless of school actions.
Incoherent instruction of policies at the district level can discourage teachers’ practices,
even in a supportive school environment with strong principal leadership. In this sense,
McLaughlin and Talbert (2001) note, “Elements of educational culture from multiple
school contexts are incorporated, or enacted, by individuals in school settings” (p. 145).

Second, local teacher communities and school settings in the embedded contexts
yield substantially different teachers’ work and cultures within the same school
organization, district administration, and state policy system. That is, teachers’ work can
be differently constructed by social relationships among colleagues, disregarding broader
contexts conditions. For example, despite rich district conditions and supportive
leadership, the positive inﬂuences of these macro contexts conditions are not transmitted
automatically into positive teachers’ professional experiences in a professional
community that values individualism and privacy (McLaughlin & Talbert, 2001). The
privacy norms and practices of such teacher communities may have different inﬂuences
on teachers’ work, mediating positive effects of the district and school leadership on

teachers’ experiences.

29

Finally, the embedded contexts assume that cross-domain linkages among
contexts conditions can affect teachers’ work. That is, different combinations of context
conditions can have qualitatively different impacts on teachers’ work (McLaughlin &
Talbert, 2001). For instance, states’ mandated demands for higher standards for
instruction may have the desired effect in school settings where the school culture
emphasizes the value that all students have the ability to leanr and where there is ongoing
teacher training about the best ways to implement higher standards. Yet a much less
desirable set of outcomes may occur in a school setting where teachers do not believe that
their students have the capacity to meet higher standards or where teachers are unsure of
how to change their teaching to achieve the new standards (Talbert & McLaughlin, 1999).

Consequently, these three assumptions of embedded contexts- permeability of
macro contexts; diverse inﬂuences of local teacher communities; and interactive effects
among contexts - make a distinction of embedded contexts from the nested contexts,
which is a more common view of school contexts. That is, while the notion of nested
contexts which assumes the inﬂuences of higher-level contexts conditions directly add up
to teaching practices relying on a “chain of command”(Talbert & McLaughlin, 1999), the
notion of embedded contexts assume substantial inﬂuences of each contexts conditions at
the macro- and micro-school contexts levels, and the interactive effects among diverse
contexts conditions. In such manners, the multi-perspective embedded context model
enables us to understand the inﬂuences of multiple embedded contexts that are related
one to another in complex ways and interact to shape teachers’ work.

McLaughlin and Talbert’s multi-perspective embedded context model has

several advantages for research on the relationship between school contexts and teachers’

30

work. First, it enables us to model inﬂuences on teachers’ work in a comprehensive
fashion, by simultaneously accounting for the effects of multiple contextual conditions.
That is, integration of three different theoretical perspectives (institutional, administrative
and organizational, social system) enables us to inclusively consider a wide variety of
contextual factors that relate to teachers’ work.

Second, the multi-perspective embedded context model makes a contribution to a
better understanding of contextual inﬂuences on teachers’ work via the construct of
embedded contexts in contradistinction to the simple nested model. Three notions of
embedded contexts - permeability of macro contexts; inﬂuences of local teacher
communities; and interactive effects among contexts - make it possible to conceptualize
highly interactive and quite complex contextual inﬂuences on TOL.

Therefore, it is a useﬁrl framework for studying the inﬂuences of the complex and
highly interactive school contexts on teachers’ work. Drawn from the main tenets
(multiple theoretical perspectives and embedded contexts) of the multi-perspective
embedded context model, the current study develops the hypotheses regarding the

relationship between TOL and contexts in the next chapter.

31

CHAPTER III
LITERATURE REVIEW (PART II)

This chapter continues the exposition of the literature on TOL, working out
speciﬁc hypotheses regarding the relationship between district and school contexts and
TOL. In this ﬁrst section, this study presents the deﬁnitions of TOL and indicators of
effective TOL used for this study. The second section provides ﬁve hypotheses and
rationale to test them, as grounded in the literature, and as a basis for use of the multi-

perspective embedded context model.

TOL Definitions and Indicators

While numerous studies have emphasized the importance of TOL for improving
the quality of teaching and student achievement, to say that teachers should have learning
opportunities in order to implement instructional reform successfully leaves much
unspeciﬁed and under-explored. Do instructional improvements relate to the formalized
TOL generally provided from districts or to the spontaneous learning opportunities
occurring in teachers’ daily work? What evidence demonstrates particular forms of
effective TOL for the improvement of teaching and learning? As expectations about TOL

increase, these issues and conceptions require clariﬁcation.

Deﬁning TOL

TOL is often equated with the term ‘professional development’, which is deﬁned
in turn as the formal learning opportunities that are pre-planned and designed in some
way to improve the skill, knowledge, and dispositions of teachers (Bellanca, 1995;

Fenstermacher & Berliner, 1985). These learning opportunities unfold through

32

participation in the formalized structures within the educational system (e.g., federal,
state, and district agencies, universities, professional networks, and others). Formal
learning opportunities take place outside of classroom practice and are often mandated. A
vast array of speciﬁc activities, such as district-or school-sponsored workshops,
university courses, summer institutes and seminars encompass these learning
opportunities.

Some scholars have asserted that learning opportunities for teachers need to be
broadened to include not only formal but also informal leaming activities, pointing out
that formal learning opportunities are narrow and detached from real-time learning during
the normal workday (Fullan, 1995; Knapp, 2003). Fullan (1995) says, "Professional
development is the sum-total of formal and informal learning pursued and experienced by
the teacher in a compelling learning environment under conditions of complexity and
dynamic change" (p.265). In contrast to formal learning opportunities, informal learning
opportunities are less structured and planned. The goals and objectives may or may not
be obvious. These learning opportunities often occur while a teacher is teaching, when a
teacher reﬂects on a particular instructional activity, or when a teacher discusses content
or pedagogical issues with colleagues. For instance, a teacher might have learning
opportunities through an unprepared reﬂective discussion among colleagues about
strategies to improve the performance of struggling readers during a lunch break. A
teacher also might have learning opportunities by reading and processing cues from
students during a lesson and adjusting her practice accordingly. In this sense, informal
learning opportunities tend to be more spontaneous and reﬂective in that they are

typically generated out of real-time experiences (Smylie, 1995). Since these learning

33

opportunities take place within teachers’ practice itself, they tend to draw more easily
implications for their daily work or in their individual classroom.

Yet, deﬁning TOL as conceptions that include only these formal or informal
learning opportunities is still not enough, for explaining the nature of teacher learning
may vary ﬁom teacher to teacher. That is, formal or informal learning opportunities may
be provided, but not all teachers learn from these opportunities (Sykes, 2005). Some
teachers may learn from them, but others may not, depending on an individual teacher’s
psychological aspects or an individual teacher’s social interactions.

TOL can be constructed through an individual teacher's psychological aspects.
The prior knowledge and beliefs that teachers have heavily inﬂuence what teachers learn
and how they interpret and organize their own learning opportunities (Borko & Putnam,
1996; Bransford, Brown, & Cocking, 1999). The logic behind this point is that, because
teachers can best judge their own leaming needs, this enables them to construct their own
learning opportunities based on prior knowledge and beliefs about learning. For example,
a teacher may be intrinsically motivated to engage in learning opportunities when they
feel the deﬁcits about their content knowledge and pedagogical skill (Scribner, 1999).

A cognitive psychological perspective on teacher learning makes a signiﬁcant
contribution to this notion of TOL (e.g., Bransford et al., 1999; Confrey, 1990; Putnam &
Borko, 2000). In this perspective, teachers are active learners who reconstruct their
existing knowledge structures, rather than passively assimilating the learning process. All
knowledge teachers experience is constructed by the teachers. Teachers do not simply
absorb. Due to the personal activity and the inevitable autonomy of intellectual work

done by teachers, the concepts, skills, and beliefs they create are unique (von Glasersfeld,

34

1984). Thus, teachers as active learners may accept, transform, or even disregard diverse
learning opportunities and sources through their own knowledge and beliefs about
teaching, learning, and subj ect-matter. For example, some teachers may feel the needs
and responsibilities to promote professional knowledge and skills, and engage in learning
opportunities, given the provision of learning opportunities. In contrast, others may not
be motivated by their own needs, and do not engage in learning opportunities.

In the cognitive psychological perspective, reﬂection is a key component to
facilitating teachers’ learning opportunities. Teachers transform their knowledge, and
beliefs through reﬂection (Confrey, 1990; Schon, 1983). In reﬂection, teachers determine
whether previous ways of thinking are workable given new information learned through
experience and interaction (Cobb & Bauersfeld, 1995). When dissonance arises between
what was previously thought and new experiences, personal reﬂection is critical in
spurring the creation of new ways of thinking, believing, and acting. Reﬂection may also
account for dissonance in ways that do not require such new developments. In this way,
reﬂection assists teachers in making sense of their knowledge and practices, seeing the
difference between new information and current beliefs, questioning those experiences,
and then feeling the need to engage in learning opportunities. Without reﬂection to
facilitate learning opportunities, teachers would be forced to either build up an increasing
collection of unrelated experiences or reject all experiences that do not perfectly align
with their prior knowledge and beliefs. In this sense, TOL is most likely to occur in
favorable learning environments, in which teachers have been encouraged to reﬂect on
teaching experiences.

TOL also can be constructed through social interactions with colleagues. Social

35

interactions, such as discussion concerning problems of practice where teachers work
strongly affect teachers’ engagement in TOL (King & Newmann, 2000; Lieberman, 1996;
Scribner, 1999; Smylie, 1995). The logic here is that TOL springs not only from the
individual teachers’ thought process, but also from social interaction with colleagues
because teachers can construct ideas related to instruction and student work while
working together with other colleagues.

This concept of TOL is signiﬁcantly inﬂuenced by social learning theories,
which focus on the social relationship that comprises learning (e.g., Bandura, 1986;
Vygotsky & Cole, 1978). In this view, teachers learn from and with one another.
Knowledge is not considered as an individual property but is in nature communal and
constructed through social interaction (Lave & Wenger, 1991). The view of knowledge as
distributed, situated, and socially constructed highlights the importance of others as well
as the cultural and social environment in the process of learning (V ygotsky & Cole,
1978). Colleagues, especially experienced ones, play important roles in the learning and
development of an individual. For example, a teacher's current level of understanding can
be elevated to a more complex level when that teacher socially engages with colleagues
who possess more advanced levels of understanding (such as veteran teachers). Therefore,
teachers are more likely to learn and change their practice if they engage in rich
deliberations and conversations with colleagues about their practice (Spillane & Zeuli,
1999)

In this perspective, teacher communities are a key factor in supporting teachers’
leaming and practices. Teacher communities serve as psychological tools that mediate

teachers' thoughts and practices (McLaughlin & Talbert, 2001). These communities

36

motivate teachers to engage in learning in order to develop and sustain learners' identities
in the communities in which they participate. By participating in teacher communities,
teachers have opportunities to participate in practices of inquiry and learning, and to
develop disciplinary practices of argument and discourse. Furthermore, the communities
grow and change over time through the participation of their members. In this sense, the
community is both a place for members to acquire knowledge of the community, and a
place where participants collectively generate new knowledge and practices within the
community. Teachers, therefore, tend to learn by engaging in a "culture" or "teacher
community" that supports favorable social interaction. For instance, teachers can learn
about the ways to change their teaching practices by trying new ideas surfaced in peer
observations, helpful feedback, or grade-level conferences around particular problems of
instruction.

Therefore, TOL can be deﬁned using the four concepts: formal, informal,
psychologically constructed, and socially constructed TOL. Figure 3-1 presents the
concepts of TOL deﬁned for this study. TOL is divided into two different types of
learning opportunities - formal and informal. These learning opportunities can also be
constructed psychologically or socially.

Speciﬁcally, G) represents formal learning opportunities shaped through individual

teacher's prior knowledge and beliefs about teaching, learning, or subj ect-matter. For
example, formal learning opportunities such as seminars or university courses may be
provided, and some teachers who believe that these leaming opportunities are useful for

improving teaching practices may take on and learn from these formal opportunities.

37

However, other teachers who believe these learning opportunities are not useful for
teaching practices may not participate or learn from them.

Figure 3.1 TOL Deﬁnition used for This Study

 

® ®
Informal & Formal &
Psychologically Constructed Psychologically Constructed
TOL TOL

ED (3

Informal & Formal &

Socially Constructed Socially Constructed

TOL TOL

 

C2) represents informal learning opportunities shaped through individual

teacher's prior knowledge and beliefs about teaching, learning, or subject-matter. For
example, informal learning opportunities, such as reading an educational journal or
conducting and independent study may be given, and some teachers who believe these
informal learning opportunities are useful for promoting teaching practices may learn
from them. However, other teachers who believe that these informal learning

opportunities are not useﬁrl for teaching practices may not learn from them. Therefore,
both formal & psychologically constructed TOL (Q) and informal & psychologically
constructed TOL (C2)) are more likely to occur when teachers are encouraged to reﬂect
about ways of their knowledge and practices.

6) represent formal learning opportunities constructed through social

interactions with colleagues. For example, formal learning opportunities such as school-

38

level subject workshops may be provided, and some teachers in teacher communities
which encourage rich deliberations and conversations with colleagues may be more likely
to take on these formal learning opportunities and learn from them. However, teacher
communities which value individualism and do not encourage interactions with

colleagues are less likely to promote these learning opportunities.
@ represent informal learning opportunities shaped through social interactions

with colleagues. For example, informal learning opportunities, such as for participation in
an informal study group may be offered, and some teachers may experience these
informal learning opportunities and leam from them. However, other teachers in teacher
communities which value individualism and do not encourage interaction with colleagues
may be less likely to experience these learning opportunities. Hence, both formal &

socially constructed TOL (C3)) and informal & socially constructed TOL (ED) are more

likely to occur in a school culture or teacher community that supports favorable social
interaction.

In sum, provision of learning opportunities and learning are often treated as one
and the same, but this is obviously not the same. When we deﬁne TOL as conceptions
that include not only these provisions of formal or informal learning opportunities but
also psychologically or socially constructed learning opportunities, a more in-depth
discussion of school contexts that affect these provisions of learning opportunities as well

as constructed learning opportunities opens up.

Literature on Effective TOL Indicators

During the past decade, a considerable body of literature has identiﬁed the

39

features of effective TOL for improving teaching practices and student learning (e. g.,
Darling-Hammond & McLaughlin, 1995; Elmore, 2002; Hawley & Valli, 1999; Knapp,
2003; Locks-Horsley, Hewson, Love, & Stiles, 1998; Sykes, 2002). For example, Knapp
(2003) describes the main features of effective TOL based on a range of theoretical and
empirical studies.

To be powerﬁrl, professional learning experiences should: 1) Concentrate on

classroom teaching that emphasizes high learning standards and on evidence of

students' learning to standard. 2) Focus on building teachers' pedagogical content
knowledge. 3) Model "preferred" instructional practices (e.g., active learning),
both in classrooms and in adult learning situations. 4) Locate professional
learning in collaborative, collegial - and generally school-based- learning
environments. 5) Offer rigorous and cumulative opportunities for professional

learning over time. 6) Align with reform initiatives (p.172).

Still, much of the literature on the features of effective TOL is not grounded in
empirical evidence about its effects on teaching practices or on student learning.
Furthermore, some studies about effect of TOL are compromised, by methodological
limitations that rely on a variety of approximations for indicators of student learning or
overestimate actual implementation due to teachers’ self-reports of their teaching

practices (Knapp, 2003). Nevertheless, there are several useful empirical studies that

explore indicators of effective TOL and/or how effective TOL is measured.

Content- ocus

One line of empirical research on effective TOL focuses on the content. What
teachers actually learn matters for changes in teaching practices and student achievement.
These studies indicate that TOL is effective on teaching practices and student learning

when it is directly linked to speciﬁc content knowledge and how students learn that

40

content (e.g., Cohen & Hill, 2001; Garet, Porter, Desimone, Birman, & Yoon, 2001;
lngvarson, Meiers, Beavis, 2005; Kennedy, 1998).

For example, in a study of mathematics instructional reform and TOL in
California, Cohen and Hill (2001) show that TOL focused on speciﬁc mathematics
curriculum content impacts classroom practices and student achievement. In this study,
the researchers examine the effects of two categories of the TOL curriculum provided
from the formal workshops. One is speciﬁc mathematics curriculum teachers would be
teaching to students, which focused on student mathematics thinking, instructional
standards, and mathematics subject matter. Another is general curriculum, which is not
focused directly on mathematics curriculum, such as gender equality in the math
classroom. Using cross-sectional teacher-level data with OLS regression, the researchers
found that teachers who participated in formal learning opportunities focused on speciﬁc
mathematics cuniculum were more likely to adopt new, reform-oriented teaching
practices.

Further, TOL focused on speciﬁc mathematics cuniculum is also positively
associated with student achievement. In the second part of the research, Cohen and Hill
(2001) explore whether students of teachers who received learning opportunities focused
on subject matter and state assessments achieve at higher levels on the state mathematics
achievement test. Using student achievement on the California Learning Assessment
System, the researchers found that schools were more likely to have high student
achievement when teachers took part in formal learning opportunities that focused on
subject matter and state assessments. By contrast, when teachers spent most of their time

on learning opportunities studying general curriculum, their students did not perform as

41

well. These ﬁndings suggest that TOL is effective on teaching practices and student
achievement when it focuses on the speciﬁc subject curriculum focal to state assessments,
student mathematics thinking, and subject matter they would teach.

Another study examining student achievement on the California Learning
Assessment System also supports the effects of the content-focused TOL. Wiley and
Yoon (1995) explored the relationship between ‘reform-oriented curriculum TOL’ and
teachers’ classroom practices and student performance in mathematics. In their research,
two outcome variables were used to measure the ‘reforrn-oriented cuniculum TOL’. One
variable was the degree of teachers’ familiarity with reform-oriented documentation, and
another was teachers’ participation in in-service training involving curriculum activities
(e. g., grade-level workshops led by mentor teachers, or mathematics specialists). The
study found that students perform at higher levels across all grades when their teachers
participated in opportunities to leam about speciﬁc and standards-based curriculum and
instruction.

Some case studies also show that TOL focused on speciﬁc content has an effect
on teaching and leanring. McCutchen and colleagues (2002) examined the relationship
between TOL focused on reading-content and teaching practices, and student learning. In
this study, the researchers provided formal opportunities for one teachers group to learn
speciﬁc reading content knowledge (such as knowledge of word sounds and structure),
whereas the other group had no additional learning opportunities regarding speciﬁc
subject content. At the end of the school year, teachers who had learning opportunities
focused on speciﬁc reading content knowledge spent more time explicitly using reform-

oriented instructional strategies such as building blocks of words and language and

42

encouraged students to seek different methods of ﬁnding answers. Moreover, these
teachers’ students did better on tests of word reading, spelling, and comprehension. By
contrast, teachers who did not have additional learning opportunities on speciﬁc reading
content knowledge did not try out new practices in the classroom.

In sum, several large-scale studies and small case research indicate that to be
effective on teaching practices and student learning, the contents of TOL matter. These
contents of TOL should focus on specific subject knowledge, subject curriculum
standards, and understanding about students’ learning such as student assessments that

require higher-level thinking skills.

Active Learning: Pedagogy of T 0L

The empirical studies about the effective TOL focus not only on the content but
also on the pedagogy of the TOL. That is, how teachers have been taught also matters for
changes in teaching practices and student achievement. These studies indicate that
improved teaching practices and student achievement result when teachers engage in
intellectual and active learning opportunities that allow teachers to understand deeply any
new instructional approaches (Garet et al., 2001; Ingvarson et al., 2005; Lieberman,
1996).

In the evaluation study of the Eisenhower program for professional
development (l996~1998), for example, Garet and colleagues (2001) show that active
learning opportunities provided formally or informally have signiﬁcant effect on the
changes in teaching practices. In this study, the researchers explore the relationship

between teachers’ knowledge, skills, and teaching practices and six features of high

43

quality TOL. These six indicators encompass (1) the degree to which the activity is
focused on subject content knowledge; (2) the degree to which the activity is active; (3)
the degree to which the activity is coherent with other activities, standards, and
assessments; (4) the form of the activity (reform type vs. traditional type); (5) the degree
to which the activity emphasizes collective participation of teachers ﬁom the same school,
department, or grade level; and (6) the duration of the activity. In particular, active
learning includes four dimensions in their research: (1) observing teaching; (2) planning
for classroom implementation; (3) reviewing student works; and (4) presenting, leading,
and writing. On the basis of a nationally representative sample of teachers, Garet and
colleagues (2001) found that content-focus, coherence of learning, and active learning
had a direct effect on teacher outcomes variables (teachers’ knowledge, skills, and
teaching practices). The other features, which were often regarded as structural features
of TOL (such as duration, collective participation, and reform activity) had indirect
effects on teacher outcomes variables. These ﬁndings suggest that TOL is more effective
when it focuses not only on content speciﬁc but also on active learning in terms of the
pedagogy of the TOL.

Ingvarson and colleagues (2005) also conﬁrm the importance of active learning
in the effectiveness of T OL. In the study of the Australian Government Quality Teacher
Program, they (2005) examined whether the features of effective TOL are related to
teachers’ knowledge, practice, and efﬁcacy, and student learning outcomes. To measure
effective TOL, the researchers used six indicators of TOL: (1) emphasis on content, (2)
active learning, (3) the amount of time for feedback, (4) collaborative examination of

student work, (5) follow-up and ongoing assistance in the classroom, and (6) total hours

of TOL. In particular, active learning includes two dimensions: reﬂecting on practice, and
testing new teaching practice. Using the teacher-level survey data, the researchers found
that content-focus, active learning, and time for follow-up have signiﬁcant direct effects
on teacher and student outcomes. Feedback and collaborative examination have the least
inﬂuence. Thus, Ingvarson et a1. (2005) also support that active learning is an important
indicator of TOL to improving teaching practices and students learning.

In sum, several empirical studies emphasize teachers’ active learning as an
important indicator of high quality learning experiences for teachers. This active learning
mainly includes talking back, raising questions, observing expert teachers, engaging in
local experiments, study groups and research, and leading discussion. By engaging in
these activities, teachers are encouraged to criticize ideas and resources, and to construct
their own learning on the basis of their experience, and in turn teachers upgrade their

knowledge and skills and change their teaching practices (Garet et al., 2001; Sykes, 2002).

Duration

Some studies on effective TOL argue that how TOL is designed (structure of
TOL) is also an important aspect of high quality TOL, even though this structure is not
directly linked to effects on teaching practices nor to student learning. In particular, the
time teachers spend for TOL is expected to impact teacher and student outcomes when it
is related to the core aspects of effective TOL (i.e., content-focus, or active learning)
(Cohen & Hill, 2001; Garet et al., 2001; Parsad, Lewis, & Farris, 2001).

Garet and colleagues (2001) explore the relationship between teaching practices

and two dimensions of time for TOL: total contact hours of TOL, and the span of time

45

which the activity was spread. The researchers found that time span and total hours had
less impact on teachers’ practices than core features of TOL (such as content-focus or
active learning opportunities), but they had substantial effects indirectly on teacher
outcomes by affecting the substance aspects of TOL. That is, a longer duration of TOL
can provide opportunities for in-depth discussion of content and pedagogical strategies
while span time (learning activities that extend over time) enables teachers to try out new
practices in the classroom and obtain feedback on their teaching.

The important inﬂuences of combination of time spent and content focused or
active learning are found in the Cohen and Hill’s study (2001). They found that teachers
who spent more time on mathematics from a new cuniculum or assessment and on how
to teach the subject, reported practices that were more closely linked to reform objectives.
Even large investments of time in less content-focused workshops were not associated
with more of the practices that reformers proposed. Further, shorter-term learning
opportunities did not allow teachers to learn about either student curricula or their work
on assessments, and these teachers did not report practices close to reformers’ goals.

In summary, these studies about the effectiveness of duration of TOL suggest
that duration is also an important indicator of effective TOL. In particular, TOL is likely
to be more effective if it is sustained over time and involves a signiﬁcant number of hours.
These dimensions of time for TOL inﬂuence indirectly teacher or student outcomes by
combining with the substantive aspects of TOL.

From the several empirical studies, three indicators of effective TOL appear.
The ﬁrst indicator would be content focus TOL. It includes the dimensions of speciﬁc

subject knowledge, standards, and learning about student learning such as student

46

assessments. When teachers have such learning activities, they can understand new
instructional approaches, reﬂect more deeply on their own practices and assumptions, and
make substantive changes in their instructional practices. The content-focused TOL can
be provided formally or informally. In particular, content-focus TOL is more likely to
relate to the formal learning opportunities provided from educational institutions directly
linked to the state and to district standards, assessments, and curriculum guidance.

The second indicator of effective TOL would be active learning. Active learning
means that teachers actively participate in meaningful opportunities to engage with new
instructional ideas and resources. When teachers have such learning activities, they can
understand new instructional approaches, more deeply reﬂect on their own practices and
assumptions, and make substantive changes in their instructional practices. More
importantly, when teachers experience such active learning, they are more likely to teach
in ways that the policy expects to provide for students (Cohen & Barnes, 1993).
Opportunities for active leaming can take a number of forms, including the opportunity to
observe expert teachers and to be observed teaching; to plan how new curriculum
materials and new teaching methods will be used in the classroom; to review student
work in the topic areas being covered; and to lead discussions and engage in written work
(Garet et al., 2001; Lieberman, 1996). Active learning can be provided formally or
informally. Further, these learning opportunities are more likely to be constructed through
social interaction with colleagues such as peer-coaching or mentoring. For instance, Garet
and colleagues (2001) emphasize active learning which happens when teachers meet
informally with other colleagues or leaders in order to discuss or review classroom

implementation.

47

Finally, the durgtion of TOL is also an important indicator of effective TOL.
Although it does not exert a strong direct effect on teacher or student outcomes, duration
can enhance effects of content or pedagogy of TOL. The duration includes two
dimensions such as total number of contact hours spent in the learning opportunities, and
period of time which the activity was spread.

Several large-scale quantitative studies reviewed in this chapter capture formal or
informal learning opportunities, and socially constructed TOL. However, it is difﬁcult to
articulate the aspect of how teachers construct TOL through their prior knowledge and
belief. Qualitative studies with in-depth interviews are required to learn more about the
psychological aspects of TOL. Therefore, the current study does not examine the
psychological aspects of TOL empirically because the SASS data do not provide
information on this aspect. Time span, one aspect of duration used in the previous

research, is also omitted because it is not available in the SASS dataset.

Hypotheses about the Relationship between District and School Contexts and TOL
The purpose of this study is to explore how contexts relate to three TOL

indicators, which were identiﬁed in the previous section. This study proposes that the
multi-perspective embedded model reviewed in Chapter H is a useful conceptual
framework for exploring the relationships between embedded contexts and the three TOL
indicators. Drawn from the main notions of the model (for example, the notions of
multiple theoretical perspectives and embedded contexts), this study develops and tests
ﬁve hypotheses regarding the relationship between contexts and TOL. It illustrates but

does not fully model an embedded context, multi-perspective approach by concentrating

48

on district and school levels. Other contexts such as states or subject departments also
contribute to variations in TOL (McLaughlin & Talbert, 2001 ), but the primary interest of
this study is to demonstrate the utility but not fully represent the model that McLaughlin
and Talbert developed. The current study is restricted to these two levels based on the
data available in the SASS survey.

Figure 3-2 shows how the main notions of the multi-perspective embedded
context model are operationalized into hypotheses for this study.

Figure 3-2 Main Notions of the Multi-Perspective Embedded Context Model and

 

 

 

 

 

 

 

 

 

Five Hypotheses
CD ®
/ Main notions of the Model\ / Hypotheses \
Direct inﬂuences of multiple ’ H1
contexts on teachers’ work @
Multiple theoretical perspectives > H2, H3
on school contexts 6))
Interactive relationships of » H4
K multiple contexts / (9
H5

 

 

 

k J

Panel (D represents the main notions of the multi-perspective embedded context
model. The ﬁrst assumption is direct effects of multiple contexts on teachers’ work. A
second assumption is multiple theoretical perspectives (social system perspectives,
administrative perspectives, and institutional perspectives) on school contexts. A third
assumption is interactive effects of multiple contexts on teachers’ work. Panel ®

represents ﬁve hypotheses operationalized from these assumptions of the multi-

49

perspective embedded context model. Speciﬁcally, based on the ﬁrst notion that each of
the multiple contexts directly inﬂuences teachers’ work, this study develops the ﬁrst
hypothesis that three TOL indicators (content-focus, active learning, and duration) vary
across schools and districts (see the arrow 69). In other words, this study assumes that
district and school level contexts directly contribute to the variation of TOL.

Second, based on the second notion that multiple theoretical perspectives on
school contexts provide a more complete understanding for teachers’ work, this study
develops the second hypothesis that social system contextual factors, administrative
contextual factors, and institutional contextual factors are associated with TOL at the
district-and school-level (see the arrow (ID). In other words, this study expects that the
variations of TOL may be explained by a wide variety of contextual factors, drawn from
the three theoretical perspectives (social system, administrative, and institutional
perspectives) on teachers’ work.

Third, based on the third notion of interactive relationships among multiple
contexts, this study develops the fourth hypothesis that the interactive relationships
between district level administrative contexts and school level social system contexts
inﬂuence the three TOL indicators.

Finally, this study also assumes that the inﬂuence of contexts at the school and
district level on TOL vary across the three TOL indicators. Although this hypothesis is
not directly drawn from the model, several recent empirical studies report that some
contextual factors have different effects on TOL, depending on the aspects of TOL(e.g.,
Desimone et al., 2002; Scribner, 1999). In the following section, this study provides the

rationale for theses ﬁve hypotheses in detail.

50

The Rationale for Hypothesis 1

The multi-perspectives embedded context model (McLaughlin & Talbert, 2001;
Talbert & McLaughlin, 1999) assumes that each of the multiple contexts directly affects
teachers’ work. For example, school settings may yield different professional cultures.
States or districts’ policies also may affect directly teachers’ daily school life. Further, the
cultures of local communities or economic conditions of labor markets may permeate
teachers’ belief about teaching practices.

Based on this idea of the model, the current study assumes that school-and
district-level contexts signiﬁcantly inﬂuence the variability of TOL. A large body of
scholars who have looked inside schools (e. g., Bryk & Driscoll 1988; Little 1982, 1999a,
1999b ; Metz 1990; Newmann, 1996; Rosenholtz 1989) found that TOL varies from
school to school, affected by characteristics of teacher communities, strength of school
mission, or school organizational conditions. For example, different levels of collegiality
norms, respect, and trust among teachers generate signiﬁcantly different school cultures
that either bolstered or eroded teachers’ sense of professional learning (Little 1982;
McLaughlin 1994). Also, district differences in policy strategies or organizational
contexts contribute to the variations in TOL. Recent studies, including those by Spillane
(2004) and Firestone et al. (2004) conﬁrm the importance of district inﬂuences, while a
great body of prior research has shown the importance of school level effects. For
example, Firestone and colleagues (2004) through an analysis of three urban school
districts in New Jersey, show that district leaderships in relatively similar demographic
and policy contexts take very different approaches to offering TOL.

Building on this rationale, the ﬁrst hypothesis about the relationship between

51

contexts and TOL states that:
H 1: Three TOL indicators (content-focus, active learning, duration) vary

across schools and districts.

The Rationale for Hypothesis 2

Another notion of the multi-perspective embedded context model, multiple
theoretical perspectives on school contexts provide a more complete understanding for
teachers ' work, is important for the current study because it enables us to consider a wide
variety of contextual factors related to TOL. The model suggests three theoretical
perspectives to assessing school contexts; 1) the social system perspective assuming that
professional norms and collegial relations within social system contexts, such as teacher
communities make differences in teaching and learning between schools, 2) the
administrative and organizational perspective, assuming that educational policies,
resource levels and allocation patterns, and programs within administrative contexts
affect teachers’ work, 3) the institutional perspective, assuming that cultures or
conditions of institutional environments (such as local community circumstances or
cultures, and parents’ socio-economic conditions) permeate teachers’ experiences.

Building on these ideas of the model, this study presumes that three categories of

contextual factors must be associated with TOL at the school level.

Social system contextual factor

Professional learning community From the social system perspective,

professional community is often considered a critical context of teachers’ work. A

52

substantial body of research argues that professional communities characterized by
shared beliefs about the school mission and collaboration with peers are essential for
promoting TOL (Bryk, Cambum, & Louis, 1999; Little 1982, 1990b; Newmann, 1996).

Yet, some recent studies assert that effects of professional community on TOL
require considerably more than mere cordiality among staff (Little, 1999b; McLaughlin
& Talbert, 2001). For example, McLaughlin and Talbert (2001) distinguish traditional
communities from learning communities, assuming that not all professional communities
encourage TOL. In the empirical research combining case studies and large-scale surveys,
they found that learning communities that value innovation and learning and encourage
staff to design new practices are more likely to support teachers’ on-going inquiries and
to organize professional learning collectively. Hence, the current study assumes that

“professional learning community’ ’ is a critical contextual factor for TOL.

School organizational factors

From the administrative and organizational perspective, several school
organizational factors are regarded as critical contexts related to TOL. Drawing on the
existing theoretical and empirical literature, this study identiﬁes ﬁve organizational
factors related to TOL.

Supportive principal leadership: It is frequently observed that principal leadership
plays a key role in nurturing T OL by creating learning environments. However, speciﬁc
aspects of principal leadership that promote TOL are elusive to pin down. Some analysts
argue that facilitative principal leadership who leads to a better school climate is

associated with promoting TOL (e.g., Clement & Vandenberghe, 2000; Zheng, 1996).

53

Other studies have found that instructional supervision of principals, guiding
achievement of school goals, and close monitoring of instruction, are more likely to
motivate TOL (e.g., Scribner, 1999; Youngs & King, 2002). Further, other scholars
underscore the attitude of principals who place a high value on teacher learning and seek
to connect TOL to school improvement (e. g., Sykes, 2002). This study assumes that
supportive leadership style may be signiﬁcantly associated with TOL.

Teacher autonomy Two forms of teacher autonomy are in general highlighted for
teachers’ learning and practices. One is faculty inﬂuence, which means the inﬂuence
collectively wielded by faculties over school policy-making. Another is individual
autonomy, which means teachers' freedom to determine planning and teaching in their
classrooms (Ingersoll & Alsalam, 1997). Proponents of increases in faculty inﬂuence and
individual autonomy assert that teachers will be highly committed to their learning and
instruction if they participate in making decisions about instruction and school policy
(Ingersoll & Alsalam, 1997; Poulson & Avramidis, 2003; Rosenholtz, 1989). However,
some scholars suspect that individual autonomy in the self-contained classroom serves to
sustain teachers' isolation from peers and to limit their sense of community, which brings
about negative effects on TOL (Little, 1990a). The current study assumes that the
directions of effects of these two forms of teacher autonomy (faculty inﬂuence and
individual autonomy) on TOL may be different respectively.

Teachers’ workload It is often presumed that teachers’ workload negatively
inﬂuence the possibility of creating TOL. Over-working places limitations on available
teachers’ time and energy to participate in learning opportunities. Little (1990a) notes

two aspects of teachers’ workload related to TOL. One is teacher-student ratio. The

54

greater the number of student teachers must teach, the greater the amount of time
required to evaluate student assignments, or manage student paperwork, and accordingly
the time for teachers’ learning opportunities decrease. Another aspect of teachers’
workload is planning and preparation burden created by course-loads. The greater the
amount of time required to prepare for daily work, the less time available to reﬂect on
how that work is going or to pursue new ideas, materials, or methods. Given these
arguments, the current study expects the results of the negative relationships between
teachers’ workload and TOL, in terms of two aspects.

School size School size has been shown to have a signiﬁcant effect on the
teachers’ work and student outcomes, yet the direction of the effect is inconsistent. Two
conﬂicting perspectives can be found in the literature on the effects of school size. One
perspective underscores the positive effects of increased size, drawing on the assumption
of economies of scale. Another perspective highlights negative effects, centering on the
more impersonal social interactions as impeding success (Lee, Smith, & Bryk, 1993).
Recently, research evidence is accumulating to support the argument that teachers’
learning decreases in bigger schools. Greater school size is shown to be associated with
reduced interpersonal relationship and teacher collaboration on interdisciplinary units,
and increased teachers’ resistance to innovation (Klonsky, 2002). Based on these research
ﬁndings, the current study assumes that there is a negative relationship between bigger
school size and TOL.

School level Elementary and secondary schools generally exert different
inﬂuences on teachers’ work as a result of the differences in structural and institutional

factors (McLaughlin & Talbert, 2001; Firestone, Herriott, & Wilson, 1984; Miller, 2003).

55

In American education, it is often assumed that teachers in secondary schools that
generally have larger enrollments and a more complex and differentiated structure (i.e.,

departmentalized) than do elementary schools are less likely to participate in TOL.

Institutional contextual factors’

From the institutional perspective, TOL also can be shaped through the broad and
diverse institutional contexts outside of educational systems. Although little research has
been conducted on the effects of institutional contexts on teacher learning, two contextual
factors are considered as critical institutional contexts for TOL at the school level.

Local Community Circumstances It is often observed that local community’ socio-
economic circumstances or cultures are highly salient to teachers and constitute key
contexts of their work. For instance, in suburban schools, local communities tend to have
much more impact on the schools and give abundant material resources such as
additional funding, amenities, and professional support (Metz, 1990). Teachers in these
schools are inclined to have low expectations for student performance and teachers’
willingness to learn increase. By contrast, local communities located in metropolitan
areas tend to have fewer resources to share with schools (Metz, 1990). Teachers in these
schools are inclined to perceive their students as lacking in motivation, and instituted a
large number of rules and procedures governing cuniculum, and testing procedures (Metz,

1990; Smylie, 1994). In such an environment, teachers’ willingness to learn may decrease.

 

’ The term “institutional context” is often used by sociologists, political scientists, and economists to
describe the phenomena that individual’s preferences and choices are shaped by institutional settings. These
institutional conditions generally include economic institutions (such as the competitive markets),
government regulations, the rules and norms of the formal organizations (For more review, see North,

1990; Ostrom, 1990). McLaughlin and Talbert extend the concept of “institutional context” to the contexts
including the broader cultural arena, local communities’ circumstances, and parents’ demographics and
educational preferences that can affect teachers’ work.

56

For this reason, the current study assumes that local community circumstances have
signiﬁcant relationships with TOL.

Diversity of students ’ race some ethnographic studies (e.g., McLaughlin & Talbert,
2001; Metz, 1990) note that diversity of students’ body in terms of race, culture, and
language within a school is signiﬁcantly related to teachers’ expectations of student
performance and sense of professional effectiveness. For example, Metz (1990) found
that teachers working in the schools that have racially diverse student body often assume
that minority students are less likely to be successful for their academic achievement
because of cultural obstacles attributed to language barriers, learning styles, or any other
factors. In these schools, teachers’ willingness to learn is more likely to decrease. By
contrast, McLaughlin and Talbert (2001) argue that diversity of students’ body can create
pressure for development and teacher learning. In other words, with diversity of students’
race, culture, and language, teachers have been more often challenged to innovate and are
required to participate in more learning opportunities in order to increase these students’
learning. Therefore, the current study expects the positive relationship between diversity
of students’ race and TOL.

Building on these ideas, the second hypothesis about the relationship between

contexts and TOL states that:

H 2: At the school level, social systems contextual factors (professional

learning community), organizational contextual factors (such as supportive

principal leadership, teacher autonomy, teachers’ workload, school size, and school
level), and institutional contextual factors (such as local community circumstances

and diversity of students’ race) are associated with the three TOL indicators.

57

The Rationale for Hypothesis 3

The notion of multiple theoretical perspectives on school contexts also enables us
to identify several district level contextual factors associated with TOL. Despite
skepticism about the potential of district policy actions for supporting educational
improvement, recent studies on school reform indicate that districts have considerable
human, social, and physical resources to inﬂuence teaching practices and learning
opportunities (e.g., see Desimone et al. 2002; Elmore & Burney 1999; Firestone et al.
2004; Hightower, Knapp, Marsh, & McLaughlin 2002; Spillane, 2004). Drawing on the
existing theoretical and empirical literature, this study assumes that six administrative

contextual factors are significantly related with TOL at the district level.

Administrative contextual factors

Districts 'eﬂorts for continuous improvement Several researchers assert that
district efforts for continuous improvement (for example, establishing school goals, using
performance indicators for school improvements, or accountability systems) contributes
to increasing TOL (Desimone et al., 2002; Knapp, 2003; Rosenholtz, 1989). For example,
districts’ efforts for school goal setting have substantial effects on increasing TOL. In a
study of eight districts, for example, Rosenholtz (1989) found that in districts requiring
schools to set instructional goals, teachers were more likely to have improved TOL and
high levels of commitment. Districts’ efforts for school goal setting help schools and
districts to communicate about target local priorities. Through these communications,
districts and schools can provide consistent learning opportunities aligned to district goals

and standards to teachers (Knapp, 2003; Rosenholtz, 1989).

58

District accountability strategies such as performance indicators also lead to
increased teachers’ learning through providing useful information about teachers’
weaknesses (Desimone et al., 2002; Gusky, 1997). These accountability strategies have
the primary assumption that would participate in learning opportunities to make better
progress when they are held accountable for outcomes of TOL and when TOL is
evaluated based on teacher and student outcomes (Desimone et al., 2002). Using a
nationally representative district level data, Desimone and colleagues (2002) found that
district strategies of continuous improvement (including performance indicators for
school improvements) have statistically signiﬁcant relationships with increased
opportunities for active learning and increased teachers’ learning opportunities for at risk
students. Given these empirical supports, the current study expects that district efforts for
continuous improvement are positively related to TOL.

District coordination of multiple funding sources for PD Some evidence
indicates that coordination of multiple funding sources signiﬁcantly affects the quality of
TOL (Desimone et al., 2002; Togneri & Anderson, 2003). For example, Desimone et
al.(2002) found that district coordination of multiple funding sources is the most
important variable to explain differences in the types of TOL (i.e., workshop, mentoring,
and coaching). Case studies also indicate that a pervasive use of external resources
enables support for diverse types of TOL, such as mentoring or inviting expertise
(Togneri & Anderson, 2003).

However, other researchers point to the weak linkage among multiple funding
sources that can yield fragmented and uncoordinated local practices. Further, over-

dependence on external funding sources has been criticized for making long range

59

planning difﬁcult (Massell, 1998). The current study expects negative relationship
between district coordination of multiple ﬁrnding sources and TOL.

Time and ﬁnancial support Although district resources such as time and money
for TOL are frequently considered obvious correlates for teachers’ engagement in
leaming opportunities, there is little evidence about the effects of resources on TOL. One
case study (David et al., 2000) in a study of three districts in Kentucky shows that
teachers in a district which paid stipends for in-service teacher professional development
are more likely to pursue reform ideas grounded in curriculum and instruction. Darling-
Hammond & McLaughlin (1995) note that by having adequate time for learning, teachers
can engage in opportunities to acquire, practice, and reﬂect on new concepts and skills, as
well as opportunities to collaborate and interact with peers. Also, funding for additional
personnel or for additional hours for existing personnel would enable teachers to
participate in new programs, and to collaborate around instruction. Based on these
arguments, the current study assumes that two resources (time and money) for TOL
would be positively related to TOL.

Professional development for Principals It is ﬁ'equently observed that districts’
provision of professional development programs for principals is associated with
improved TOL. For example, the research on New York Community District #2 shows
that principals engaging in professional observation and inter-visitation tend to support
and encourage teachers’ engagement in learning opportunities (Elmore & Burney, 1999;
Stein, D’Amico, & Israel, 1998). The primary assumption about these strategies is that if
districts seek to augment the expertise of principals through formal professional

development, instructional capacity of principals can support teachers’ professional

60

learning.

Yet, the speciﬁc aspects of professional development for principals that foster
TOL are less clear. Some studies document that instructionally-focused training programs
for principals, (such as learning opportunities for critical inquiry, reﬁning pedagogical
repertoires, rigorous discourse, and internal visitation) make improvements for TOL
(Knapp, 2003; Rosenholtz, 1989; Sykes, 2002). Hence, the current study assumes that
principals’ professional development “focused on instructional matters” would constitute
meaningful assistance for TOL.

Decentralization of PD Some analysts believe that district decentralization of
responsibilities for both planning and implementing TOL will promote TOL. The
assumption here is that locally guided decision-making motivates teachers to participate
in their learning opportunities and helps address the knowledge and skills relevant to
teachers’ needs (Desimone et al., 2002; Guskey, 1997; Loucks-Horsley et al., 1998).

However, the effects of district decentralization have been highly contested.
Recent studies indicate that too much control at the school site over TOL is as dangerous
as providing too much control at the district level (Knapp, 2003). For this reason, Knapp
(2003) advocates a balance of responsibilities that enables schools and districts to
complement each other's strengths and weaknesses in the design and conduct of TOL.
Based on this rationale, the current study assumes that decentralization of PD would be
signiﬁcantly related to TOL.

District Size In general, large districts have been associated with both
inefﬁciencies and ineffectiveness in teacher quality and instructional practices. Some

analysts assert that students in large districts too often are taught by novice teachers with

61

emergency certiﬁcation or by experienced teachers who have systematically been
deprived of TOL throughout their careers (Darling-Hammond, 1995).

Other studies argue however that successful investment for TOL in large districts
can lead to improved TOL (Desimone et al., 2002; Firestone, Mayrowetz, & Fairman,
1998; Hannaway & Kimball, 1998). That is, large districts can provide high quality TOL
due to their better infrastructure for planning and delivering TOL, their investments of
greater amounts and sources of funds, and having more access to expertise for supporting
TOL. For example, Desimone et al. (2002) show that larger districts are more likely to
provide much aligned TOL with standards, and to have multiple sources and personnel
such as subj ect-area specialists for supporting TOL. Given the inconsistent empirical
ﬁndings about the impact of district size, the current study will examine the direction of

the effects of district size on TOL in an exploratory way.

District level institutional contextual factors

Parents ’socio-economic condition Some recent studies argue that districts with a
higher proportion of students in poverty are more likely to provide higher quality
professional development for teachers. For example, Desimone et al.’s (2002) study show
that high-poverty districts can employ more continuous improvement efforts for
professional development because they tend to receive funding from multiple federal
programs. The existence of multiple sources of funding may create greater opportunities
for co-funding and create opportunities for supporting teachers’ learning (Desimone et al.,

2002). Furthermore, high-poverty districts are expected to conduct more continuous

62

improvement in TOL than other districts because the funding supports from these federal
programs require the district’s use of accountability strategies.

However, it is frequently assumed that teachers in high-poverty districts
generally often become demoralized and lose their enthusiasm for the difﬁcult task of
improving student performance under difﬁcult conditions (Darling-Hammond, 1995).
That is, institutional contexts such as parents’ low socio-economic conditions are more
likely to negatively affect teachers’ attitudes toward students and instructional practices.
Based on this rationale, the current study assumes that a signiﬁcant relationship between
parents’ socio-economic conditions and TOL would be found at the district level.

Building on these ideas, the third hypothesis about the relationship between

contexts and TOL states that:
H 3: At the district level, administrative contextual factors (such as districts’

efforts for continuous improvement, district coordination of multiple funding
sources for PD, time and financial support, professional development for principals,
decentralization of PD, and district size) and institutional contextual factors (such as

parents’ socio-economic conditions) are associated with the three TOL indicators.

The Rationale for Hypothesis 4

Another main notion of the multi-perspective embedded context model is
‘interactive relationships among multiple contexts’. That is, different combinations of
context conditions can have qualitatively different effects on teachers’ work (McLaughlin
& Talbert, 2001). For instance, states’ mandates for higher standards for instruction are

more likely to lead to the positive effect in school settings where the school culture

63

emphasizes the value that all students have the ability to learn and where there is ongoing
professional learning about the best ways to implement higher standards. Yet, a much less
positive effect on outcomes may occur in a school setting where teachers do not believe
that their students have the ability to meet higher standards or where teachers do not
know how to change their teaching to achieve the new standards (McLaughlin & Talbert,
2001).

Based on this idea of the model, the current study assumes that there are
interactive relationships between district administrative contexts and school level social
system contexts and these interactive relationships make different effects on three TOL
indicators. There must be other interactive relationships among multiple contexts, but a
full explication of all possible interactions is beyond the scope of this study. The current
study only examines interactive relationships between district administrative contexts and
social system contexts in order to explore the possibility of interactions between these
two contextual conditions.

Some studies (Knapp, 2003; McLaughlin & Talbert, 2001; Rosenholtz, 1989)
underscore these interactive relationships between district level administrative contexts
and school level social contexts. For example, Knapp (2003) notes that particular district
policies can support and guide professional learning through strengthening professional
communities. For instance, teacher communities would encourage teachers’ learning if
districts establish school goals, send the consistent message about the importance of
learning, or set norms of collegiality.

Building on these ideas, the fourth hypothesis about the relationship between

contexts and TOL is that:

64

H 4: Interactive relationships between professional learning community and

districts level administrative contextual factors affect three TOL indicators.

The Rationale for Hypothesis 5

The ﬁfth hypothesis of this study is that the inﬂuences of contexts at the school
and district level on TOL vary across the three indicators of TOL. This hypothesis was
not drawn from the multi-perspective embedded context model, but some empirical
studies (Desimone et al., 2002; Scribner, 1999) indicate the different contextual
inﬂuences on TOL depending on the different aspects of TOL. For example, Desimone et
al. (2002) found that teachers’ participation in planning professional development is more
related to active learning for teachers than other aspects of professional development
while district level planning is more related to longer time for professional development.
Scribner’s (1999) case study also shows that school wide factors, such as professional
communities or teachers’ autonomy are more likely to affect teachers’ active learning
than content-focused TOL because local professional communities can encourage
teachers to engage in active learning, such as intellectual discussion and meaningful
feedback occurring during the process of classroom instruction rather than formal
learning environments for teachers. These variations across three TOL indicators may
provide some useful implications for policy strategies for improving the quality of TOL.
Thus, it is reasonable to develop and examine this hypothesis in this study. The current
study assumes, based on some empirical supports, that the effects of district and school
contextual conditions on TOL are diverse across the three TOL indicators (content-focus,

active learning, and duration).

65

Building on these ideas, the ﬁfth hypothesis about the relationship between
contexts and TOL predicts that:

H 5: The relationships between contextual conditions and TOL vary across
the three TOL indicators.

In Chapter V, hypotheses l to 5 are tested to determine the relationship between
district and school contexts and TOL. The general conclusions reached by reviewing the

results of this study are discussed in Chapter VI.

66

CHAPTER IV
RESEARCH METHODS

As noted in the previous chapter, the current study adopts the multi-perspective
embedded context model as a conceptual approach. This model calls for an integration of
quantitative and qualitative research methods in order to capture teachers’ experiences
within the multiple embedded contexts. That is, using quantitative measurement
techniques enables detection of overall tendencies of the effects on teachers’ schooling
experiences at each level of a large number of contrasting multiple contexts while using
qualitative techniques helps account for interdependency or interaction among complex
contexts (McLaughlin & Talbert, 2001)

The current study suggests that using a new quantitative technique (Hierarchical
Linear Model) with a nationally representative data set also enables large-scale
comparative analysis of embedded contexts. Speciﬁcally, the nationally representative
Schools and Stafﬁng and Survey (SASS) enables the McLaughlin and Talbert model to
be extended to the analysis of national samples of teachers within national samples of
schools and districts in embedded school contexts. Further, applying the Hierarchical
Linear Model (HLM) to the multi-perspective embedded context model captures not only
overall tendencies of the effects on teachers’ experiences at each level but also direct or
interactive inﬂuences of embedded contexts. For example, in the preliminary step, HLM
can detect which contextual levels make differences in TOL, by estimating models that
partition the variation in TOL across educational level at the same time. In the next step,
the HLM can detect what speciﬁc factors within each contextual level affect TOL, by

building an explanatory model to account for this variability of TOL. Then, HLM can

67

allow estimation of interactive models, in which the effects of diverse contexts interact
with other contexts to affect TOL. Therefore, this study extends the multi-perspectives
embedded context model methodologically, by utilizing the HLM methodology with
nationally representative SASS data.

Based on such a study rationale, this chapter describes the research design and
methods. It begins with a description of the data source used for this study. It is followed
with a discussion of construction, and properties of the measures used in each research
model. This chapter concludes by describing the statistical methods applied for each

particular analysis.

Data Sources

To estimate a model of the relationship between TOL and embedded district and
school contexts in a broader context, this study uses the 1999-2000 Schools and Staffing
Survey (SASS) sponsored by the National Center for Education Statistics (NCES). The
SASS data is a cross-sectional and large-scale data set that includes public, private, and
charter schools. The SASS data selects a nationally representative random sample of
public schools stratiﬁed by state, sector, district, and school level. A detailed description
of the procedures used to gather these data is found in Tourkin, et al. (2004).

The 1999-2000 SASS includes public sector sample data on 42,086 teachers in
8,432 school and 4,690 districts. However, not all schools had teachers who participated
in the study, not all districts had schools that participated in the study, and not all teachers
could be matched to schools and districts. Thus, the ﬁltering procedures left a total of

32,951 teachers in 6,718 in 3,992. Since the purpose of this study is to estimate the

68

national population of teachers, schools, and districts, this study use weighting functions
assigned by NCES to each teacher, school, district. The weighting produces unbiased
estimates of the national population of public school teachers, including adjustment for
non-response using respondents’ data, and adjustment of the sample totals to the ﬂame
totals to reduce sampling variability (See Tourkin, et al., 2004) Therefore, the current
study uses a weighted data set of 12,159 districts, 68,841 schools, and 2,402,862 teachers.
The SASS provides an advantage over other available datasets in the number and
range of schools and districts included in the survey. It is the only national data set in
which large portions of the survey target schools and districts serve as distinct units of
analysis. Thus, the SASS data offer researchers a powerful tool to evaluate TOL. Second,
the SASS data is also useful because it facilitates multilevel analyses and links data
provided by schools with their respective teachers, principals, and districts. Finally, the
SASS provides appropriate data for analyzing not only teachers’ learning opportunities
and teachers’ perceptions of their school social and organizational conditions but also
districts’ organizational characteristics and policies related to TOL. Yet SASS lacks the
full range of variables one would need to replicate all prior research on TOL. For
example, the SASS data do not provide information on time span, which is one of

variables of TOL used in the previous studies.

Measures
This section describes the variables used to measure effective TOL, district
contexts, school contexts, and teacher background. In addition, this section also discusses

the process of how each of these variables is constructed. A summary of the 2000 SASS

69

items used in constructing these measures is presented in Appendix A. This study begins

by describing the three dependent outcomes measuring three TOL indicators.

Measures of TOL

In Chapter IH, this study identiﬁes three indicators for effective TOL. The ﬁrst
indicator is content-focus, which means that TOL should be directly linked to speciﬁc
subject knowledge, subject curriculum standards, and understanding about students
learning. A second indicator is active learning, which means that TOL should be active
engagement in meaningful discussion, planning, and practice. A third indicator is
duration, which include two dimensions such as total number of contact hours spent in
the learning opportunities, and period of time which the activity was spread. This study
will use theses three indicators as measures of TOL.

The ﬁrst variable of TOL is content-focus, which indicates the degree to which
teachers participate in learning opportunities focused on content. It is constructed by
summing three items that indicated if a teacher had a learning opportunity coincided
with : (1) in-depth study of the content in the teachers’ main assignment ﬁeld; (2) content
and performance standards in teachers’ main assignment ﬁeld; (3) student assessment,
such as methods of testing, evaluation, and performance assessment.

The items used for the content focus variable correspond nearly to the content
focus indicators discussed in the literature. For example, all items used in the current
study were used in similar items by Cohen and Hill (2001). Garet et al. (2001) and
Ingvarson et al. (2005), and Wiley and Yoon (1995) also included the items concerning

in-depth study of the content and content standards.

70

Factor analysis’ conﬁrms that all of these items are associated with a single factor
indicating an eigenvalue of 1.593 and explaining 53.11 percent of the combined variance.
This statistical evidence suggests that the three items work together to measure content
focus TOL. The scale value for the content focus TOL variable ranges ﬂom 0 to 3, with 3
indicating that teachers have participated in learning opportunities focused on content to a
large extent, and 0 indicating that teacher have not participated in learning opportunities

focused on content at all.

k'l"

A second variable of TOL is active learning, which indicates the degree to which

tr! a_. (V. .51-:

teachers engage in active and meaningful learning opportunities such as observation and
presentation. This variable is constructed by summing four items that indicated whether
or not a teacher engaged in learning opportunities of : (l) observational visits, (2)
individual or collaborative research professionally, (3) mentoring, peer observation, and
coaching, and (4) presenting in workshops or conferences.

The items used for this variable correspond with those used by Garet et al. (2001)
and Ingvarson et al. (2005). For example, the ﬁrst item, observational visits was used in
similar form by Garet et al. (2001) and the second item, individual or collaborative
research, was also used in Garet et al.’s study (2001) and in Ingvarson et al.’s study
(2005). The third item, mentoring or peer observation was used by Garet et al. (2001),
and by Ingvarson et al.(2005). The fourth item, presenter in workshops was used by Garet

et al. (2001).

 

2 A factor analysis is an important tool when working with variables that seem to measure similar
constructs. It is employed to deﬁne speciﬁc constructs, or factors, ﬂom the available variables. Through
this analysis, a few general constructs, or factors, can be extracted from a set of variables. This study
employs a principal components analysis as part of an exploratory factor analysis. The principal
components analysis involves the estimation of the variance that each variable shares with the other
variables. This procedure assumes that all of the variance in a given measure can be explained by the
factors resulting ﬂom the analysis.

 

71

Factor analysis conﬁrms that all of these items are associated with a single factor
indicating an eigenvalue of 1.551 and explaining 33.77 percent of the combined variance.
This statistical evidence suggests that the active learning can be measured using these
four items. The scale value for this variable ranges ﬂom 0 to 4, with 4 indicating that
teachers have engaged in active and meaningful learning opportunities to a large extent,
and 0 indicating that teachers have engaged in active and meaningful learning
opportunities at all.

The previous studies have used two variables as measure of duration (for example,
Garet al., 2001; Ingvarson et al., 2005). One is ‘total number of hours’ that teachers spent
in activities related to the professional development program. Another is ‘time span’ that
indicates the total time the professional development activity covered. To deal with
indicating the proximate duration, both variables (total number of hours and time span)
should be used. However, the current study only includes total hours as a third variable
of TOL because the SASS data do not provide information on time span.

It would be ideal to examine total number of hours teachers spent in all
components of the activity related to the TOL during the l-year period. But the items in
SASS only cover the time teachers spent in the six areas of TOL. Thus, to construct this
variable, I summed 6 items concenring the hours teachers spent in the six areas of TOL.
The dimensions covered by these items include how many hours teachers spend on the
following activities in the past 12 months: (1) in-depth study of the content in the
teachers’ main assignment ﬁeld; (2) content and performance standards in teachers’ main
assignment ﬁeld; (3) student assessment, such as methods of testing, evaluation, and

performance assessment; and (4) methods of teaching, (5) discipline and management in

72

the classroom, and (6) use of computers for instruction. (each item scale: 0 = did not
participate; 1 = 8 hours or less; 2 = 9 to 16 hours; 3 = 17 to 32 hours; and 4 = 33 hours or
more) The reliability analysis conﬁrms that it is possible to conclude that the scale of this
variable is reliable for the purpose intended. The reliability estimates for the SASS data
for total hours of TOL variable is 0.66. Table 4-1 provides the operational deﬁnitions of
each of three TOL variables.

Table 4-1 Dependent Variables

 

Variable Names What Is Measured

 

Content-focus The degree to which teachers participate in learning
opportunities focused on content (three dimensions: in-depth
study of the content, content and performance standards, and

student assessment).

 

Active learning The degree to which teachers engage in active and meaningful
learning opportunities (four dimensions: observation, individual

and collaborative research, mentoring, and presentation).

 

Total hours Total hours that teachers have spent on TOL during the 1-year
period in the six areas: content, standards, student assessment,
teaching methods, classroom management, and computers for

instruction.

 

School Level Variables

This study assumed one social system contextual factor (professional learning
community), ﬁve school organizational contextual factors (supportive leadership, teacher
autonomy, teacher’s workload, school level, and school size), and two institutional
contextual factors (community circumstances and diversity of student’s race) are
associated with three TOL indicators at the school level. These factors are operationalized

into eight school level variables. The scales of school level variables are based on items

73

appearing in a number of studies about school level social or organizational features
associated with teachers’ learning or practices advocated by traditional social system
theorists and organizational theory (e.g., Bryk & Driscoll, 1988; Darling-Hammond &
McLaughlin, 1995; Grodsky & Gamoran, 2003; Little, 1999a).

First, professional learning community is measured by one professional learning
community variable indicating the extent to which school staﬂ’ collaborates concerning
instruction and share beliefs about school mission (McLaughlin & Talbert, 2001). This
variable is a mean of ﬁve items. The dimensions covered by these items include the
extent to which staff members (1) share beliefs about the central mission of the school,
(2) cooperate with fellow staff, (3) coordinate course content, (4) plan with the library
media specialist for the integration of media services into teaching, and (5) principals talk
frequently about instructional practices. The items used here are based on components of
professional learning community developed by McLaughlin and Talbert (2001) and these
items appeared in some studies on staff cooperation or collaborative climate for
supporting teacher learning (e.g., Byk et al., 1999; Smith & Rowley, 2005).

Factor analysis conﬁrms that all of these items are associated with a single factor
(or construct) indicating an eigenvalue of 3.234 and explaining 72.6 percent of the
combined variance. The internal consistency of this scale produces an alpha reliability .59.
The school-level measure for this variable is created by taking the mean of teachers’
responses within a school and aggregating at the school level. The scale value for this
variable ranges ﬂom 1 to 4, with 4 indicating that school staff collaborates concerning
instruction and shares beliefs about school mission to a large extent, and 1 indicating that

school staff does not.

74

 

 

Second, supportive leadership is measured by one supportive leadership
variable indicating the extent to which principals are perceived to employ a supportive
leadership by teachers (Byk et al., 1999; Smith & Rowley, 2005). This variable is a mean
of ﬁve items. The dimensions covered by these items include the extent to which the
principal (1) is supportive, (2) lets staff members know what is expected of them, (3)
backs up when teachers as needed, (4) communicates what kind of school s/he wants to
the staff, and (5) school staff are recognized ﬂom teachers a job well done. The items
used here are derived ﬂom the studies about principal leadership supporting learning
environment (e. g., Smith & Rowley, 2005; Zheng, 1996). For example, all items used in
this study were used in identical form by Smith & Rowley (2005) to access supportive
administrations.

Factor analysis conﬁrms that all of these items are associated with a single factor
(or construct) indicating an eigenvalue of 3.234 and explaining 64.68 percent of the
combined variance. The internal consistency of this scale produces an alpha reliability .76.
The school-level measure for this variable is created by taking the mean of teachers’
responses within a school and aggregating at the school level. The scale value for this
variable ranges from 1 to 4, with 4 indicating that principals are perceived to employ a
supportive leadership to a large extent, and 1 indicating that principals are not.

Third, teacher autonomy is measured by two variables, one indicating the degree
of teachers’ collective faculty inﬂuence over school policy, and another indicating the
degree of teachers’ individual autonomy over classroom policy (Ingersoll & Alsalam,
1997). Collective autonomy is a mean of seven items. The items cover a wide-range of

policy dimensions including (1) performance standard, (2) curriculum, (3) in-service

75

professional development programs, (4) evaluation, (5) hiring, (6) discipline policy, and
(7) school budget. All items used in this study correspond with the items used by
Ingersoll & Alsalam (1997).

Factor analysis conﬁrmed that all of these items are associated with a single factor
indicating an eigenvalue of 3.205 and explaining 45.78 percent of the combined variance.
The internal consistency of this scale produces an alpha reliability .80. The school-level
measure for this variable is created by taking the mean of teachers’ responses within a
school and aggregating at the school level. The scale value for this variable ranges ﬂom 1
to 5, with 5 indicating a great deal of teachers’ collective autonomy over school policy,
and 1 indicating no inﬂuence.

Individual autonomy is a mean of six items. The items cover a wide-range of
classroom policy dimensions including (1) course texts, (2) course content, (3) teaching
technique, (4) evaluating students, (5) disciplining students, and (6) determining
homework. All items used in this study were used by Ingersoll and Alsalam (1997) in
identical form to measure individual autonomy.

Factor analysis conﬁrms that all of these items are associated with a single factor
indicating an eigenvalue of 2.921 and explaining 48.69 percent of the combined variance.
The internal consistency of this scale produces an alpha reliability .77. The school-level
measure for this variable is created by taking the mean of teachers’ responses within a
school and aggregating at the school level. The scale value for this variable ranges ﬂom 1
to 5, with 5 indicating a great deal of teachers’ individual autonomy over classroom
policy, and 1 indicating no autonomy.

Fourth, teachers’ workload is measured by two variables, one indicating the

76

amount teachers scheduled school time for planning, and another indicating the number
of students per teacher in the school (Little, 1990a). Planning time is a composite
continuous variable created by NCES. The school-level measure for this variable is
created by aggregating at the school level and standardized. Teacher-student ratio is a
continuous variable that indicates the number of students per full-time equivalent teacher
in the school, also created by NCES.

Fifth, secondary school variable is used in order to indicate the effect of school
level because the current study assumes that teachers in the secondary school are less
likely to engage in TOL. Secondary school is a dummy variable, which takes the value 1
if the school is secondary, 0 otherwise.

Sixth, school size is measured by student total enrollment. Student total
enrollment is the total number of students enrolled in the school in grades K-12.

Seventh, suburban variable’ is used to indicate community circumstances
because the current study assumes that teachers in suburban schools are more likely to
engage in three TOL indicators. Suburban is a dummy variable that indicates the school is
located at suburban area.

Eighth, there is little research that measures ‘diversity of students’ race’ at the
school level. Although there might be several ways to measure ‘diversity of students’
race’, the current study uses percent of minority students as a measure of ‘diversity of
students’ race’. It is a continuous variable indicating the percent of minority students
within schools.

Table 4-2 provides the operational deﬁnitions of each of school level variables

 

3 Conceptually, it makes sense that ‘suburban’ is a district-level variable rather than school-level variable.
However, due to multicollinearity with ‘percent of ﬂee lunch students’ variable at the district level, this
study uses ‘suburban’ as a school-level variable.

77

used in this study.

Table 4-2 School Level Variables

 

School variable

What Is Measured

 

Professional learning
community

Supportive leadership

Collective autonomy

Individual autonomy

Planning time

Teacher-student ratio

Secondary school

Student total enrollment

Suburban

Percent minority

The extent to which school staff collaborates concerning
instruction and share beliefs about school mission (Range
ﬂom 1 to 4)

The degree to which principals are perceived to employ a
supportive leadership style (Range from 1 to 4)

The degree of teachers’ collective faculty inﬂuence over
school policy (Range ﬂom l to 5)

The degree of teachers’ individual autonomy over classroom
policy (Range ﬂom 1 to 5)

The amount of which teachers have scheduled school time
for planning (Range ﬂom 0 to 50 hours)

The number of students per full-time equivalent teacher in
the school

Dummy variable indicating secondary school (second =1,
otherwise = 0)

The total number of students enrolled in the school in grades
K-12

Dummy variable that indicates the school is located at
suburban area (suburban = 1, otherwise = 0)

Continuous variable indicating the portion of minority
students within schools

 

District Level Variables

This study assumed that administrative contextual factors (time and ﬁnancial

resources for TOL, districts’ coordination of multi-funding sources, professional

development for principals, district efforts for continuous improvement, decentralization

of professional development, and district size) and institutional contextual factors

(parents’ socioeconomic condition) are associated with three TOL indicators at the district

level. These factors are operationalized into eleven district level variables. The scales of

78

district level variables are based on items that have appeared in recent discussions of
school district leadership in TOL and research on school district reform (for example,
David et al., 2000; Desimone et al., 2002; Elmore & Burney, 1999; Togneri & Anderson,
2003)

First, time and ﬁnancial resource for TOL is measured by four variables (time
support 1, time support H, money support 1, money support H), which indicate whether
teachers receive time or ﬁnancial support for their professional development. Time
support I is a dummy variable, which takes the value 1 if the teacher receives release
time for teaching (i.e., regular teaching responsibilities are temporarily assigned to
someone else). Time support II is a dummy variable, which takes the value 1 if the
teacher receives scheduled time in the contract year for professional development.
Monetary support I is a dummy variable, which takes the value 1 if the teacher receives
stipends for professional development. Monetary support II is a dummy variable, which
takes the value 1 if the teacher receives reimbursement of college tuition, conferences, or
travel for professional development.

Second, district coordination of multiple ﬁrnding sources is measured by co-
funding variable, which indicates the extent to which districts combine multi-funding
sources for professional development (Desimone et al., 2002). Co-funding variable is
constructed by summing seven items. The dimensions covered by these items include
whether districts use (1) state professional development funds, (2) special project budgets,
(3) school improvement funds, (4) Title I, (5) Eisenhower program, (6) other federal
programs, and (7) private sector grants for TOL. The items used here were used in similar

form by Desimone et al. (2002). They construct a co-funding variable measuring the

79

extent to which districts coordinate funding sources, by summing 10 items. In the current
study, the co-funding variable is a composite variable with a range ﬂom 0 to 7, which
indicates the total number of funding sources for TOL that districts have combined (of a
possible 7).

Third, districts’ provision of professional development for principals is measured
by principal learning variable, which indicates the extent to which districts provide
professional development opportunities for principals involving instructional matter.
Principals learning variable is constructed by summing ﬁve items. The dimensions
covered by these items include whether districts provide (1) evaluation and supervision,
(2) technology for planning, decision-making, and reporting, (3) cuniculum and
assessment, (4) networking Opportunities, and (5) serving as mentors.

The items used here are derived ﬂom studies of the training programs of
principals for augmenting the instructional capacity of district and school leadership
(Elmore & Burney, 1999; Pink & Resnick, 2001; Rosenholtz, 1989). Factor analysis
conﬁrms that all of these items are associated with a single factor indicating an
eigenvalue of 2.49 and explaining 49.74 percent of the combined variance. The scale
value for this variable ranges ﬂom 0 to 5, with 5 indicating that the district provides
professional development programs for principals involving instructional matter to a
large extent, and 0 indicating which to no support.

Fourth, districts’ effort for continuous improvements is measured by two variables,
with one indicating whether the district uses performance indicators to evaluate the
progress of students, and another indicating whether the district reward or sanction

related to performance (Desimone et al., 2002). Performance indicator is a dummy

80

variable, which takes the value 1 if the district uses performance reports. Accountability
is a dummy variable, which takes the value 1 if the district use rewards or sanctions
related to performance.

Fifth, district decentralization of professional development is measured by
teachers’ responsibility for PD variable, which indicates the extent to which teachers
have primary responsibilities for the in-service professional development activities.
Teachers’ responsibility for PD is constructed by summing three items. The dimensions
covered by these items include whether teachers have primary responsibility for
professional development in (1) deciding the content, (2) designing and planning the
activities, and (3) conducting the activities.

The items nearly correspond to the items used by Desimone et al. (2002). They
used the items concerning teacher involvement in planning and conducting the learning
opportunities in order to measure the degree of teachers’ participation in decision-making
for PD. Only teachers’ responsibility for TOL in deciding the content is new for this study.
Factor analysis conﬁrms that all of these items are associated with a single factor (or
construct) indicating an eigenvalue of 1.995 and explaining 66.49 percent of the
combined variance. The scale value for this variable ranges ﬂom 0 to 3, with 3 indicating
that teachers have primary responsibility for professional development to a large extent,
and 0 indicating that teachers do not have.

Sixth, district size is measured by the total number of students enrolled in the

district". District total enrollment is a continuous variable indicating the total number of

 

" There are several ways to measure ‘district size’. For example, Hannaway and Kimball (1998) measure
district size by dividing the total number of students by four levels (such as districts with enrollments
between 300 and 2500, between 2501 and 10,000, and between 10,001 and 25,000+). Desimone et al.
(2001) categorize school districts by three levels of the total number of teachers in the district, such as

81

 

students enrolled in the district in all grade level.

Seventh, parents’ socio-economic conditions is measured by percent of free-

lunch students, a continuous variable indicating percent of K-12 students in district who

were approved for ﬂee or reduced-price lunches. It was created by NCES.

Table 4-3 provides the operational deﬁnitions of each of district level variables

used in this study.

Table 4-3 District Level Variables

 

Variable Names

What Is Measured

 

Time support I

Time support H
Money support I
Money support H
Co-funding

Principal learning

Performance indicator
Accountability

Teachers’
responsibility for PD

District
total enrollment

Percent free-lunch

Whether the teacher receives release time for teaching (0 orl)

Whether the teacher receives scheduled time for professional
development (0 orl)

Whether the teacher receives stipends for professional
development (0 orl)

Whether the teacher receives reimbursement of college tuition,
conferences, or travel for professional development (0 orl)

The extent to which the district combines multi-funding
sources for professional development (Range ﬂom 0 to 7)

The extent to which the district provides professional
development opportunities for principals involving
instructional matter (Range ﬂom 0 to 6)

Whether the district uses performance indicators (0 orl)

Whether the district uses rewards or sanctions related to
performance (0 orl)

The extent to which teachers have primary responsibility for
the in-service professional development activities (Range ﬂom
0 to 3)

Total number of students enrolled in the district in all grade
level

Percent of K-12 students in district who were approved for ﬂee
or reduced-price lunches

 

 

small (1-249 teachers), medium (250-499 teachers), and large (500 or more teachers). The present study
uses the total number of students enrolled in the district, as measured by Bickel and Howley (2000).

82

 

Teacher Background Variable

It is reasonable to expect that some teachers are more predisposed to seek out and
participate in TOL than others and that their personal background characteristics may
explain some of this variance. For example, it is often argued that female teachers tend to
have high expectations and commitment for their instruction and professional leaming
(Miller, 2003). From the literature review, the current study considers how three teachers’
background characteristics (such as gender, teaching experience, and subjects taught) are
related to the TOL indicators. Each of the teacher background variables is constructed as
follows. Female is a dummy variable, which takes the value 1 if the teacher is female.
Teaching experience is a continuous variable indicating the teacher's total number of
years teaching full or part-time in public schools. It was created by NCES. Math is a
dummy variable, which takes the value 1 if the teacher is teaching math. English is a
dummy variable, which takes the value 1 if the teacher is teaching English.

Table 4-4 provides the operational deﬁnitions of each of the teacher background
variables.

Table 4-4 Teacher Background Variables

 

 

Variable Names What Is Measured
Female Female teacher
Total years of experience Teacher's total number of years teaching full or part-
time in public schools
Math Teacher who is teaching math
English Teacher who is teaching English

 

83

 

Statistical Modeling

This study uses Hierarchical Linear Modeling (HLM) to test hypotheses
developed in Chapter HI. By using hierarchical linear models, it is possible to
simultaneously estimate the effects of teacher-, school-, and district-level on three TOL
variables while estimating the effects of district and school level variables across contexts
on three TOL variables. Furthermore, HLM facilitates modeling for cross-level
interaction effects. The current study, using the analytical software package HLM 6 for
Windows estimates the parameters of these models.

Before beginning to model the contextual effects on three TOL variables, it is
important to note the advantage of HLM over other statistical approaches. The traditional
approaches (i.e., OLS Regression) to hierarchically organized data (i.e., teachers sampled
within schools) are to aggregate the lower level units (i.e., teacher or school) to the higher
level (i.e., school or district) or to disaggregate the higher level units to the lower level.
The ﬁrst approach, “higher level as the unit of analysis approach” would lead to
aggregation bias in the estimated effects of lower-level variables on the dependent
outcomes. In such an approach, the dependent variables would be district or school
averages of the TOL variables for each teacher. The use of aggregating data tends to
inﬂate any relationship found between variables. Consequently, such an aggregation
leads to the aggregation bias because of the reduction of random error.

The second approach, “lower level as the unit of analysis,” suffers ﬂom other
methodological problems. In such an approach, variables measuring properties of schools
would be indexed to each teacher and all analyses conducted at the teacher level. This

procedure ignores the fact that groups of teachers within the sample share membership in

84

the same schools. In other words, the second approach assumes that levels of individual
professionalism are independent of any "nesting" of the individual within groups.
However, teachers are inﬂuenced by the school environment in which they work, and
much of the variation in individual levels of professionalism can be attributed to
variations between schools in the overall level of commitment and community among the
teachers. As a result, estimates of school level variables would not yield statistically
independent effects on the teacher repots of TOL. This violation of statistical
independence in ordinary least squares (OLS) regression leads to mis-estimated precision,
where the standard errors of regression coefﬁcients will be deceptively small, leading
analysts to mistakenly assume statistically signiﬁcant effects (Raudenbush & Bryk ,
2002). Because of these problems with the traditional OLS approaches, educational
researchers have increasingly used multilevel statistical models to analyze the effects of
different-level variables. By using hierarchical linear models, we can simultaneously
estimate the effects of different level variables on an outcome without having to choose
between either aggregation or dis-aggregation of data. This study, with a more advanced

statistical method, examines three sets of models in order to test the ﬁve hypotheses.

HLM Analysis for Testing Hypothesis 1

To test the ﬁrst hypothesis that three TOL indicators vary across schools and
districts, this study adopts an unconditional HLM. An unconditional HLM analysis
provides information about how much of the variation in the three TOL variables lies at
the teacher level, at the school level, and at the district level.

The form of these models is presented in equation 1 through equation 3 below.

85

The ﬁrst equation is for level 1 model (within-school). For the level 1 model, the level of
each of three TOL variables for a teacher is modeled as a function of a school mean
across all teachers plus the deviation of each of three TOL variables ﬂom the school
mean (level 1 random error)

Unconditiogrl HLM level-1 equation

Y.. =rt

rjk Ojk + eij

k
(Equation 1)
YU.k : level of each of the three TOL variables (content-focus, active learning, and

total hours) for teacher; in school j in district k

no. : school mean of each of the three variables of TOL in school ,- in district k

jk

eijk : random error of teacher, in school ,- in district k , normally distributed with a

mean of 0 and variance 0'2
The second equation is for the level 2 model (between-school). For the level
2 model, school mean of each of three TOL variables is modeled as an outcome of
district mean of each of three TOL variables plus the deviation of the school mean
ﬂom the district mean (level 2 random error).

Unconditiongl HLM level-2 eguation

“01k = B00k + rOjk
(Equation 2)

trojk: school mean of each of the three variables of TOL in school j in district k
800k : district mean of each of three TOL variables of TOL in district k

rOjk : random error of school j in district k, normally distributed with

mean of 0 and variance Tn.

The third equation is for the level 3 model (between-district). The level 3 model

presents the variability among districts. The district means vary randomly around a grand

86

mean and the deviation of a district’s mean ﬂom the grand mean (level 3 random error)

Unconditiongl HLM level-3 eguation

500k = y000 + u00k
(Equation 3)

000k: district mean of each of three TOL variables in district k

7000 : grand mean of each of three TOL variables

“00k: random error of in district k, is normally distributed with mean of O and

variance 1:13.

The computer program HLM 6 provides estimates for the values of variance for
each level (02, t,I , and 13) and these variance components are used to estimate the

percentage of variance in three TOL outcome variables that lies with-schools, between-

schools, and between-districts.

HLM Analysis for Testing Hypothesis 2 and 3

To test the second and third hypotheses about district and school level factors
related to three TOL variables, an intercepts as outcomes HLM analysis is used. This
analysis estimates the separate effects of each independent variable at the teacher, school,
and district level on three TOL variables. The model of the analysis is presented in
equation 4 through equation 6 below.

The fourth equation is the within-school (level-l) equation of intercepts-as-
outcomes analysis. At level 1 each of three TOL variables for a teacher is modeled as a
function of school mean across all teachers plus the deviation of each of three TOL
variables from the school mean (level 1 random error), and teacher background variables

(female, teaching experience, math, and English).

87

An intercepts Q outcomes HLM level-l equation

Yij" = “Ojk+7tljk *(F)+1t2jk*(Tch. Ex.)+1t3jk*(Math)+n4jk*(Eng)+eUk
(Equation 4)

Yijk’ level of each of the three TOL variables (content-focus, active learning, and

total hours) for teacher; in school ,- in district k

no : school mean level of one of three TOL variables in school j in district 1,,

jk
controlling for the independent variables

mjk ~ 1:4,}: regression coefficient expressing the direction and strength of
relationship between teacher background variables (female, teaching
experience, math, and English) and each of the three TOL variables in

school ,- in district k

e : random error of teacher i in school j in district k

ijk

The ﬁfth equation is the between-school (level 2) equation of intercepts-as-
outcomes analysis. At level 2 expected school mean of each of three TOL variables is
modeled as a function of district mean, the deviation of the school mean ﬂom the district
mean (level 2 random error), and school level variables (professional learning community,
supportive leadership, collective autonomy, individual autonomy, planning time, student-
teacher ratio, secondary school, student total enrollment, suburban, and percent minority).

An intercepts as outcomes HLM level-2 eguation

 

nojk= [500k 4.1301" *(professional learning community) +1302k *(supportive
leadership) +BO3k *(collective autonomy) +1304k *(individual autonomy) +13

05k *(planning time) +B06k *(st_ratio) +BO7k *(secondary school) +1508k

*(student enrollment) +00% *(suburban) +130l 0k *(percent minority) + r0jk

(Equation 5)

nojk : school mean of each of three TOL variables in school j in district k

[300k : district mean of each of three TOL variables (content-focus, active

learning, and total hours) in district k , controlling for the independent

variables

[301k ~00] 0k: regression coefﬁcient expressing the direction and strength of
relationship between school level variables (professional learning

88

 

community, supportive leadership, collective autonomy, individual
autonomy, planning time, student-teacher ratio, secondary school,
student total enrollment, suburban, and percent minority) and

school mean of each of the three TOL variables in district k
rOJ.k : random error of school ,- in district k

The sixth equation is the between-school (level 3) equation of intercepts-as-
outcomes analysis. At level 3 the district mean of each of three TOL variables are viewed
as a function of the grand mean, the deviation of the district mean from the grand mean,
and district level variables (time support 1, time support H, money support I, money
support H, co-funding, principal learning, performance indicator, accountability, teachers’
responsibility for PD, district total enrollment, and percent ﬂee-lunch)

An intercepts as outcomes HLM level-3 equation

[300k = 7000 + 7001*(time I) + 7002*(time H) + y003*(money I) +7004*(money H)
+ 7005*(co-funding) +yOO6*(principal learning) +y007*(perforrnance
indicator) +7008*(accountability) +7009*(teachers’ responsibility for PD)
+yOOIO*(district total enrollment) +700” *(percent ﬂee-lunch) +u00k
(Equation 6)

[300k : district mean of each of three TOL variables (content-focus, active

leaming, and total hours) in district k

7000 : grand mean of each of three TOL variables (content—focus, active learning,

and total hours)

700]... 70011 : regression coefﬁcient expressing the direction and strength of
relationship between district level variables (time 1, time H, money 1,
money H, co-ﬁrnding, principal learning, performance indicator,
accountability, teachers’ responsibility for PD, district total

enrollment, and percent ﬂee-lunch) and district mean TOL level

n : random error of district k

00k

HLM Analysis for Testing Hypothesis 4
To test the fourth hypothesis about interactive relationships between school level

professional learning community and district level administrative contextual factors, a

89

slope-as-outcomes model is used. This analysis estimates cross-level interaction effects
between district level variables and school-level variables. In particular, this study
focuses on analysis of the cross level interactions between district level two variables
(performance indicator and accountability) and the professional learning community
variable. The form of these analyses is presented in equation 7 through equation 9 below.

The seventh equation is the within-school equation (level-1) of a slope-as-
outcomes analysis. This level 1 equation is the same the equation 4 (level 1 equation of
the intercept-as outcomes analysis).

A slope-as-(nrtcomes HLM level-l Equation

Yuk = trojk+1tl jk *(Female)+1t2jk*(Tch. Ex.)ttt3jk'”(Math)+*rr4jk'"(Eng)+eijk
(Equation 7)

Yijk’ level of each of the three TOL variables (content-focus, active learning, and

total hours) for teacher, in school j in district k
ﬂojk : school mean level of one of three TOL variables in school ,- in district k

Ttljk ~ 1t4jk : regression coefﬁcient expressing the direction and strength of
relationship between teacher background variables (female, teaching
experience, math, and English) and each of the three TOL variables in

school ,- in district k

e : random error of teacher, in school ,- in district k

ijk

The eighth equation is the within-school equation (level-2) of slope-as-outcomes
analysis. This equation is also the same as equation 8 (the level 2 equation of the

intercept-as outcome analysis)

A slope-as-outcomes HLM level-2 eg’ uation

nojk= 000k +1301k *(professional learning community) +00% *(supportive
leadership) +3031: *(collective autonomy) +Bo4k *(individual autonomy) +13

05k *(planning time) +B06k *(st_ratio) +Bo7k *(secondary school) +00%

90

*(student enrollment) +13 M *(suburban) +130] 0k *(percent minority) + rOjk

(Equation 8)

“Ojk : school mean of each of three TOL variables in school j in district k
B0011: district mean of each of three TOL variables (content-focus, active

learning, and total hours) in district k
[301k 4301011 : regression coefﬁcient expressing the direction and strength of

relationship between school level variables (professional learning
community, supportive leadership, collective autonomy, individual
autonomy, planning time, student-teacher ratio, secondary school,
student total enrollment, suburban, and percent minority) and

school mean of each of the three TOL variables in district k

rOjk : random error of school j in district k

The ninth and tenth equations are the between-district equation (level-3) of
slope-as-outcomes analysis. They address cross-level interactions between district level
and school level. The ninth equation remains the same as the equation 6 (the level 3
equation of intercept-as outcome model), but the tenth equation adds at the level-3 model.
The tenth equation indicates performance indicator (or accountability) variable at the

district level of analysis moderates regression coefﬁcient (801k ) expressing the

relationship between professional learning community variable and school mean total

hour (or active learning) variable.

A slope-is—outcomes HLM level-3 equation

[300k = 7000 + 7001 *(time I) + 7002*(time H) + 7003*(money I) +7004*(money H)
+ 7005*(co-funding) +7006*(principal learning) +7007*(performance
indicator) +7008*(accountability) +7009*(teachers’ responsibility for PD) +

*(district total enrollment) +700” *(percent ﬂee-lunch) +u00k

(Equation 9)

70010

= * . . . .
[501k 7010 + 7010 (performance rndrcator or accountability) + "01k

(Equation 10)

91

7000 : grand mean of total hour (or active learning)

7001~ 70011 : regressron coefﬁcrent expressrng the direction and strength of

relationship between district level variables (time I, time H, money 1,
money H, co-funding, principal learning, performance indicator,
accountability, teachers’ responsibility for PD, district total
enrollment, and percent ﬂee-lunch) and district mean TOL level

“00k : random error of district k

[301k : regression coefﬁcient expressing the direction and strength of the

relationship between school level professional learning community
variable and school mean total hour (or active learning) in district k

In the models estimated here, all school- and district-level predictors are
centered around their sample grand mean values because this study assumes that
variations in TOL are estimated across the entire sample of schools, not within each
school. When grand mean centering is used, the intercept can be interpreted as an
adjusted mean ﬂom school j and the variance of the intercept is the variance among

schools with adjusted means (for more review, see Kreft & Leeuw, 1995).

Analytical Model for the Study

Figure 4-1 displays an analytical model to test ﬁve hypotheses developed in this
study. This model was developed based on McLaughlin and Talbert’s multi-perspective
embedded context model.

First of all, the horizontal headings indicate the multiple level of educational
system in this model: district level, school level, and teacher level. These multiple levels
were developed based on the assumption of embedded contexts that each of multilevel
contexts shapes teachers’ work. Based on this assumption, this study tests the ﬁrst

hypothesis that three TOL indicators vary at the embedded district and school level.

92

Second, the vertical headings indicate multiple theoretical perspectives on school
contexts, such as institutional perspective, administrative perspective, and social system
perspective. I integrated these three theoretical perspective in the model based on
McLaughlin and Talbert’s argtunent that theories must be integrated to capture the
embedded characters of school contexts. The boxes in the ﬁgures 4-1 indicate contextual
factors (variables) of each level of educational system, drawn ﬂom these multiple
perspectives. For instance, professional learning community is a social system contextual
factor at the school level. Solid lines indicate direct effects of these contextual factors
(variables) on three TOL indicators.

Third, the bold lines indicate the interactions between district administrative
factors (variables) and school level professional learning community. In these
interactions, administrative factors (variables) at the district level moderate the effects of
professional learning community on three TOL indicators. With this analytical model,
this study will test the hypotheses regarding the relationship between district and school

contexts and three TOL indicators in the next chapter.

93

 

‘ Eng—Eco .83 «5th J

.30. geooom
BE 2333280...
2:: mesa...
>529...” 3:239:
.8855“ 0260—30

@5530. EEoEn oztonqem

k

 

 

 

W 4
ﬂ a8: .85 p

h 36832841 ,. .. .......... _ , ,, .............
73......er .._g§.....§.__§_....&

asses doe k
a

532
.235.
35:35 .«o man» :38.
285.;

AH>HA ¢m=0<mh AH>HA AOOIUm

 

 

 

 

 

 

 

 

btoEE “neocon—
535:5

   
  

 

40H 98 35:80 Eon—om new drama 5.233 Enmcowﬂem on“ ouch—wee 9 new: 682.4 g

Soto 830835 Ere—-380 ”
Soho 825 ”

 

\ Eves—930 .82 again—J .

Dn— Bu bani—5&2 .Eonomok

 

      
     

553582
.2865 cue—agate;
wage. 3305.5
365.8

= a. _ ton—96 >252

 

= d _ conga 08:. 38:3 o>€¢h$£ﬁp<

. k
_

25289 589mm Boom

 

 

Amman-— 828m:—

TO
I@

ﬁgment 2:83; 382.8 3:32.33:—

-41 use...

94

CHAPTERV

RESULTS AND DISCUSSION

This chapter presents the test results of hypotheses proposed in Chapter HI and,

provides a discussion of the ﬁndings regarding the relationship between district and

school contexts and TOL. It begins with a brief presentation of the descriptive statistics

and correlation analysis for all variables included in the data analysis. The second section

presents the results ﬂom the HLM analysis. Finally, this chapter concludes with a

discussion of signiﬁcant ﬁndings.

Figure 5-1 represents the roadmap, which helps the reader to understand how the

chapters proceed.

Figure 5-1 Roadmap representing the Structure of How the Chapters proceed

 

Chapter II \

CD
gain notions of Model

 

 

 

Direct inﬂuence

 

 

 

 

(43
/ Chapter V \

Results &Discussion

 

 

D Result 1, Discussion 1

Result 2&3, Discussion 2

 

 

 

Multiple-theoretical
perspectives

 

Interactive relationship

 

Results 4

/'

 

® 6)
( Chapter III \ K Chapter IV \
Hypotheses HLM analysis
# H1 >Un-conditional HLM
; H2, H3 Intercepts-outcomes
HLM
47 H4 Slopes-outcomes HLM
H5

 

 

 

k J

 

L._J

 

 

k J

P Result 5, Discussion 3

 

 

 

The panel (.1) represents the main notions of the multi-perspective embedded

context model -— direct inﬂuences of each multiple contexts, interactive relationship

among contexts, and multiple theoretical perspectives on school contexts(social system

95

perspectives, administrative perspectives, and institutional perspectives). All of these
were reviewed in Chapter H. The panel ® represents how these three main assumptions
of the model are operationalized into ﬁve hypotheses to explore the relationship between
district and school contexts and TOL. Four of them are directly drawn ﬂom the multi-
perspective embedded context model while the ﬁfth hypothesis is drawn ﬂom some
empirical studies of contextual inﬂuences on TOL. All of these were discussed in Chapter
H1. The panel ® represents the statistical modeling used for testing the ﬁve hypotheses.
Three HLM analysis models are used for this study; unconditional HLM, an intercepts as
outcomes HLM, and a slopes as outcomes HLM. All of these were noted in Chapter IV.
The panel C4) represents the results from each statistical modeling and a discussion of the
signiﬁcant ﬁndings. The results and discussions will be provided in the second and third
section of this chapter (Chapter V) respectively.

Speciﬁcally, ﬂom the notion of direct inﬂuences of each the multiple contexts on
teachers’ work, this study developed the ﬁrst hypothesis that three TOL indicators
(content-focus, active learning, and duration) vary across schools and districts (see the
arrow 69). This hypothesis was tested using an unconditional HLM, which provides
information about how much of the variation in the TOL outcome variables lies at the
teacher level, at the school level, and at the district level(see the arrow @). Then, ﬂom the
HLM analysis, this study provides result 1 in the second section, and discussion
regarding district and school level variability of TOL (discussion 1) in the third section of
this chapter (see the arrow O).

From the notion of integration of multiple theoretical perspectives, this study

developed the second hypothesis that at the school level, social system contextual factors,

96

administrative contextual factors, and institutional contextual factors associate with TOL,
and the third hypothesis that at the district level, administrative contextual factors and
institutional contextual factors associate with TOL (see the arrow CID). These two
hypotheses were tested using an intercepts as outcomes HLM, which estimates the
separate effects of the each independent variable at the teacher, school, and district level
on TOL outcome variables (see the arrow (9). From the HLM analyses, this study
provides results 2 and 3 in the second section, and discussion regarding contextual factors
related to TOL regarding district and school level variability of TOL (discussion 2) in the
third section of this chapter (see the arrow 61)).

From the notion of interactive relationships among multiple contexts, this study
developed the fourth hypothesis that interactive relationships between professional
leaming community and districts administrative contextual factors affect three TOL
indicators (see the arrow ©). This hypothesis was tested using a slopes as outcomes HLM,
which estimates the cross-level interaction effects between district level administrative
factors and school-level professional learning community (see the arrow CD). From the
HLM analyses, this study provides result 4 in the second section, and discussion
regarding contextual factors related to TOL regarding district and school level variability
of TOL (discussion 2) in the third section of this chapter (see the arrow (D).

Finally, based on some empirical studies on the relationship between contexts and
TOL, this study develops the ﬁfth hypothesis that the relationship between contextual
conditions and TOL varies across the three indicators of TOL. The results of this
hypothesis test is reported in result 5 and discussed in discussion 3 (see the arrow (D).

Before testing the proposed hypotheses, this study presents the descriptive

97

statistics for all variables included in the data analyses. These statistics are particularly
interesting as they are derived ﬂom a large, nationally representative data of teachers,
schools, and districts in the United States. This chapter next presents the correlations
among selected variables in the study as they provide useﬁrl information about

relationships among the independent variables.

Descriptive Statistics
Table 5-1 shows the descriptive statistics (such as mean, standard deviation,
maximum value, and minimum value) for the teacher, school, and district-level variables
in this study. Teacher-level descriptive statistics are weighted using the SASS teacher
weights, school-level descriptive statistics are weighted using the SASS school weights,
and district-level descriptive statistics are weighted using the SASS district weights. As
this is a large, nationally representative database, some of the descriptive data are worthy

of review.

98

Table 5-1 Descriptive Statistics (The entries in this table are weighted)

 

 

 

 

Variable Mean SD Min Max
Teacher level Content-focus l .95 1 .03 0 3
(N=2,402,862) Active learning 1.98 1.2 0 4
Total hours 7.53 4.56 0 24
Female .75 .43 0 1
Teaching experience 14.71 10.1 1 57
Math .07 .25 0 1
English .11 .30 0 1
School level Prof. Learning community 2.89 .35 l 4
(K=68,841 ) Supportive leadership 3. 1 7 .49 1 4
Collective autonomy 2.58 .55 l 5
Individual autonomy 4.04 .44 l 5
Planning time 9.93 4.26 0 50
Student-teacher ratio 15.56 13.04 .55 745.60
Student total enrollment 544.66 432.37 2 5380
Secondary school .25 .43 O 1
Suburban .43 .49 0 1
Percent minority 32.71 33.07 0 100
District level Time support I (release) .75 .24 o 1
”—12:59” Time support 11 (scheduled) .59 .28 o 1
Money support I (stipend) .40 .30 0 1
Money support H (reimburse) .60 .29 0 l
Co-funding 4.91 l .70 0 7
Tchr. Responsibility for PD .72 1.06 0 3
Principal learning 3 .3 l 1.72 0 5
Accountability .20 .3 1 0 1
Performance Indicator .96 .19 0 1
District total enrollment 3530.77 6658.27 2 1,093,071
Percent ﬂee-lunch 33.44 23.57 0 100

 

99

In terms of teacher characteristics, almost 75% of the US. teachers are female.
Teachers average 14.71 years of teaching experience. English teachers compose 11% of
the US. teachers while math teachers compose 7% of the US. teachers.

The descriptive statistics also show that the US. teachers report a medium level
of professional learning community (mean: 2.89, range ﬂom 1 to 4), while they report a
somewhat high level of supportive leadership of principal (mean: 3.17, range ﬂom 1 to 4).
The US. teachers report a very high level of their individual control over classroom
policy (mean: 4.04, range ﬂom 1 to 5) while they report a medium level of collective
faculty inﬂuence over school policy (mean: 2.58, range ﬂom 1 to 5). The average level of
student-teacher ratio in US. is approximately 15 and the US. teachers have on average
10 hours scheduled time for planning in a week. The mean of student enrollment is
545.66. The percentage of secondary school in US. is 25% while the percentage of
schools located at suburban is 43%. The school average of minority student enrollment is
32.71%.

The descriptive statistics indicate that 75% of teachers in US. receive release
time for teaching at the district level while 59% of teachers receive scheduled time for
professional development. 40% of teachers in US. receive stipends for professional
development at the district level while 60% of teachers receive reimbursement for college
tuition, conferences, or travel for professional development. The data show that the US.
districts provide a somewhat high level of professional development opportunities for
principals regarding instructional matters (mean: 3.31, range ﬂom O to 5) and coordinate
high level of multiple funding sources for professional development (mean: 4.91, range

from 0 to 7). The average level of districts’ report of the degree of which teachers have

100

primary responsibility for professional development is 0.72, with range ﬂom 0 to 3. 96%
of the districts report that they use performance indicators to evaluate the progress of
students while 20% of the districts reports that they use rewards or sanctions related to
performance. The mean of district enrollment is 3530.77. The average of students

participating in the national school ﬂee lunch program is 33.44%.

Correlations
The correlation matrix representing all variables used in this study is presented
on Table 1 and is placed in Appendix B’. to improve the efﬁciency of data presentation.
The correlation matrix includes the complete list of items and scales used in this research,
but the following discussion will be limited to a select set of measures. Table 1 in
Appendix B. shows the bivariate correlations among all variables included in this study
presented at the teacher-level. These correlations are a preliminary form of analysis

providing an estimate of expected associations and strength of relationships.

Correlations among Outcome Measures

Content-focus is correlated with total hours at .66 (p < .01), and active learning is
correlated with total hours at .37 (p <.01). The relationship between content-focus and
active learning shows strength at .34 (p <.01). The correlation results indicate that each of

these TOL variables is related constructs, but not to the extent that multi-collinearity

 

5 This study measured the correlation between two variables, using the Pearson’s r, the most common type
of measure of correlation. It also used two special types of correlation to handle the special characteristics
of such types of variables as dichotomies. For example, when both variables are dichotorrries (i.e., female
vs. math teacher), this study used Phi and when one variable is a true dichotomy and the other is continuous
variable (i.e., female vs. teaching experience), it used Point-biserial correlation (for more review,
http://www2.chass.ncsu.edu/garson/PA765/correl.htm)

101

issues would arise in advanced data analyses. Moreover, each item is ﬂoated as a separate
independent variable in the series of advanced statistical models used in this study.

While these three outcomes are ordinal, linear models were used. While not
standard, Goodman (1978) suggests that the use of linear models with ordinal outcomes

is merited in cases where the distribution of responses ﬁt a particular pattern.

Correlations of Teacher Background to Three TOL Variables

All teacher background variables were signiﬁcantly related to three TOL
variables (p <.01). Although effect size for these correlations were small, the correlation
coefﬁcients of the teacher background variables with the three dependent variables
ranged from -.03 to .14 for content-focus variable, ﬂom -.05 to .02 for active learning
variable, and ﬂom -.03 to .11 for total hours variable.

Among the teacher background variables, female variable had the highest
positive correlation coefﬁcients for content-focus (.14), and total hours (.11). The
correlation between female and active learning was also positive but very small (.02).
This ﬁnding indicates that female teachers appear more likely than male teachers to
report higher levels of content-focus learning opportunities and to spend more time on
their learning.

The relationships of years of teaching experience and three TOL variables were
signiﬁcantly correlated but the correlation coefﬁcients are small (content-focus: .06,
active learning: .02, and total hour: .05). English teacher variable also shows quite small
correlations with all three TOL variables (content-focus: .02, active learning: .01, total

hours: .01). Math teachers and all three TOL variables were negatively correlated with

102

low correlation coefﬁcients (content-focus: -.03, active learning: -.05, total hour: -.03).

Correlation of School Level Variables to Three TOL Variables

All school level variables, except one variable (suburban) were signiﬁcantly
related to three TOL variables (p <.01). The correlation coefﬁcients of the school level
variables with the three dependent variables ranged ﬂom -. l 7 to .11 for content-focus
variable, ﬂom -.06 to .12 for active learning variable, and ﬂom -.11 to .11 for total hours
variable.

Among the school level variables, secondary level variable had the highest
correlation coefﬁcients for content-focus (-.17) and total hours (-.11). Professional
learning community had positive and high correlation coefﬁcients for content-focus (.11),
active learning (. 12), and total hours (.11). These ﬁnding indicate that teachers in
secondary schools are less likely to participate in content-focus and to spend less time on
their learning opportunities while teachers in professional learning community are more
likely to engage in content-focus, and active learning and to spend more time on learning.

Collective autonomy variable and all three TOL variables were positively
correlated (content-focus: .06, active learning: .1, and total hours: .08), while individual
autonomy variable and all three TOL variables were negatively correlated with small
correlation coefﬁcients (content-focus: - .07, active learning: .003, and total hours: -.O4).
Percent of minority students was positively related to all three TOL variables. In
particular, the strength of the correlations between percent of minority students and three
TOL variables were modest ( .1, .04, .1 respectively). The relationships of supportive

leadership and three TOL variables were signiﬁcantly correlated but the correlation

103

coefﬁcients were modest (content-focus: .05, active learning: .08, and total hours: .05).
While planning time for instruction, student-teacher ratio, and school size were
signiﬁcantly related to three TOL variables, the strength of the correlation between these
school level variables and TOL variables were small. Suburban was positively correlated
with active learning (.02), and negatively related with total hours variable (-.02) but the
strength of these correlations were very weak. Furthermore, suburban was found to be

insigniﬁcantly related to content-focus variable.

Correlation of District Level Variables to Three TOL Variables

Most district level variables were signiﬁcantly related to three TOL variables (p
<.01). The correlation coefﬁcients of the district level variables with the three dependent
variables ranged ﬂom .003 to .19 for content-focus variable, from -.01 to .15 for active
learning variable, and from .001 to .20 for total hours variable.

Overall, time or money supports for TOL were more moderately correlated with
three TOL variables than were any other district variables. For example, both time
support I (release) and time support H (scheduled time) had moderate correlation
coefﬁcients for content-focus (.16, .15), active learning (.15, .10), and total hours
(. 14, .09). The relationships of money I (stipend) and money H (reimbursement) to all
three TOL variables were also moderately correlated (m I: .2, .13, .20, m H: .10, .03, .14,
respectively), except the correlation between money H and active learning (.03). These
ﬁndings suggest that time or money supports for TOL would be important factors for
TOL.

Principal learning was correlated to marginally higher levels of all three TOL

104

variables (content-focus: .07, active learning: .04, and total hours: .05). Co-ﬁrnding and
accountability had modest correlation coefﬁcients for content-focus (.05, and .04
respectively), and total hours (.04, and .03 respectively) while these independent
variables had very small correlation coefﬁcients for active learning (.007, and .008
respectively). Performance indicator and teachers’ primary responsibility for PD were
positively correlated with three TOL variables, but correlation coefﬁcients were quite
small. Furthermore, the relationship between teachers’ responsibility and total hours
variable was insigniﬁcant.

Percent of ﬂee lunch students was positively correlated with content-focus and
total hours but this variable is negatively related to active learning. However, the
strengths of these correlations were modest.

District total enrollment had marginally high correlation coefﬁcients for content-
focus (.07) and quite small correlation coefﬁcients for total hours (.02). The relationship
between district total enrolhnent and active learning was insigniﬁcant.

It should be recalled that this preliminary analysis is performed at the teacher
level. A properly controlled multilevel model, performed at the appropriate levels of

aggregation, may yield the expected results in a more consistent manner.

HLM Model Results
This section presents the results of the HLM analysis. It begins with the results of
the ﬁrst hypothesis in which I partitioned each of the TOL indicators for each analysis

series into within-and between-school, and between-district variance components.

105

Result 1: Results of the First Hypothesis Test

The ﬁrst hypothesis for this study was that the three TOL indicators vary across
schools and districts. The analysis results from an unconditional (i.e., no predictors at the
teacher, school, or district levels) Hierarchical Linear Model (HLM) conﬁrms the ﬁrst
hypothesis. An unconditional HLM analysis provides useful information about how much
of the variation in the three TOL outcome variables lies at the teacher level, at the school
level, and at the district level.

For the level 1 (within-school) model, the level of each of three TOL variables
for a teacher is modeled as a ﬁrnction of a school mean across all teachers (level-1
intercept) plus the deviation of each of three TOL variables ﬂom the school mean (level-
1 random error term). For the level 2 model, school mean of each of three TOL variables
is modeled as an outcome of district mean of each of three TOL variables (level-2
intercept) plus the deviation of the school mean ﬂom the district mean (level-2 random
error term). The level 3 model presents the variability among districts. The district means
vary randomly around a grand mean and a random district effect, that is, the deviation of
a district’s mean ﬂom the grand mean. In all cases, the random effects are assumed to be

normally distributed with a mean of O. The computer program HLM 6 provides estimates
for the values for each level variance (0’, t,t , and t3) and these variance components are

used to estimate the percentage of variance in three TOL outcome variables that lies
within-schools, between-schools, and between-districts.

Table 5-2 reports the results of this variance decomposition. From the HLM
analysis results, variances in the three TOL variables are found across three levels -

teachers, schools, and districts. Although the largest percentage of the total variance is

106

among teachers, a substantial, though smaller, percentage lies between schools within
districts, and a yet smaller percentage lies between districts. These variations across
levels are statistically signiﬁcant.

Speciﬁcally, for the content-focus, approximately 69% of the variance in
teachers’ reports lies in teachers, 6% of the variance lies among schools, and
approximately 25% of the variance among districts. For the active learning variable,
approximately 69% of the variance in teachers' reports lies in teachers, 6% of the
variance lies among schools, and approximately 25% of the variance among districts. For
the total hours variable, approximately 74% of the variance in teachers' reports lies in
teachers, 6% of the variance lies among schools, and approximately 20% of the variance
among districts. Hence, as hypothesized in this study, teachers’ opportunities to learn

vary across schools and districts.

107

Table 5-2 Unconditional HLM Analysis for Three TOL Variables (Content-focus, Active

learning, and Total hours)

 

Model I Content-focus

 

Fixed Effects
Model intercept
(Average district mean)

Coefﬁcient
l .822

P- value
0.000

 

 

Percent of Total

 

 

 

Random Effect Variance Component Variation
Level 1- Teachers 0.684 68.9%
Level 2 - Schools 0.056 5.6%
Level 3- Districts 0.252 25.5%
Model 11 Active learning

Fixed Effect_s_ Coefﬁcient P value
Model intercept 1.846 0.000

(Average district mean)

 

 

Percent of Total

 

 

 

 

R__andom Effect Variance Component Variation
Level 1- Teachers 0.908 69.5%
Level 2 — Schools 0.074 5.6%
Level 3- Districts 0.325 24.9%
Model III Total hours
Fixed Effecﬁ Coefﬁcient P value
Model intercept 7.136 0.000
(Average district mean)

Percent of Total
Random Effect Variance Component Variation
Level 1- Teachers 13.737 73.6 %
Level 2 — Schools 1.101 5.9 %
Level 3- Districts 3.829 20.5 %

 

108

Result 2: Results of the Second Hypothesis Test

The second hypothesis for this study was that at the school level, social system
contextual factors (such as professional learning communities), administrative contextual
factors (teacher autonomy, supportive leadership, teachers’ workload, school level, school
size), and institutional contextual factors (such as community circumstances and diversity
of students’ race) are associated with three TOL indicators. The results ﬂom ‘an
intercepts as outcomes HLM’ analysis conﬁrm the second hypothesis. This HLM model
allows estimation of the separate effects of the each independent variable at the teacher,
school, and district level on three TOL indicators.

Building on the previous model (unconditional HLM), teacher background level
variables, school level variables, and district level variables are introduced as separate
predictors at the ﬁrst, second, and third levels, respectively. At level 1 each of three TOL
variables for a teacher is modeled as a function of school mean across all teachers plus
the deviation of each of three TOL variables ﬂom the school mean, and teacher
background variables (such as female, teaching experience, math teacher, and English
teacher). At level 2 expected school mean of each of three TOL outcomes is modeled as a
function of district mean, the deviation of the school mean ﬂom the district mean, and
school level explanatory variables (e.g., professional learning community, supportive
leadership, and collective autonomy). At level 3 the district means of each of three TOL
outcomes are viewed as a function of the grand mean, the deviation of the district mean
from the grand mean, and district level explanatory variables (e. g., time support 1, money
support I, and principal learning).

In multilevel analysis, the intercept is interpreted as the expected value of the

109

outcome value, when all explanatory variables have the value zero. For instance, the
intercept at level 1 is interpreted as the expected outcome of TOL variable for the average
person, disregarding teacher background variables such as female. To aid 3 model
interpretation, all variables at the teacher, school, and district level are grand mean
centered. When the explanatory variables are grand mean centered, the mean of each
variable is subtracted ﬂom each teacher’s raw score to rescale each variable to have mean
of 0 (Hox, 2002). By centering around the grand mean, the interpretation of coefﬁcient of
dichotomous variables such as gender (coded as 0=male, l=female) become how much
the outcome variable is expected to change as the ratio of male and female teachers
changes within a school. The intercept of the model at any level becomes expected value
for the average unit within the context, controlling for the various independent variables
(for more review, see Hox, 2002).

The results of this study show that the variability of TOL at the school is
associated with social system contextual factors, administrative contextual factors, and
institutional contextual factors. Before presenting the analysis results of school level
contextual factors, it is also meaningful to examine the relationship between teacher
background variables and TOL. Table 5-3, 5-4, and 5-5 present the relationships of
teacher background variables used in this study with three TOL variables.

Most teacher background variables are signiﬁcantly related to TOL variables.
First, female teacher is signiﬁcantly related to three TOL variables. The coefﬁcients for
female teacher are .17 /-.07 / .58 for content-focus, active learning, and total hours
respectively. This means that as the ratio of females to males increases, teachers in a

school report on average experiencing a level of content-focus 0.17 units high and

110

spending a 0.5 hours more on their learning, while they report on average experiencing a
level of active 0.07 units low. These results imply that female teachers are more likely to
participate in the opportunities for content-focus and spend more time on their learning
than male teachers while they are less likely to participate in the opportunities for active
learning than male teachers.

Second, teaching experience has a positively signiﬁcant relationship to three
TOL variables. The coefﬁcients for teaching experience are .007/ .005/ .024 for content-
focus, active learning, and total hours respectively. These results are interpreted to mean
that as years of teacher experience increase by l, the level of three TOL variables
increase by .007, .005, and .024 units. These results imply that teachers with more
experience tend to participate slightly more in opportunities for content-focus and active
learning, and spend slightly more time on their learning than those with less teaching
experience.

Third, math teacher is only signiﬁcantly related to active learning. The
coefﬁcients for math teacher are -.11 for active learning. This means that as the ratio of
math teachers to other subject teachers increase, teachers in a school report on average
experiencing a level of active learning 0.11 units lower. This implies that math teachers
are less likely to participate in the opportunities for active learning than those taught other
subjects.

Finally, English teacher is only signiﬁcantly related to content-focus variable.
The coefﬁcients for English teacher are .11 for content-focus. This means that as the ratio
of English teachers to other subject teachers increase, teachers in a school report on

average experiencing a level of active learning 0.11 units higher. This implies that

111

English teachers are more likely to participate in the opportunities for content-focus than
those teaching other subjects.

Table 5-3, 5-4, and 5—5 also present the results of HLM analyses regarding the
relationships of school level contextual factors to the three variables of TOL. It provides
the HLM estimates of coefﬁcients for the school-level contextual factors related to the
three TOL variables. Each coefficient provides a measure of direct relationship of a given
independent variable to the speciﬁed outcome variable. The results of the second
hypothesis test are here represented in terms of three contexts: social system contexts,

administrative contexts, and institutional contexts.

Social system contextual factor

This study hypothesizes that social system contextual factors are associated with
three TOL indicators. In this study, the social system contextual factor was measured by
professional learning community, which indicates the extent to which school staff
collaborates concerning instruction and share beliefs about school mission. The

dimensions of this variable are as follows.

 

Dimensions of professional learning community variable

 

The extent to which staff members (1) share beliefs about the central mission of the
school, (2) cooperate with fellow staff, (3) coordinate course content, (4) plan with the
library media specialist for the integration of media services into teaching, and (5)
principals talk ﬂequently about instructionaliractices (Range from 1 to 4)
As hypothesized in this study, professional learning community is signiﬁcantly
related to three TOL variables. The coefﬁcients for professional learning community

are .22 / .32 / 1.1 for content-focus, active learning, and total hours respectively. This

means that as the professional learning community level increase by one unit, three TOL

112

indicators are expected to increase by .22 units of content-focus, .32 units of active
learning, and 1.1 hours for learning opportunities. The results imply that teachers who
experience a higher level of professional learning community are more likely to engage
in opportunities for content-focus, and active learning and spending more time on their
learning.

The results from this study support the assumption that building professional
learning communities which cooperate concerning instruction and share beliefs about
school mission are important for fostering TOL. The discussion regarding the important

ﬁndings will be provided in detail in the discussion section.

School organizational contextual factors

This study assumed that school organizational contextual factors (such as
supportive leadership, teacher autonomy, teacher’s workload, school size, and school
level) are associated with three TOL indicators. In this study, these school organizational
contextual factors are measured by the following variables. First, supportive leadership
was measured by supportive leadership variable, which indicates the extent to which

principals are perceived to employ a supportive leadership style.

 

Supportive leadership variable: The extent to which the principal (1) is supportive, (2)
lets staff members know what is expected of them, (3) backs up when teachers as
needed, (4) communicates what kind of school s/he wants to the staff, and (5) school
staff are recognized for a job well done. (Range ﬂom 1 to 4)

 

Second, teachers’ autonomy was measured by two variables, one indicating the
degree of teachers’ collective faculty inﬂuence over school policy, and another indicating

the degree of teachers’ individual autonomy over classroom policy.

113

 

Collective autonomy variable: The extent to which teachers have autonomy over
school policy dimensions including (1) performance standard, (2) curriculum, (3) in-
service professional development programs, (4) evaluation, (5) hiring, (6) discipline
policy, and (7) school budget. (Range ﬂom l to 5)

 

Individual autonomy variable: The extent to which teachers have control over
classroom policy dimensions including (1) course texts, (2) course content, (3) teaching
technique, (4) evaluating students, (5) disciplining students, and (6) determining
homework. (Range ﬂom 1 to 5)

 

Third, teachers’ workload was also measured by two variables, one indicating the
amount of which teachers have scheduled school time for planning, and another

indicating the number of students per teachers.

 

Planning time for instruction: The amount of time which teachers have scheduled

school time for planning (Range ﬂom 0 to 50 hours)

Student-teacher ratio: The number of students per full-time equivalent teacher in the
school

Fourth, school size was measured the total number of students enrolled in this
school in grades K-12. Finally, school level was measured by one dummy variable
indicating secondary school.

Among school organizational contextual factors proposed in this study, teacher
autonomy and school level are signiﬁcantly related to three TOL indicators. Speciﬁcally,
collective autonomy is positively related to three TOL variables. The coefﬁcients for
collective autonomy are .17 / .25 / .69 for content-focus, active learning, and total hours
respectively. This means that as the collective autonomy increase by one level, three TOL
indicators are expected to increase by 0.17 units of content-focus, 0.25 units of active
learning, and 0.69 hours of learning opportunities. This implies that teachers who have
more collective autonomy in school policy are more likely to engage in opportunities for

content-focus, and active learning and spending more time on their learning.

By contrast, individual autonomy is negatively related to three TOL variables.

114

The coefﬁcients for individual autonomy are -. 12 / -. 14 / -.36 for content-focus, active
learning, and total hours respectively. This means that as individual autonomy increases
by one unit, three TOL indicators are expected to decrease by 0.12 units of content-focus,
0.14 units of active learning, and 0.36 hours of learning opportunities. That is, teachers
who have more individual autonomy in classroom instruction are less likely to engage in
opportunities for content-focus, and active learning and spending less time on their
learning.

Secondary school level is also signiﬁcantly related to three TOL variables. The
coefﬁcients for secondary school are -.25 / -.10 / -.52 for content-focus, active learning,
and total hours respectively. This means that as the ratio of secondary school to
elementary school increases, teachers report on average experiencing a level of content-
focus 0.25 units and active learning 0.1 units low, and spending 0.49 hours less in their
learning opportunities. These results imply that teachers in secondary school are less
likely to participate in opportunities for content-focus and active learning, and less likely
to spend time on their learning, than those in elementary school. Student total enrollment
is signiﬁcantly related to opportunities for active learning, but the strength of the
coefﬁcient for student total enrollment is too small to assume signiﬁcant relationship
between these two variables.

While these results support the hypothesis that school organizational contextual
factors are associated with TOL, some relationships of organizational factors (such as
supportive leadership, planning time, and student-teacher ratio) to TOL are not precisely

as predicted in the hypotheses stated in this study. Speciﬁcally, this study did not ﬁnd

statistically signiﬁcant relationships of supportive leadership with all three TOL variables.

115

Planning time for instruction and student-teacher ratio also do not have signiﬁcant
relationships with three TOL variables. Theses results provide limited support for the
hypothesis that school organizational contextual factors are associated with TOL.

Consequently, some school organizational variables (such as collective autonomy,
individual control, school level, and student total enrollment) are signiﬁcantly associated
with three variables of TOL, but not all organizational factors proposed in the literature
have a signiﬁcant relationship to TOL. In particular, it is noteworthy that supportive
leadership does not associate with the variability of TOL. Teachers’ workload also does
not account for this between-school variance of TOL. The discussion regarding these

ﬁndings will be provided in the next section.

Institutional contextual factors

This study hypothesized that institutional contextual factors at the school level
are associated with three TOL indicators. In this study, local community circumstances
and diversity of students’ race were regarded as important institutional contexts for TOL.
Community circumstance was measured by one dummy variable indicating suburban and
diversity of students’ race was measured by percent of minority students.

The analysis results show that suburban is signiﬁcantly related to active learning.
The coefﬁcient for suburban is .14 for active learning. This means that as the ratio of
suburban school to other areas (such as urban or rural) increases, teachers report on
average experiencing a level of active learning 0.13 units higher. This implies that
teachers in suburban schools are more likely to participate in opportunities for active

learning than those in other areas such as urban or rural. As hypothesized in this study,

116

one possible interpretation for this result may be that teachers working in suburban
schools are more likely to engage in opportunities for active learning because material
and professional support in suburban communities encourage teachers to engage in
professional development that involves hands-on and inquiry-oriented learning.

The data also indicates that the percent of minority students has a signiﬁcant
relationship with all three TOL variables. The coefﬁcients for percent of minority student
are .002/ .002/. 012 for content-focus, active learning, and total hours respectively, which
mean that as the percentage of minority students increases by 1, content-focus level,
active leanring level, and total hour can be expected to increase by 0.002, 0.002, and
0.012 and correspondingly that as minority percent increase by 10% above the average
minority students, the content-focus level, active learning level, and total hour would be
expected to increase by 0.02, 0.02, and 0.012. However, the magnitude of theses
coefﬁcients is quite small.

Overall, the ﬁndings provide evidence supporting the hypothesis that at the
school level, social system contextual factors, administrative contextual factors, and
institutional contextual factors are associated with TOL, although some relationships of
organizational factors (such as supportive leadership, planning time, and student-teacher

ratio) to TOL are not precisely as predicted in the hypotheses stated in this study.

117

Table 5-3 Intercepts as Outcomes HLM Analysis (I): All Teacher, School, and District
Level Variables and Content-focus as the Dependent Variables

 

 

Content focus TOL
Predictor Coeﬁicients Standard errors
Teacher Level
Female 0167*" 0.044
Teaching experience 0.007" 0.002
Math 0.173 0.090
English 0113* 0.042
School Level
Social system context
Professional learning community 0215*" 0.058
Administrative context
Supportive leadership -0.058 -0.034
Collective autonomy 0.171 *** 0.042
Individual autonomy -0.122** 0.041
Planning time -0.002 0.003
Student-teacher ratio -0.001 0.001
Student total enrolhnent 0.001 0.001
Secondary school -0.248*** -0.032
Institutional context
Suburban 0.053 0.033
% minority 0.002'Ml 0.001
District Level
Intercept 1658*" 0.054
Administrative context
Time support I (Release) 0.076 0.041
Time support H (Scheduled) 0.110" 0.039
Money support I (Stipend) 0.151 *** 0.036
Money support H (Reimburse) 0.038 0.035
Co-funding 0.013 0.012
Principal learning 0.095 0.058
Performance indicator -0.004 -0.094
Accountability -0.03 1 -0.044
Teachers’ responsibility for PD 0004 -0.027
District total enrolhnent 0.001 0.001
Institutional context
% free lunch -0.003 0.001
Variance Components
Within schools 0.67 1*"
Between schools 0048*“
Between districts 0194*"

 

* indicates signiﬁcance at p S .05, " indicates signiﬁcance at p S .01, "* indicates signiﬁcance at pg .001
Note. Numbers in parentheses are standards errors

118

Table 5-4 Intercepts as Outcomes HLM Analysis (H): All Teacher, School, and District
Level Variables and Active learning as the Dependent Variables

Active learning

 

Predictor Coeﬂicients Standard errors
Teacher Level
Female 0074* 0.033
Teaching experience 0.005" 0.001
Math -0.106* 0.044
English 0.033 0.052
School Level
Social system context
Professional learning community 0321*" 0.081
Administrative context
Supportive leadership -0.024 0.043
Collective autonomy 0245*" 0.062
Individual autonomy -0.138** 0.049
Planning time -0.004 0.005
Student-teacher ratio 0.001 0.001
Student total enrollment 0.001*** 0.001
Secondary school -0.103** 0.033
Institutional context
Suburban 0.139" 0.039
% minority 0.002“ 0.001
District Level
Intercept 1833*" 0.049
Administrative context
Time support I (Release) 0.062 0.047
Time support H (Scheduled) 0.133“ 0.051
Money support I (Stipend) 0.061 0.044
Money support H (Reimburse) 0.129" 0.042
Co-funding 0.031 0.025
Principal learning 0.189* 0.075
Performance indicator 0.022 0.074
Accountability -0.012* 0.054
Teachers’ responsibility for PD -0.024 0.022
District total enrollment 0.001 0.001
Institutional context
% ﬂee lunch -0.001 (.001) 0.001
Variance Components
Within schools 0.901 *"‘*
Between schools 0068*"
Between districts 0254*"

 

* indicates signiﬁcance at p S .05, " indicates signiﬁcance at p S .01, *" indicates signiﬁcance at pg .001

Note. Numbers in parentheses are standards errors

119

Table 5-5 Intercepts as Outcomes HLM Analysis (HI): All Teacher, School, and District
Level Variables and Total hours as the Dependent Variables

Total hours

 

Predictor C oqjicients Standard errors
Teacher Level
Female 0.576" 0.203
Teaching experience 0.024” 0.008
Math 0.384 0.417
English 0.192 0.176
School Level
Social system context
Professional learning community 1070*" 0.237
Administrative context
Supportive leadership -0.187 0.154
Collective autonomy 0685*" 0.169
Individual autonomy -0.356* 0. l 82
Planning time -0.021 0.017
Student-teacher ratio 0.001 0.004
Student total enrollment 0.001 0.001
Secondary school -0.515*** 0.140
Institutional context
Suburban 0.148 0.139
% minority 0012*" 0.003
District Level
Intercept 7.097*** ' 0.067
Administrative context
Time support I (Release) -0.003 0.147
Time support H (Scheduled) 0.197 0.160
Money support I (Stipend) 0.761 *** 0.138
Money support H (Reimburse) 0.238 0.157
Co-funding 0.026 0.048
Principal learning 0.749" 0.237
Performance indicator 0492* 0.243
Accountability related to performance -0.099 0.153
Teachers’ responsibility for PD -0.055 0.061
District total enrollment 0.001 0.001
Institutional context
% ﬂee lunch 0.002 0.004
Variance Components
Within schools 13.607***
Between schools 1007*"
Between districts 2963*"

 

* indicates signiﬁcance at p S .05, ** indicates signiﬁcance at p S .01, **"‘ indicates signiﬁcance at p5 .001
Note. Numbers in parentheses are standards errors

120

Result 3: Results of the Third Hypothesis Test
This study assumed that at the district level, administrative contextual factors

(time resources, money resources, professional development for principals,
decentralization of PD, coordination of multiple funding sources, districts’ efforts for
continuous improvement, district size), and institutional contextual factors (parents’
socio-economic condition) are associated with three TOL indicators. The results ﬂom ‘an
intercepts as outcomes HLM’ conﬁrm the third hypothesis as well. To aid a model
interpretation, all variables at the teacher, school, and district level are grand mean
centered. When the explanatory variables are grand mean centered, the mean of each
variable is subtracted ﬂom each teacher’s raw score to rescale each variable to have mean
of 0 (Hox, 2002). The results show that the variability of TOL was associated with
district level administrative factors, and district level institutional factors.

Table 5-3, 5-4, and 5-5 provide the HLM estimates of coefﬁcients for the district-
level contextual factors related to the three variables of TOL. The results of the third

hypothesis test are here represented in terms of administrative and institutional contexts.

District level administrative factors

In this study, district level administrative factors were measured by the following
variables. First, resources for professional development were measured by four variables
(time support I, time support H, money support I, money support H), indicating whether

teachers receive time or money support for their professional development.

 

Time support I: Whether teachers receive release time for teaching (i.e., regular
teaching responsibilities were temporarily assigned to someone else)

 

121

 

Time support II: Whether teachers receive scheduled time in the contract year for
professional development

 

Money support I: Whether teachers receive stipends for professional development

 

Money support II: Whether teachers receive reimbursement of college tuition,
conferences, or travel for professional development

 

Second, districts’ coordination of multiple funding sources was measured by co-
funding, which indicates the extent to which districts combine multi-ftmding sources for

professional development.

 

Co—funding (Sum of seven items): Whether districts use (1) state professional
development funds, (2) special project budgets, (3) school improvement ﬁrnds, (4) Title
I, (5) Eisenhower program, (6) other federal programs, and (7) private sector grants for
TOL.

 

Third, districts’ provision of professional development for principals was
measured by the principal learning variable, indicating the extent to which districts

provide professional development for principals involving instructional matter.

 

Principal learning (sum of six items): Whether districts provide (1) evaluation and
supervision, (2) technology for planning, (3) decision-making and reporting, (4)
curriculum and assessment, (5) networking opportunities, and (6) serving as mentors

 

Fourth, districts’ effort for continuous improvements was measured by two
variables, with one indicating whether the district uses performance indicators to evaluate
the progress of students, and another indicating whether the district uses rewards or

sanctions related to performance.

 

Accountability: Whether the district uses rewards or sanctions related to performance

 

Performance indicator: Whether the district uses performance indicators to evaluate
the progress of students

 

Fifth, districts’ decentralization of professional development was measured by

teachers’ responsibility for PD, indicating the extent to which teachers have primary

122

responsibilities for the in-service professional development activities.

 

Teachers’ responsibility for PD (Sum of three items): Whether teachers have primary
responsibility for professional development in (l) deciding the content, (2) designing
and planning the activities, and (3) conducting the activities.

 

Sixth, district size was measured by district total enrollment.

The results of this study show that most district administrative variables proposed
in this study are signiﬁcantly related to three TOL variables. Among ten variables, six
variables were found to be signiﬁcantly related to TOL variables.

Speciﬁcally, time support H is signiﬁcantly related to content-focus and active
learning variables. The coefﬁcients for time support H (scheduled time) are .11 / .13 for
content-focus and active learning respectively. This means that as the ratio of teachers
receiving scheduled time to those who are not receiving increases, teachers report on
average experiencing a level of content-focus 0.11 units, and active learning 0.13 units
high. These imply that teachers who receive scheduled time for professional development,
are more likely to participate in opportunities for content-focus, and active learning than
those who do not receive scheduled time.

Money support I (stipend) is also signiﬁcantly related to content-focus and total
hours variable. The coefﬁcients for money support I (stipend) are .15 / .76 for content-
focus and total hours respectively. This means that as the ratio of teachers receiving
stipends to those who do not receive increases, teachers report on average experiencing a
level of content-focus 0.15 units high and spending 0.76 hours more on their learning.
This implies that teachers who receive stipends for professional development are more
likely to engage in opportunities for content-focus, and spend more time on their learning.

Money support H (reimbursement) has a signiﬁcant relationship only to active

123

learning variable. The coefﬁcient for money support H is .13 for active learning. This
means that as the ratio of teachers receiving reimbursement to those who do not receive
increases, teachers report on average experiencing a level of active learning 0.13 units
high. This implies that teachers who receive reimbursement of college tuition,
conferences, or travel for professional development are more likely to engage in
opportunities for active learning than those who do not receive.

Principal leaming is signiﬁcantly related to active learning and total hours. The
coefﬁcients for principal learning are .19 / .75 for active learning and total hours
respectively. This means that as the principal learning variable increases by one level,
active learning level and total hours can be expected to increase by 0.19 units, 0.75 hours.
That is, teachers in the district that provides more professional development for principals
are more likely to engage in opportunities for active learning, and spending more time on
teachers’ learning.

Performance indicator is signiﬁcantly related to total hours. The coefﬁcient for
performance indicator is 0.49 for total hours, which means that as the ratio of districts
that use performance indicators to districts that do not use increases, teachers in a district
report on average spending 0.49 hours more on their learning. This result implies that
teachers in the districts that use performance indicator are more likely to spend more time
on their learning than those in the districts that do not use performance indicator.

Accountability related to performance has a signiﬁcant and negative relationship
to active learning. The coefﬁcient for accountability is - 0.12 for active learning, which
means that as the ratio of districts that use rewards or sanctions related to performance to

districts that do not increase, teachers report on average experiencing a level of active

124

learning 0.12 units lower. This result implies that teachers in the districts that use rewards
or sanctions related to performance are less likely to participate in active learning than
those in the districts that do not use rewards or sanctions.

Consequently, these results provide sufﬁcient evidence for supporting the
hypothesis that administrative contextual factors are associated with TOL at the district
level, although this study did not ﬁnd statistically signiﬁcant relationships of some
administrative variables with three TOL variables. The discussion regarding these

ﬁndings will be provided in the next section.

Institutional contactual factors

This study hypothesized that parents’ socio-econornic condition is a critical
institutional contextual factor for three TOL indicators. Parents’ socio-economic
condition as a factor was measured by percentage of ﬂee lunch at the district level. The
data shows that percent of ﬂee lunch students is not signiﬁcantly related to any variable
of TOL although this variable is negatively related to all three variables. Thus, this
ﬁnding does not provide evidence that supports for the hypothesis that at the district level,
institutional contextual factors are associated with TOL indicators. That is, the
relationships of the percent of ﬂee-lunch students to three TOL variables are not
precisely as predicted in the hypotheses. The discussion regarding this ﬁnding will be

provided in the next section.

Result 4: Results of the Fourth Hypothesis Test

The fourth hypothesis for this study was that interactive relationships between

125

 

professional learning community and districts administrative contextual factors affect
three TOL indicators. This study limited the analysis to the cross interactions between
district level administrative factors and school level social system conditions for the
reason that a full explication of all possible interactions is beyond the scope of this study.
A slopes-as-outcomes HLM provides useful information about the effects of cross-level
interactive relationships between professional learning community and districts level
factors on three TOL indicators. This model estimates the degree to which district level
administrative variables inﬂuence the strength of the relationship between school level

professional learning community and TOL (i.e., a cross-level interaction), holding

constant all of the variables included in the prior model (an intercepts-as-outcomes HLM).

The results ﬂom the HLM analysis show that some district administrative variables have
cross interaction effects with professional learning community on TOL variables.

First, this study found that accountability variable has statistically signiﬁcant
interaction effects with professional learning community and main effects on active
learning. Table 5-6 presents the estimates ﬂom the model with this cross-level interaction
between accountability and professional learning community.

Professional learning community still predicts high level of active learning, with
increased coefﬁcient ﬂom 0.32 of the previous model to 0.36 of the current model. This
means that as the professional learning community level increases by one unit, active
learning is expected to increase by 0.36 units. Accountability variable is also signiﬁcantly
related to active learning, with a much more decreased coefﬁcient ﬂom - 0.12 of the
previous model to -.401 of the current model. This means that as the ratio of districts that

use rewards or sanctions related to performance to districts that do not, teachers report on

126

thus-mun-” __ a

average experiencing a level of active learning 0.40 units lower.
The regression coefﬁcient for the cross-level interaction is -.401, which is very
big and signiﬁcant. This interaction is formed by multiplying the levels for the variables
‘professional learning community’ and ‘accountability,’ and the negative value of
interaction term (-.401) means that in districts that use rewards or sanctions related to
performance, the advantage that teaches working in the professional learning community
are more likely to participate in active learning, is smaller than expected ﬂom the direct 5

effects of professional learning community only. Thus, the effects of professional

[1147' U '

learning community on teachers’ opportunities for active learning are considerably
smaller in districts that uses rewards or sanctions related to performance.

Table 5-6 Cross-Level Interactive Effects between District-Level Accountability and
School- Level Professional Learning Community on Active Learning

 

Model for active learning outcome variable,
Selected Term

 

Intercept 1.833 (.049)***

Learning community (G010) .356 (.073)***
Interaction with accountability (001 l) -.401 (.169)*

Accountability (G001) -.106 (.048)*

 

Note. Numbers in parentheses are standards errors.
"‘ indicates signiﬁcance at p S .05,
*** indicates signiﬁcance at p 5.001

Another district level variable, performance indicator also has statistically
signiﬁcant interaction effects and main effects on total hours. Table 5-7 presents the
estimates ﬂom the model with this cross-level interaction between performance indicator
and professional learning community.

Professional learning community still predicts high level of total hour, with

increased coefﬁcient from 1.07 of the previous model to 1.10 of the current model. This

127

 

means that as the professional learning community level increases by one unit, total hours
of TOL are expected to increase by 1.07 hours. Performance indicator is also
signiﬁcantly related to total hours, with positive coefﬁcient of 0.48. This means that as
the ratio of districts that use performance indicators to districts that do not use, teachers
report on average spending 0.48 hours more on their learning.

The regression coefﬁcient for the cross-level interaction is 1.65, which is big
and signiﬁcant. This interaction is formed by multiplying the scores for the variables
‘professional learning community’ and ‘performance indicator,’ and the positive value of
interaction term (1.65) means that in the districts with district performance indicator, the
advantage that teachers working in the professional learning community spend more time
on their leaming is bigger than expected from the direct effects of professional learning
community only. Thus, the relationship of professional learning community to total hours
is stronger in districts that use performance indicators.

Table 5-7 Cross-Level Interactive Effects between District Level Performance Indicator
and School Level Professional Learning Community on Total Hours

 

Model for total hours outcome variable,
Selected Term

 

Intercept 7.09 (0.07)***

Professional learning community (G010) 1.10 (0.24)***
Interaction with district performance indicators (G011) 1.65 (0.61)"

Performance indicators (G001) 0.48 (0.20)"‘

 

Note. Numbers in parentheses are standards errors
* indicates signiﬁcance at pS .05,

** indicates signiﬁcance at p501,

*** indicates signiﬁcance at p S .001

In sum, through the cross-level interaction, this study found that the strength of
this relationship varies across different types of district policies, such as accountability

related to performance or performance indicator. For example, professional learning

128

community is a stronger predictor of three TOL variables when districts use performance
indicators for school improvement. In contrast, districts’ use of rewards or sanctions
related to performance results in professional learning communities having a weaker
relationship with active learning. Consequently, the results of this study support the
hypothesis that interactive relationships between professional learning community and
districts administrative contextual factors affect three TOL indicators. The discussion

regarding the important ﬁndings will be provided in detail in the discussion section.

Result 5: Results of the Fifth Hypothesis Test
The ﬁfth hypothesis for this study was that the relationships between contextual

conditions and TOL vary across the three TOL indicators. The results of this hypothesis
are provided by synthesizing the results of the previous hypotheses 2 and 3 rather than by
performing a particular HLM analysis. The analysis conﬁrms that some relationships
between district or school level contextual conditions and TOL vary across the three TOL
indicators. Table 5-8 presents coefﬁcients and signiﬁcance for the school and district
variables, which indicate the different relationship across three TOL variables (content-
focus, active learning, and total hours).

First, among school level variables, student total enrollment and suburban are
only signiﬁcantly related to active learning. The signiﬁcant coefﬁcient for suburban is .14
for active learning, which means that as the ratio of suburban to other areas increases,
teachers report on average experiencing a level of active learning 0.14 units higher.
Student total enrollment is only signiﬁcantly related to opportunities for active learning,

but the strength of the coefﬁcient is so small.

129

Second, among district level administrative factors, time or money resources for
TOL have diverse relationships with three TOL variables. For example, time support H is
signiﬁcantly related to content-focus and active learning variables. The coefﬁcients for
time support H (scheduled time) are .11 / .13 for content-focus and active learning
respectively, which means that as the ratio of teachers receiving scheduled time to those
who are not receiving increases, teachers report on average experiencing a level of
content-focus 0.11 units, and active leaming 0.13 units higher. Money support I is
signiﬁcantly related to content-focus and total hours. The coefﬁcients for money support
I (stipend) are .15 / .76 for content-focus and total hours respectively, which means that
as the ratio of teachers receiving stipends to those who are not receiving increases,
teachers report on average experiencing a level of content-focus 0.15 units higher and
spending a 0.76 hours more on their learning. In addition, money support H has only
signiﬁcant relationship to active learning variable. The coefﬁcient for money support H
(reimbursement) is .13 for active learning, which means that as the ratio of teachers
receiving reimbursement to those who are not receiving increases, teachers report on
average experiencing a level of active learning 0.13 units higher.

Third, among district level administrative variables, principal learning,
accountability, performance indicators, and co-funding are only signiﬁcantly related to
active learning and total hours. For example, principal learning variable is signiﬁcantly
related to total hours and active learning variable. The coefficients for principal learning
for principal learning are .19 / .75 for active learning and total hours respectively, which
means that as the principal learning increases by one level, active learning level and total

hours can be expected to increase by 0.19 units, 0.75 hours. Performance indicator is

130

signiﬁcantly related to total hours TOL variable. The coefﬁcient for performance
indicator is 0.49 for total hours, which means that as the ratio of districts that use
performance indicators to districts that do not use increases, teachers in a district report
on average spending 0.49 hours more on their learning. By contrast, accountability,
money support H (reimbursement), and co-funding are only signiﬁcantly related to active
learning, with regression coefﬁcient at -0.12, 0.13, and 0.03 respectively, which means
that as the ratio of districts that use rewards or sanctions related to performance to

districts that do not increase, teachers report on average experiencing a level of active

learning 0.12 units lower.

Table 5-8 Different Relationships with Contextual Factors across Three TOL Variables

 

 

Content-focus Active learning Total hours
School level
Student total enrollment .000 (.000) .001 (.001)*** .000 (.000)
Suburban .053 (.033) .139 (.039)** .148 (.139)
District level
Time support H (Scheduled) .110 (.039)" .133 (.051)‘ .197(.l60)
Money support I (Stipend) .151 (.036)*** .06l(.044) .761(.138)***
Money support H (Reimburse) .038 (.035) .129(.042)** .238(.157)
Co-funding .013 (.012) .031 (.015) .026 (.048)
Principal learning .095 (.058) .189 (.075)‘ .749 (.237)**
Performance indicator -.004 (.094) .022 (.074) .492 (.243)*
Accountability -.031 (.044) -.117 (.054)* -.099 (.153)

 

Consequently, although not all contextual factors are related to these variations
across three variables of TOL, the results of this study have sufﬁcient evidence
SUpporting the hypothesis that the relationship between contextual conditions and TOL
Varies across the three indicators of TOL (content-focus, active learning, and total hours).

In the discussion section, this study will discuss these important ﬁndings more in detail.

131

Summary Remarks

From the results of the three separate HLM analyses, this study found mixed
evidence for the ﬁve hypotheses regarding the relationship between district and school
contexts and TOL. First, this study found evidence supporting the ﬁrst hypothesis that
three TOL indicators vary across schools and districts. As proposed in this study, a
substantial proportion of the variation in three TOL indicators is attributed to district-and
school-level contexts.

Second, not all of the stated contextual factors were confumed in their
relationship with TOL. But the results support the notion of multiple perspectives and of
embedded contexts as affecting TOL, by providing sufﬁcient evidence that multiple
contexts conditions (such as social system contexts, administrative and organizational
contexts, and institutional contexts) are signiﬁcantly associated with three TOL indicators
at the district or school level.

Furthermore, although this study did not explore all possible interactions among
contextual factors, the results conﬁrm the notion of embedded contexts that assume
interactive relationships among contexts conditions, by providing sufﬁcient evidence that
districts’ administrative contexts conditions moderate the relationship between schools’
social system contexts conditions and TOL. The results also conﬁrm the ﬁfth hypothesis
that the relationship between contextual conditions and TOL varies across the three

indicators of TOL (content-focus, active learning, and total hours).

132

Discussion

The ﬁrst purpose of this discussion is to suggest some possible interpretations for
the results of this study. Although the primary focus of this study was to test the utility of
the conceptual model, there are some implications for policy and practice that may be
addressed ﬂom the ﬁndings.

This section presents three topics: First, a discussion regarding district and
school-level variability of TOL is presented based on signiﬁcant ﬁndings ﬂom the ﬁrst
hypothesis test. Then, a discussion about contextual factors related to TOL is presented
on the basis of the important ﬁndings ﬂom the second, third, and fourth hypotheses tests.
Finally, a discussion regarding diverse inﬂuences of contexts across TOL indicators is

presented based on the ﬁndings ﬂom the ﬁfth hypothesis.

Discussion 1: Relationship between Multilevel School Contexts and TOL

This study hypothesized that teachers’ opportunities to learn vary across schools
and districts. If very little or none of the variance was the result of between-school or
between-district differences, then there was no practical reason to study multilevel school
contexts for TOL. The results of this study conﬁrmed that a substantial proportion of the
variation in three TOL indicators is attributed to district-and school-level contexts
although much of this is attributable to differences in teacher background characteristics
within schools. Speciﬁcally, approximately 20% of the variance in three TOL indicators
was explainable by between-district differences and approximately 6% of the variance in
three TOL indicators was explainable by between-school differences.

Several aspects of these ﬁndings are noteworthy. First, although a substantial

133

 

percentage of the variance is found at the district-and school level, almost 70% of the
variance is still at the teacher level. If we assume that TOL is a multi-level contextual
construct, then the evidence that most of the variance is found at the teacher level raises
an issue regarding the precise estimation of variance.

One possible explanation for these ﬁndings is that small within-school samples
of the SASS data may make it difﬁcult to pick up the variation between schools. The
SASS data has on average of only ﬁve or six teachers per school. Smith and Rowely’s
study (2005) using the 2000 SASS also show the similar results in the variation of TOL.
In this study, they found that only 4.1% of the variance in total hours of TOL was
explained by school level factors (including administrative support, collaboration,
autonomy, and other structural conditions) and 2.7% of the variance in total hours was
explained by state level factors (including teacher certiﬁcation and accountability),
leaving 93% of the variance in total hours within the schools. A new national survey data
designed more carefully within a nested sample may enable us to increase the precision
of variance estimates in teacher outcome variables within schools (Talbert & McLaughlin,
1994)

Another possible explanation is that TOL would be also constructed through
subgroups within a school. For example, TOL tend to inhere in the grade level or the area
of subject speciﬁcation. The ﬁeld observations conducted by Little (1999b) and
McLaughlin and Talbert (2001) show that different TOL patterns within and across
different subject departments in the same schools. For this reason, I have reason to expect
high within-school variation in TOL.

Second, the ﬁndings of this study indicate that district contexts matter for TOL as

134

much as school contexts. In this study, differences among districts apparently contribute
more to the variability of TOL than school differences. These ﬁndings are interesting and
surprising because many educational reformers and researchers have generally argued
that schools rather than districts can make differences in teacher learning. Yet, the
ﬁndings of this study suggest the importance Of the district relative tO the school as a
primary agent of teachers’ professional development.

Some critics may contend that high between- district variance may occur because

I
rPl_

I'm-I“

the SASS data is more likely tO represent formal learning rather than informal learning.

That is, since teachers’ informal learning tends to occur spontaneously (Smylie, 1995),
large survey SASS data has limitations in capturing the aspects of informal learning
Opportunities and thus the data may include more components Of formal learning
provided at the district level than informal learning. This may be a possible explanation
for the high between-district variance in TOL because districts are generally regarded as
dominant providers of formal learning (Little, 1990a). Further, it should be noted that a
some of previous research (e.g., Desimone et al., 2002; Firestone et al., 2004; Spillane,
2004) is consistent with these ﬁndings. Although I ﬁnd little research that simultaneously
explores between — school and district variance in TOL, several studies suggest that
districts can play a facilitating role in increasing formal or informal teachers’ learning.
For example, Desimone et al.’s (2002) study using a nationally representative sample Of
district coordinators found that district policy mechanisms guide and support high quality
TOL. In particular, their study takes comprehensive indicators Of TOL, which include not
only three features used in the current study (such as content-focus, active learning

opportunities, duration), but also collective participation, coherence, and type Of activity.

135

Further, these indicators Of TOL include the component of informal as well as formal
learning.

Third, this study found that by simultaneously estimating the effects of two
embedded contexts, each one has a signiﬁcant relationship with TOL. These ﬁndings
challenge the results Of prior research that report the effects Of district and school
contexts as mutually exclusive. Findings ﬂom this study suggest that the variability of
three TOL indicators may be attributable to multi-level school contexts. Although this
study is restricted to two levels due to the data available in the SASS survey, it supports
the assumption that each Of the multiple embedded contexts (including states or subject

departments) may affect TOL.

Discussion 2: Relationship between Contextual Factors and TOL

This study assumed that multiple contexts (social system contexts, administrative
contexts, and institutional contexts) at the district and school level are associated with
TOL. Results from the HLM analyses conﬁrmed that most of the proposed contextual
factors are signiﬁcantly associated with three TOL indicators (content-focus, active
learning, and total hours). Further, regarding a range of speciﬁc contextual factors
assumed to inﬂuence TOL, there was a lot of complexity, under-speciﬁcation, and
interactive effects.

This section provides a discussion regarding the complex and interactive
relationships between contextual factors and three TOL indicators. Because contextual
factors are often interrelated with one another, synthesizing these factors with common

themes can provide a better interpretation Of the results and implications for policy and

136

 

practices (Knapp, 2003). Here, this study discusses the relationships between contextual
factors and TOL using six themes.

Table 5-2 illustrates how contextual factors proposed in this study are related to
six themes. Each theme is elaborated below.

Figure 5-2 Contextual Factors Proposed in this Study and Six Themes

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Contextual factors proposed Themes for discussion
Professional learning community I:> Professional learning community
Supportive leadership . . .

, , Prrncrpals’ leadership
PD for prrncrpals
Collective autonomy
Individual autonomy Teacher autonomy
Decentralization of PD
Accountability related to
performance Accountability strategy

Performance indicator

 

 

 

 

Time support

Money support

Coordinating multiple funding
sources

Time and ﬁnancial resources

 

 

 

 

 

 

 

 

Local community circumstances

Parents’ sociO-economic condition Institutional environments

ll 11 ll 0 ll

 

 

 

Diversity Of students’ race

 

 

 

137

mar—1"

an n-uun...

 

Professional learning community

The results of this study show that professional learning community is a critical
context for TOL. Speciﬁcally, this study found that teachers working in the professional
learning community are more likely to engage in learning opportunities for content-focus
and active learning, and are more likely to devote their time tO learning. These ﬁndings
support the notion that building and nurturing professional learning community where
teachers engage socially with colleagues and share the sources Of knowledge can
contribute to increasing TOL. In particular, the professional learning communities should
be connected to the learning or instructional matter rather than to simply helping teachers
feel good about each other or their working situation.

These ﬁndings are consistent with ﬁeld Observations Offered in Little (1999b)
and McLaughlin and Talbert (2001) that underscore the importance of professional
leaming community in teacher learning and teaching practices. For example, McLaughlin
and Talbert (2001) found that strong learning communities that value innovation for
instructional practices and teachers’ collaboration for learning are more likely to support
teachers’ on- going learning. By contrast, strong traditional communities that value
conservatism about learning and instruction isolate teachers ﬂom peers and discourage
teachers’ learning. Although the current study did not include the relationships Of other
kinds of professional communities (such as traditional communities) to TOL, it provides
empirical conﬁrmation on consensus assumptions about the effects Of professional
learning community on TOL.

The results Of this study also extend the prior literature on professional learning

community, by ﬁnding the interactive relationships between professional learning

138

 

community and district level administrative contextual factors. That is, this study found
that the effects Of professional learning community on TOL are stronger or weaker
depending on different types of districts’ accountability strategy. In the current study,
teachers working in the professional learning communities tend to spend more time on
their professional development when districts adopt performance indicators for school
improvement. Knapp (2003) explains by noting that professional communities that
encourage teachers’ leaming can be established or strengthened through district policies.
For instance, the collegial relations and the value of inquiry in professional communities
would be strengthened if districts mandate tasks that are to be performed collaboratively,
such as planning for school improvement or establishing goal setting (Knapp, 2003).
While participating in these tasks, teachers may spend time together, and learn ﬂom each
other.

By contrast, the results Of this study show the effects of professional learning
communities on TOL can be weakened through district strategies. In the current study,
teachers in the professional learning communities are less likely to engage in active
learning when districts adopt rewards or sanctions related to performance. District
rewards or sanctions related performance may undercut the general value Of collegiality
and the speciﬁc commitment to an improvement agenda by isolating teachers ﬂom peers,
or inhibiting teachers’ attempts to undertake innovative practices and learning (Firestone
et al., 2004).

Consequently, it is certain that professional learning communities contribute
substantively to TOL. When teachers collaborate to design new instructional practices

and have shared norms concerning learning, teachers are more likely to engage in

139

 

content-focused and active learning. They are also more likely to devote time to their
learning. In particular, professional communities can be effective on TOL when they are
strongly connected tO the learning or instructional matters. Furthermore, these
professional communities have interactive relationships with district level administrative
contextual factors. For example, professional learning communities would lead to
increased TOL if district strategies convey norms and expectations regarding teacher

learning and teaching practices. By contrast, the effects Of professional leaming

“r

communities on TOL would decrease if district strategies destroy the value of collegial

‘F'.’ .

norms and innovation.

Principals ’ leadership

The conventional wisdom is that the principal matters for teachers’ learning. In
some parts, the results of this study support these arguments. For example, this study
found that in districts where principals participate in professional development program,
teachers are more likely to engage in active leaming and devote their time for
professional development.

This is consistent with several studies (Rosenholtz, 1989; Fink & Resnick, 2001;
Firestone et al., 2005) that report the importance of principals’ professional development

for teachers’ learning. For instance, Fink and Resnick (2001) found that in districts that

 

Offer professional development programs for principals (such as visiting other schools,
classroom Observations, discussion involving curriculum and pedagogical repertories, or
learning from exemplary instructional practices), principals are more likely to develop

their understanding of high quality instruction and provide practical advice for

140

overcoming Obstacles to realizing reform goals and performing new teaching practices. In
particular, these programs for principals are effective for teacher learning when the
programs are directly related to instructional matters (Rosenholtz, 1989). That is,
principals’ instructional capacity developed through professional development programs
focused on instructional matters can improve TOL. This is a critical linkage that is Often
overlooked in conventional professional development programs.

Despite the importance of principals in the enhancement Of teacher learning, this
study did not ﬁnd signiﬁcant relationships between supportive leadership Of principals
and increased TOL. This puzzle turns up in several other studies which contend that
supportive principals’ leadership does not directly lead to increased TOL although it
encourages collaborative relations among teachers (Firestone et al., 2005; Little, 1999b).
Rather, teachers are more likely to experience their learning cumulatively and
substantively when they have regular contact with principals who have the requisite
expertise to diagnose and exhibit productive instructional practices, guide achievement Of
school goals, and give teachers feedback on classroom practices (Knapp, 2003; Youngs
& King, 2002). “Supportive leadership” as measured in this study, is apparently not
connected tightly and speciﬁcally enough to teacher learning or instructional matters.

Consequently, the ﬁndings of this study suggest that the speciﬁc aspects Of
principal leadership that promote TOL are elusive to pin down. Although we speculate
that principals’ instructional capacity developed through professional development
provided by districts tends to increase TOL, the principals’ leadership affecting TOL is
still left unspeciﬁed and under-explored. More information is needed on the aspects Of

principals’ leadership available to improve effective TOL.

141

 

 

Teacher autonomy

Although common wisdom posits that teacher autonomy leads to increased TOL,
the results Of this study indicate a complex relationship between teacher autonomy and
TOL. For instance, this study found that teachers’ collective autonomy predictor increase
in three TOL indicators (content-focus, active learning, and total hours) while individual
autonomy leads to decrease in three TOL indicators. Speciﬁcally, when teachers exercise
a great deal of collective inﬂuence on school policy and instruction, teachers tend to

engage in content-focus and active learning and spend more time on their learning. These

 

ﬁndings are consistent with the contention Of several scholars (Little, 1990a; Marks &
Louis, 1999; Newmann, 1993; Smlylie, 1994) that collective inﬂuence of teachers on
making decisions affects professional development and instructional practices. For
example, Smylie (1994) argues that teachers are more likely to examine their current
assumptions and practices and increase their responsibilities for professional
enhancement while experiencing collective autonomy.

By contrast, the ﬁndings Of this study show that teachers’ individual autonomy
over classroom processes decreases effective TOL. This ﬁnding can be explained by
Marks and Louis’s (1999) contention that increasing the individual autonomy of teachers
who then exercise their ﬂeedom in disparate directions may undermine a uniﬁed focus on
curriculum and pedagogy and support teachers' isolation ﬂom peers. Teachers’ isolation

from peers has a gradually dangerous effect on teacher learning by creating invisible

 

walls among teachers and decreasing the beneﬁcial role that activities such as
collaboration can have in their practices (Scribner, 1999).

Interestingly, despite the effects Of teachers’ collective or individual autonomy

142

on TOL, this study did not ﬁnd signiﬁcant relationships between districts’
decentralization Of professional development and TOL. Marks and Louis (1999) explain
this ﬁnding by noting that once decentralization provides the school with autonomy, both
capacity for organizational learning and empowerment are largely a matter of internal
relationships among people. In other words, school autonomy that leads to increased TOL
is more closely related to school culture rather than district policies that provide a great
deal of decentralization.

Again, the district emerges as an important agent Of TOL, and this makes sense.

1rm2‘riiﬁw ‘—

When districts play a minor role in PD (i.e., via “decentralization” of this function), then
TOL is decreased as well. This is an important ﬁnding if it holds up as a central factor. At
the same time, the ﬁndings Of this study suggest that efforts tO empower teachers can
contribute to increased capacities for teacher leaming on the part of the school. Teachers
are willing to be committed to their learning in the learning environment focused on
teachers’ collective autonomy while their learning opportunities decrease in the school
culture focused on individual autonomy. Furthermore, we speculate, district engagement
in PD can enhance teacher inﬂuence by bringing them out of isolated settings to engage
in common learning. This combination of district direction with teacher collective
engagement appears to be a crucial aspect of enhanced TOL. Still, based just on these

data, this is a hypothesis that requires further study.

Accountability strategies
Some scholars presume that strong accountability strategies lead tO increased

TOL, but the empirical ﬁndings using the national data do not necessarily support this

 

143

proposition. The results of this study show that districts’ use Of performance indicators
predict teachers’ more extended professional development while district rewards or
sanctions related to performance lead tO decreased teachers’ active leaming. The positive
relationship between districts’ use of performance indicators and TOL would be
explained by the contention Of several scholars (i.e., Desimone etal., 2001; Knapp, 2003;
Murphy & Hallinger, 1988; Rosenholtz, 1989) who emphasize districts’ role in improving ,
teachers’ learning or practices. For example, Knapp (2001) argues that districts’ use of

performance indicators can lead tO increased professionalism within schools, through

 

conveying districts’ messages about general norms and expectations. These
accountability strategies also provide useful information on weaknesses that need to be
addressed (Desimone et al., 2001).

By contrast, the negative effects Of rewards or sanctions related to performance
could be explained by critics Of accountability systems (i.e., Firestone et al., 2004)
arguing that strong accountability polices can have perverse effects on teacher outcomes.
Firestone et al. (2004) note that the accountability-related performance, ranging ﬂom
extra pay for increased student performance to school closure for chronic
underperformance, leads to decreased TOL, by creating an environment which inhibits
teacher risk taking and experimentation, and isolates teachers ﬂom peers.

Consequently, the ﬁndings of this study suggest that, at least with regard to TOL,
districts’ accountability strategies do not necessarily lead to increased TOL.
Accountability strategies generally assume that if teachers are evaluated based on teacher
and student outcome, they would participate in professional development to make better

progress. But these accountability strategies would lead to decreased teachers’ aspiration

144

tO inquiry-oriented leaming opportunities if teachers feel pressured to raise test scores
due to accountability policy. Rather, these accountability strategies would be more
effective if such strategies were to focus more on fostering learning environments that
share and communicate the importance Of learning and school improvement. Further
studies using more developed measurements or in-depth case studies can provide better

information about the effects of accountability strategies on TOL.

Time and Financial Resources

The ﬁndings of this study show that the availability Of time and ﬁnancial support
for professional development signiﬁcantly increase TOL. For example, teachers receiving
scheduled time for professional development are more likely to engage in Opportunities
for content-focus, and active learning. In addition, the availability of stipends or
reimbursement for professional development is more likely tO enable teachers to
participate in Opportunities for content-focus, and active leanring and to devote their time
for professional development.

Findings of this study are consistent with the evidence Of some case studies (i.e.,
David et al., 2000; Scribner, 1999). In a study Of three districts in Kentucky, David et al.
(2000) found that teachers in a district which paid stipends for in-service teacher

professional development were more likely to pursue their professional development
focused on reform ideas. Scribner’s (1999) study shows that teachers are more likely to

conduct highly valued learning activities (such as individual inquiry and collaboration)
and engage in learning Opportunities to meet their needs when the scheduled time for

teacher learning is given. Darling-Hammond and McLaughlin (1995) note that

145

organization of time for teacher learning is a critical factor for facilitating teachers’
professional development because by having adequate time for learning, teachers can
have Opportunities to acquire, practice, and reﬂect on new concepts and skills to
collaborate and interact with peers.

Interestingly, however, the ﬁndings of the current study show that all ﬁnancial
resources or time availability are not necessarily effective for TOL. For example, this
study did not ﬁnd signiﬁcant relationships between released time for professional
development and TOL indicators. In other words, provision Of time for teacher learning
would be effective when the time is Ofﬁcially scheduled for teachers’ professional
development rather than when the time is ﬂeely given to teachers by simply releasing
teaching time ﬂom teachers. These ﬁndings suggest that the effects of time or ﬁnancial
resources on effective TOL need to be investigated cautiously.

Despite the importance of ﬁnancial resources, districts’ efforts to coordinate
multiple funding sources of professional development do not predict increased TOL. The
current study did not ﬁnd signiﬁcant effects of multiple funding sources on increased
TOL. This ﬁnding contrasts with Desimone et al. (2002) who found that districts’
coordination of funding from multiple sources supports coherent TOL activities. One
possible explanation for this would be that while Desimone et al. found the signiﬁcant
effects Of multiple funding sources on structural features Of TOL, the current study did
not include such aspects of T OL as outcome variables. That is, signiﬁcantly related
outcome variables of TOL ﬂom the study Of Desimone et a1. (2002) were reform types Of
TOL and collective participation, but this study focuses on content-focus, active learning,

and total hours of TOL as TOL outcome variables.

146

Another possible explanation for this is the weak connection between districts’
effort to coordinate multiple funding sources and the core features of TOL such as
content-focus or active learning. Because external funding sources are Often ﬂagrnented
into different funding streams and uncoordinated with other programs, over-dependence
on external ﬁrnding sources can make it difﬁcult to plan and provide the core features Of
TOL which generally require substantive and long term investments.

Consequently, it is certain that time availability and ﬁnancial resources for

teachers’ professional development can signiﬁcantly contribute to improving TOL.

However, all provisions Of these resources are not effective on teacher learning.
F urtherrnore, the impact Of ﬁnancial resources does not necessarily rely on multiple

funding sources.

Institutional environments

This study assumed that institutional contexts outside of the educational systems
(such as local community circumstances, parents’ socioeconomic conditions, and
diversity of students’ race) also matter for TOL. The results of this study are mixed. As
hypothesized, in suburban schools where local communities generally have much more
impact on the schools and give abundant material resources, teachers’ Opportunities for
active learning are more likely tO increase. In addition, in schools with a high portion of
minority students, teachers tend to more engage in the Opportunities for content-speciﬁc
and active learning. In these schools, teachers are more likely to devote their time to
professional development as well.

By contrast, the current study did not ﬁnd signiﬁcant relationships between

147

parents’ socioeconomic conditions and TOL at the district level. One possible explanation
of this ﬁnding is that although high poverty districts receive funds ﬂom multiple
programs, these provisions may not solve the problems Of generally low teachers’
engagement in TOL due to institutional contexts such as parents’ socioeconomic
conditions. In some ways, the funding may be used for providing professional
development for teachers. However, the provision of funding for professional
development is not enough to stimulate teachers’ willingness to engage in learning
Opportunities in high-poverty districts. Because low socio-economic status of students
creates teachers’ low expectations of student performance and their own eﬂ'rcacy,
teachers’ psychological discouragement would not be recovered only by providing the
funding for professional development. Furthermore, these funds Often tend tO be
fragmented into different funding streams in poverty districts. Based on these ideas, I
speculate that multiple funding sources and the institutional contexts (such as parents’
socioeconomic conditions) in hi gh-poverty districts may lead to a zero-sum impact on
teacher learning.

Taken as a whole, the evidence ﬂom this analysis suggests that institutional
contexts constitute TOL as well. The national data indicates that teachers’ willingness to
engaging in learning Opportunities is more likely to be attributed to the socioeconomic
conditions and cultures particular to different local communities, student body, and
parents. These ﬁndings are important because up to now, policy makers and policy
researchers had not paid much attention to the inﬂuences of institutional contexts on TOL.
The issue of institutional contexts is one that should be closely examined in ﬁrrther

research.

148

Discussion 3: Variability across Three TOL Indicators

The current study hypothesized that the relationships between contextual
conditions and TOL vary across the three TOL indicators. The ﬁndings of this study show
that some contextual factors have diﬂ’erent effects on three TOL indicators.

First, local community circumstances are more likely to affect teachers’ active
learning than content-focus learning or total hours. These ﬁndings seem to be attributed
to the nature Of active learning. That is, active learning referring to teachers’ active
engagement in intelligent and meaningful discussion, planning, and practices are more
likely to occur in the learning environments that encourage teachers’ new
experimentation and support teachers’ professional work. Generally, suburban schools in
which local communities Offer abundant material resources and professional support are
more likely to support and encourage teachers’ professional experimentation and
professional work.

By contrast, teachers’ active learning is more sensitive to accountability strategies
related to student performance rather than other indicators Of TOL. The ﬁndings Of this
study show that districts’ rewards or sanctions related to student performance are
negatively related to active learning. Strong accountability related to performance may
inhibit teachers’ deep understanding about new instructional approaches or criticism on
the basis Of their experiences, by emphasizing conventional didactic practice or test
preparation.

Second, time availability for professional development is more associated with
the core aspects of TOL than the structural aspects. This study found that scheduled time

for professional development has signiﬁcant relationships with teachers’ content-focus

149

and active learning. These ﬁndings make sense because increasing the core aspects of
TOL (such as content-focus, active learning) requires substantive and long term
investments.

Third, districts’ accountability strategies using the performance indicators seem to
have a relationship with total hours rather than the core aspects of TOL (content-focus
and active learning). The ﬁndings of this study suggest that despite extended time for
professional development, districts’ accountability strategies do not guarantee the quality
of TOL.

Consequently, the ﬁndings Of this study suggest that the indicators Of TOL are
sensitive to level or context effects. In particular districts’ policies that guarantee
extended time for professional development do not necessarily increase teachers’
opportunities for the content-focus and active learning. Thus, policies for improving the

quality Of TOL need to be cautiously developed.

150

H

CHAPTER VI
CONCLUSIONS

Chapter ﬁve presented and discussed the test results Of ﬁve hypotheses proposed
in this study regarding the relationship between district and school contexts and TOL.
Three kinds Of HLM analyses enabled examination Of these hypotheses. This research
was conducted using a SASS national database providing parallel measures Of TOL, and
providing the necessary information regarding school and district level contextual factors.
This ﬁnal chapter provides a brief discussion of the strengths and limitations of this study,

and presents some implications for policy research, and policy and practices.

Strengths and Limitations

This study provides new evidence on the utility of a conceptual model exploring
the relationship between contexts and TOL. Although a more complete understanding of
the relationship between contexts and TOL requires a comprehensive approach that is
capable Of capturing the diverse and interactive contextual inﬂuences on TOL, the
contemporary approaches tend to treat contextual levels of educational systems as
mutually exclusive. By simultaneously estimating the connection between district and
school contexts and TOL, this study has demonstrated the relationship between complex
contexts and TOL. Furthermore, this study also afﬁrms the value of multilevel statistical
techniques as a means to extend McLaughlin and Talbert’s model methodologically.

However, as with most research, this study also has methodological
shortcomings. First, the SASS data used in the current study are cross-sectional. All

information was collected during the 1999-2000 school year. This is problematic for two

151

 

 

 

reasons. First, the relationship between district and school level factors and three TOL
indicators (content-focus, active learning, and total hours of TOL) may be inﬂuenced by
temporally prior learning Opportunities. Second, there may be a time lag among the
relationships under investigation. For instance, districts’ investment for principals’
professional development might not be expected to affect three TOL indicators in the
year in which the investment for principals was begun, but rather in the years after the
investment projects are completed. F’

Therefore, it is not possible to draw inferences about the direction Of causation
for the relationships that were discovered. It may be that high levels of professional
learning communities cause increased TOL, or it may be that having high teachers’
leaming Opportunities create professional learning conununities. Due to these limitations
in causal inferences, the results of this study were carefully qualiﬁed and delimited. For
example, when this study addressed the results Of the relationship between contexts and
three TOL indicators, it limited the language by using ‘related’ or ‘associated’ rather than
‘affect’ or ‘inﬂuence’. TO conﬁrm the causal direction hypothesized in this study,
subsequent research should replicate the results using longitudinal data.

Second, the results Of this study are based on analysis Of secondary data limited to
variables about which information was collected by the National Center for Education
Statistics. Of course, studying the relationship between TOL and contextual factors using
such a large national data set Of teachers, schools, and districts creates a worthwhile
Opportunity. Nevertheless, only a portion of the variance in the reported TOL is accounted
for by the variables examined in each model. Although it is not Obvious ﬂom the

literature review and theory whether important variables are missing from the models, it

152

 

is always possible that variables exist that have not been identiﬁed. If such variables do
exist and are not included in the models this can lead to speciﬁcation error or biased
coefﬁcients and potentially misleading statements. Due to these limitations, this study
addressed the results cautiously.

Finally, some critics may contend that not much empirical support was found for
the proposed hypotheses regarding the relationship between contexts and three TOL
indicators, because this research employed relatively weak measures in terms Of some
independent variables. For example, as reported in Chapter IV, the reliability estimate Of
the measures for the professional learning community variable is .59, which means that
the magnitude Of this reliability estimate is modest.6 Measurement error can also lead to
bias in predictors. That is, the measure developed here is more likely to fail to tap
important dimensions of professional learning community construct and so tO be an
invalid measure of the underlying theoretical constructs. Due to this limitation, it could
be argued that better measures of the professional learning community variable would
have modest effects on teachers’ learning Opportunities. This is possible, but it is also
worth noting that a substantial body of previous research is consistent with these ﬁndings

(for a review, see McLaughlin & Talbert, 2001).

 

6 Future studies using Item Response Theory (IRT) may improve the reliability of measures for the
variables. IRT is considered the most effective way to increase reliability by providing information about
measurement error in the variables. That is, higher levels of IRT infomration indicate higher precision and
thus greater reliability. By incorporating IRT information about measurement error in hierarchical models,
Raudenbush and Bryks (2002) mobilize an advanced HLM that handles measurement error in the latent
variables and increases the reliability Of the latent variables. In this model, the level 1 analysis represents a
measurement model for the latent variables, by designing a matrix linked to the item responses with the
“true scores” or latent variables being measured for each individual. At level 2, the latent variables varied
among individuals within groups. And at level 3, the coefficients Of the level-2 model vary randomly across
higher level groups (for more review, see Raudenbush & Bryks, 2002, p. 148). However, the current study
did not adopt the advanced HLM because the model does not technically allow weighting variables.

153

Implications of the Study
Based on the results presented in the previous chapter, this study Offers
implications for policy research in support Of McLaughlin and Talbert’s multi-
perspective embedded context model as a useful conceptual approach for a better
understanding of relationships between contexts and TOL. Further, it draws policy and
practice implications for the improvement Of TOL. Yet, all Of these implications for
policy and practice should be taken cautiously due to the methodological limitations just

noted. Hence, the implications for policy and practice in this study are speculative, not

deﬁnitive.

Implications for Policy Research

This study provides several useful implications for policy researchers that
investigate the relationship between school contexts and effective TOL.

First, this study suggests that the multi-perspective embedded context model is a
useful ﬂarnework for policy researchers that examine the relationship between school
contexts and TOL. This model provides a better understanding Of the highly interactive
and quite complex contextual inﬂuences on TOL via the construct Of embedded contexts
— macro and micro contexts directly or interactively affect TOL. Throughout the series of
analyses presented in this study, empirical evidence was found supporting the general
approach drawn ﬂom McLaughlin and Talbert. The ﬁndings Of this study indicate that
diverse contexts at the school and district level are interrelated.

Further, this study suggests the complex and interactive relationships not only

between multilevel school contexts and TOL, but also between diverse contextual factors

154

and TOL and among contextual factors. By comprehensively considering a wide variety
of contextual factors drawn from the three perspectives Of the model, this study
demonstrates that diverse social system contexts, administrative contexts, and
institutional contexts factors have complex and interactive relationships with TOL. For
example, conventional wisdom is that teacher autonomy contributes to improving TOL.
However, this study, using the multi-perspective embedded context model, showed that
different aspects Of teacher autonomy can have different effects on TOL and accordingly
all aspects Of teacher autonomy are not necessarily effective for TOL. Thus, adopting the
multi-perspective embedded context model as a conceptual ﬂamework would help policy
researchers understand the complex and interactive relationships between the speciﬁc
factors in the multilevel school contexts and TOL. For instance, one possible research
question for policy researchers would be how collective autonomy and individual
autonomy affect effective TOL in secondary and elementary schools respectively.

Consequently, the multi-perspective embedded context model is a useful
conceptual approach that deserves policy researchers’ further attention. The conceptual
approach that conceives Of “context” as a complex construct contributes to a better
understanding about the relationship between multiple embedded “real” school contexts
and TOL. Although this study is restricted tO two levels based on the data available in the
SASS survey, it supports the assumption that the full model Of multiple embedded
contexts (including states or subject departments) may show the complex and interactive
relationships between these multilevel contexts and diverse contextual factors at each
level and TOL.

Second, this study demonstrates the possibility Of operationalizing the multiple

155

perspective embedded context model using the multilevel statistical technique (HLM). To
capture teachers’ experiences within the multilevel and embedded contexts, McLaughlin
and Talbert (2001) suggest a combination of both qualitative and quantitative
measurement techniques. They expect that using quantitative measurement techniques
enables detection of overall tendencies of the effects on teachers’ schooling experiences
at each school level, while using qualitative techniques helps account for interdependency
or interaction among complex contexts. Instead of a combination of both qualitative and

quantitative methods, this study captures the features of multilevel and embedded

arr-“*1

contexts using a new quantitative technique such as Hierarchical Linear Model (HLM).
First, HLM can estimate overall tendencies of the effects on teachers’ experiences at each
school level, through estimating the model that partitions the variation in TOL across
educational system levels at the same time. Then, it estimates interactive effects of
diverse contextual factors on teachers’ experiences’, by mobilizing interactive models.
The advantage Of HLM to capture the nature of multilevel and embedded contexts
is supported by literature that focuses on the utility Of the HLM technique in the complex
process Of schooling. For example, Frank (1998) points out that HLM technique can help
us achieve important insights and understanding about the nature, causes, and
consequences Of relations among contexts. By considering the effects of different levels
Of the educational system, (such as the district, school, and classroom) and by examining
relations among contexts, HLM addresses the limitations Of traditional methods for

quantitative analyses which have a moderate capacity to address these complex processes.

 

7 Traditional regression models can also estimate the interaction of two attributes using interaction terms.
However, the models fail to capture the characterization of schools as complex organizations partially
deﬁned by relations among people afﬁliated with the school due to the assumption Of independent error
terms. By contrast, the HLM technique is capable of capturing the complexity Of the organization of
schooling, by estimating one source Of dependency among teachers, the common schools with which they
are afﬁliated.

156

Further, the variance decomposition technique (nested ANOVA) used by McLaughlin
and Talbert (2001) was quite similar in principle tO the HLM procedure (Grodksy &
Gamoran, 2003), even though they did not use HLM.

Consequently, the multilevel statistical technique (HLM) is a possible
methodological approach to studying teachers’ experiences within the multilevel and
embedded school contexts. Using this technique enables policy researchers to estimate
not only overall tendencies of the inﬂuences on teachers’ experiences at each level of
school systems but also independent or interactive effects among diverse contextual

factors.

Implications for Policy and Practice

As TOL has been underscored as central to the standards-based reform, there are
increasing discussions on the potential Of contexts as critical inﬂuences for TOL. The
purpose Of the current study contributes to these efforts to understand the relationship
between contexts and TOL.

First, district units have to be given more attention in terms of the improvement
Of TOL. The evidence ﬂom the national data suggests that districts can play an important
role in providing teachers’ professional development and facilitating teachers’ learning.
Several aspects of district policies could be considered useful for effective strategies for
TOL. For example, districts’ provision Of professional development programs for
principals may help teachers in a timely manner. Principals’ instructional capacity
developed through participation in the professional development programs is more likely

to improve TOL not only by providing practical advice for teachers, but also enabling

157

teachers tO connect their learning experiences over time. Districts’ policy makers can also
stimulate teachers’ participation in learning Opportunities by considering useful strategies
for continuous improvement, such as sending messages about general expectations Of
school improvement and speciﬁc commitment to teacher learning. Furthermore, district
administrative contexts can promote the professional learning communities that value and
support teachers’ learning. By requiring tasks that are to be done collaboratively or
establishing goal setting for school improvements, districts can strengthen collegial
relationships among teachers, and support a culture of inquiry. In these communities,
teachers can have more Opportunities to improve their knowledge of subject matter
content, actively engage in intellectual and meaningful discussion, Observation, and
presentation, and spend more time on their learning. Therefore, both policymakers and
educational leaders need to consider how district strategies enhance effective TOL
independently or by combination with school level contextual factors.

Second, the importance of institutional contexts must be recognized and discussed
by policy makers. The national data suggests that differences in teachers’ learning are
partly attributable to the differences in institutional environments outside Of educational
systems. For example, teachers’ expectations of student learning or teachers’ willingness
to learn is partly shaped by communities’ circumstances. The differences in communities’
circumstances are mainly connected to the differences in sociO-economic conditions
(Metz, 1990). Therefore, policymakers must consider that educational policy will weaken
the inﬂuences ﬂom the different institutional contexts, in particular social or economic
conditions on teachers’ learning.

Finally, some speciﬁc contextual factors thought to inﬂuence TOL in the

158

literature are suspect. The evidence ﬂom the national SASS data suggests that despite the
premise of effects of teacher autonomy on TOL, teacher autonomy does not always help
teacher learning. High levels of individual autonomy may discourage teachers ﬂom
making attempts at innovate or to learn from colleagues while teachers’ collective
autonomy encourages teachers’ learning. Moreover, all kinds of district resources or
accountability strategies are not necessarily effective on TOL either. Scheduled time for
TOL rather than simply releasing teachers ﬂom their teaching would be effective on TOL.
Furthermore, while district strategies using the performance indicators would be useﬁrl
for stimulating teachers’ participation in learning Opportunities, district strategies focused
on accountability related to performance have perverse effects. Hence, policymakers and
educational leaders need to verify cautiously the effectiveness Of these factors in order to

use as strategies for improvements of TOL.

159

APPENDICES

160

Appendix A

2000 SASS VARIABLES AND ITEMS USED IN DATA ANALYSIS

TOL Measures
Content Focus [3 items]

In the past 12 months, have you participated in any professional development activities
that?

T0159 In-depth study of the content in your MAIN teaching assignment ﬁeld
T0162 Content and performance standards in your MAW teaching assignment
T0171 Student assessment, such as methods of testing, evaluation, performance

assessment, etc
Active Learning [4 items]

In the past 12 months, have you participated in the following activities related to
teaching?

T0152 Observational visits to other schools

T0153 Individual or collaborative research on a topic of interest to you
professionally

T0155 Mentoring and/or peer Observation and coaching, as part Of a formal
arrangement that is recognized or supported by the school or district

T0158 Workshops, conferences or training in which you were the presenter

Total Hours [6 items]

In the past 12 months, how many hours did you spend on the activities?

T0160 In-depth study of the content in your MAW teaching assignment ﬁeld

T0163 Content and performance standards in your MAN teaching assignment
ﬁeld

T0166 Methods Of teaching

T0169 Uses Of computers for instruction

T0172 Student assessment, such as methods Of testing, evaluation, performance
assessment, etc

T0175 Student discipline and management in the classroom

161

 

 

Teacher Background Variables

Female
T0356 Are you male or female?
Teaching Experience

TOTEXPER Teacher's total number of years teaching full or part-time in public

Math
T0102 This school year, what is your main teaching assignment ﬁeld at this
school, that is, the ﬁeld in which you teach the most classes?
(Math = 1 Others = 0)
English
T0102 This school year, what is your main teaching assignment ﬁeld at this 5

school, that is, the ﬁeld in which you teach the most classes?
(English = 1 Others = 0)
School Level Variables
Professional Learning Community [5 items]

DO you agree or disagree with each of the following statements?
(Strongly agree =1 , Strongly disagree = 4)

T0307 The principal talks with me ﬂequently about my instructional practices

T0309 Most of my colleagues share my beliefs and values about what the central
mission of the school should be.

T0311 There is a great deal of cooperative effort among the staff members.

T0316 I make a conscious effort to coordinate the content of my courses with
that of other teachers.

T0319 I plan with the library media specialist/librarian for the integration of

library media services into my teaching.
Supportive Leadership [5 items]

DO you agree or disagree with each of the following statements?
(Strongly agree =1, Strongly disagree = 4)

T0299 The principal lets staff members know what is expected Of them.
T0300 The school administration's behavior toward the staff is supportive and
encouraging.

162

T0306

T0310

T0312

My principal enforces school rules for student conduct and backs me up
when I need it.

The principal knows what kind Of school he/she wants and has
communicated it to the staff.

In this school, staff members is recognized for a job well done.

Collective Autonomy [7 items]

Using the scale 1-5, where 1 means "NO inﬂuence" and 5 means "A great deal of
inﬂuence," how much actual inﬂuence do you think teachers have over school policy at
this school in each Of the following areas?

T0286
T0287
T0288
T0289
T0290
T0291
T0292

(N O inﬂuence =1 , A great deal of inﬂuence =5)

Setting performance standards for students at this school

Establishing curriculum

Determining the content of in-service professional development programs
Evaluating teachers

Hiring new full-time teachers

Setting discipline policy

Deciding how the school budget will be spent

Individual Autonomy [6 items]

Using the scale 1-5, where 1 means "NO control" and 5 means "Complete control," how
much control do you think you have in your classroom at this school over each Of the
following areas Of your planning and teaching?

T0293
T0294
T0295
T0296
T0297
T0298

(NO control =1 Complete control =5)

Selecting textbooks and other instructional materials
Selecting content, topics, and skills to be taught
Selecting teaching techniques

Evaluating and grading students

Disciplining students

Determining the amount Of homework to be assigned

Planning Time

T0274

In your most recent full week of teaching, how much scheduled school
time did you have for planning? Hours

Student-Teacher Ratio

STU_TCH Number of students per full-time equivalent teacher in the school

163

Student Total Enrollment

80092 Around The ﬁrst Of October, what was the total number of students
enrolled in this school in grades K-12 and comparable ungraded levels?

Secondary School

SCHLEVEL NCES Created Variable
(Elementary =1 Secondary = 2 Combined = 3)

Suburban

URBANTC (Large or mid-size central city =1 Urban ﬁinge of large or mid-size city =
2 Small town/Rural = 3)

% Minority

MNENR Percent minority students NCES Created Variable

District Level Variables

Time Support I

T0179 For the professional development in which you participated in the last 12
months, did you receive the following types Of support? Release time ﬂom
teaching (i.e., your regular teaching responsibilities were temporarily
assigned to someone else)

Time Support 11

T0180 For the professional development in which you participated in the last 12

months, did you receive the following types of support? Scheduled time in
the contract year for professional development

Money Support I

T0181 For the professional development in which you participated in the last 12
months, did you receive the following types Of support? Stipend for
professional development activities that took place outside regular work
hours

Money Support 11 [2 items]

For the professional development in which you participated in the last 12 months, did you
receive the following types Of support?

164

T0182
T0183

Co-funding

Full or partial reimbursement Of college tuition
Reimbursement for conference or workshop fees

[7 items]

Are the following sources of funding for teacher professional development activities used

in this district?

D1104
D1605
D1606
D1607
D1608
D1609
D1610

Principal Learning

State professional development funds
Special project budgets

School improvement funds

Title I

Eisenhower program

Other federal program

Private sector grants

[5 items]

Does this district provide the following professional development Opportunities for
school or district administrators?

D1590
D1591

D1592
D1593

D1596

Decentralization of PD

Training in evaluation and supervision

Training to use technology for planning, budgeting, decision-making, and
reporting

Training about advances in curriculum, teaching, and assessment

F orrnal networking Opportunities for personnel with similar
responsibilities

Opportunities to serve as mentors within the district

[3 items]

With regard to the in-service professional development activities for TEACHERS in this
district, who has PRIMARY responsibility for (teacher)

Teacher = 1 Principals or other school staff = 2 District staff = 3 Outside professional
development providers (e. g., university or college faculty, professional organizations) = 4

D1599
D1600
D1601

Deciding the content
Designing and planning the activities
Conducting the activities

Performance Indicator

D1533

Does this district use performance reports to evaluate the progress of
students in your district or schools?

165

Accountability [2 items]

D1542 Does this DISTRICT reward schools for student achievement?
D1543 Does this DISTRICT sanction schools for poor student achievement?

District Total Enrollment

D0456 Around the ﬁrst of October, what was the total number of students
enrolled in this district in ALL grade levels?

% Free-lunch

NSLAPP Percent Of K-12 students in district who were approved for ﬂee or
reduced-price lunches in the National School Lunch Program.

166

Eve 1. 8. Va ..

 

:23— oBm com

 

EL mm. =85 32 8.5.0.5

3.. _ Na. aim—o. 8:665 Botan—

uuwwo. 38m. 2..me b=3§5o8<

:3. .13. :E. :Neo. wanes nausea

i8. .5er .68. :MS. :89 $528.2 use.

caomc. a... —om. . sic— —. aim—O. aeovm. 230— . maﬁa—£60

:5... ..S. :57 :28.- ..Be. :a: .- :8? mega use:

£2. twee. .. _ S. is. :NS. .38. :58. 2:8. 253 see:

eavwo. peace. ask-mo.- ..ihoo. genoc- aewmo. ELNO. E38.- ooo. Ndnnm 08C-

:2 _. teem. 2..er :89- .18.- .18.- :So. :3... .33.- 2%. 8.;

33¢. .52.— .- umeN. :vmor peeve. 2.3m. :03. 38v. uuwmm. .3388 com

3.35.- samNo. names-a cab—Oi cammo. comm — .. cowg- aimw — .1 2.33.. Dana—m

3N3: sawhor sumo _. gamma. . :3 n .3. g mo. since. 2. SN. 3:5.- 285 :82 Bow—5m

eel-ma.- ..ime. 2.30.. 2&8.- ..imoo. 2.58.- 1.er seem-o.- ..immor _o>o_ geooom

£58. to _ or same _. sum—c. page. 2&8.- ..iomor 33c. pro—o. owned-m

tog.- ..immc. summer sews. sic—o. 938.- ..im 8. :08. .38.. 95H .53

:08.- .uuwmﬁ 2&3.- ..inc. a... _ Q .- :2 n- fog.- ..iaz .- 32. _ .- >Eoeo§< 65

:95. 2&3. u... _ mo... :58. promo: .35?- ..im 8. 1&2 .- 2&3 .- EBB—83 «oozoo

:NS. 258. :ms. .5er :58. :48- ...§. rune.- ..Neor ensues. useueeaem

:MS. :35. twee.- ..oS- 2&8.- ..wmcr :48. :g .- :89- ceases wagon

2.55.. ..i _ S. u... So. So. 1&3. :25. :5.- ..imoor «av—o. ﬁzwem

aim-NO. al.—No. invoc- aavoc. {av—o. since. pics. SO. sic—0.- 532

pea—or Eva—o.- ..immqr COO. new 2v.- ..iNNOr 1.08.- camNor sic—O.- dxm an?

semwc. ec—NO. saN—o. Shoo.- ..iN—o. SL8.- ..iogc. {Ln-C.- eew~o. .0350"—

.imom. aim-mg . aims. :5. anvmo. aaNmo. 1.08. {L- ~o. .1.me mason "SOH-

suN-Nﬂ. promo. 33o. 3:5. 3:6. uswco. 2&3. o8. .3..er o>uo<

uamo~. six: . tau-VG. aim—o. uncoo- eeg. cameo. ELK-o. pawns. mace-«8:50

:53 «533 a: wane-82 b: 8339.: «58:85 no.5— out
>282 >282 3:33.00 ﬁes-convex 392:5 4985583. BOO-8m 39 BEDQ com mac—nuts»
enou-

 

esemsa as use as a muse... 25 mazes... 2a: a Bases was: He use: 8:228 pacesam mm<m 88 _.< use-F

m aux—25¢.

167

 

~5.VQ ...... m5. V5 ..

 

 

N59... 35.

1.8.. 2...... B...-

:sa- :58.- ..Eéu ......

2.5~N.u ......Ng. ......m #5.- Danna—w

......N5~. 2.5:. 2.3.5.- ....5m5r Bonum

:S... :3... :SO- :8...- :...o.- .25. 32.8%

2.555. ......mwN. ......m55. 2.555. 2.. m55.- ....vm5r ova-Hm

2.555... ......mmN. 2.52. 2.5N5. 2.555.- 2: _5. {...—N5. REP darn

......wN~ . 2.555.- ......NVN. ......v55... ...... — —5. 2.55”- ...... ~55. 2.53. >Eoco~=< .55

2.. _ mv. 2.5.5. ......mmor ......Nvof ......N w — .- 2.305. 2.5m — ... ......wwo. 2.555. hanou=< «00:00

...... :2...- ::o. .....N..- :E..- :8... :E..- :95. :2... EES... ”gonna

2.555. 2.505.- ......wmor 2..th6 .32.”- ......BN5. 2.5— ~ .- 2.555. 2.555. ban—EEO“. magnum

..iwmo. 2.555. 2.555.- 2.55~ . 2.. _ w5. 2L.~5( 2.25. 2555. 2.5N5. nmzwcm

2.53. ......mno. ......N55- 255—. 23; ~ . 2.355. 2.5N5.- 2..N—5... 2.555... 532

2.555. ......Nmo. 2.5N5.- ......hN5. 555. ......mmor 2.505.- ......NN5.- 2.25. .&m ...-Cr

2.555.: 2. _55.- 2.555.- 2.550- 2.50_ ... ......NNO. nib—5. ......wmo. 2.5M5. 0350....—

25m5.- .....hN5f ......VN5. 2..er 2.53.- «ioﬁof 2.555. 2.32. 2.555. mason Each.

.....m55. 2.555.- .....5—5. .....5m5... 2.555.- ......NN5. ......hmo. ......wv— . 255— . o>ﬂo<

2.555.- 2.53.- six-N5. 2.55 ~ .. ......vwof 555. 2.555. 2.5m _ . 2'5; . mace-28300
a 833.5»

82.92. . 96 E 8.8 GOES . .
.9: up... .83 2.89m raw—aooom .Sohmvvam p.595 ”Mm . _ answ- oEC. N .33 085-

 

€3.35... 2... as... ...... ... .05.... 2.... .528... 2d. ... 8.83... we... 8. 5...: Essence .....Ezm 3% 88 ...... 2......

168

 

 

169

SK .. 8. Va _.
.....va. maﬁa—vau— 25835
:5. :NR b32980 maﬁa:
......mmcr .....wmc.- ......NSV swam—um
238.- 308.. 2.08.- ...L 8.- £32
......NCOf ......ONO. .329. ..iwccr ...LNOB dam— .mu>
......MMOV ......awor niece. ......Vm m . .3,qu {*MNO. 03th
2.2:. 3.53. 3.35.. ......m—O. 8..— _ _ . 31.3. ......mho. mun—o: 33H.
......mnm. {ﬁx—O. Sea—o. 2.63... ......woc. ......m— m . ......mbO. tuna: . u>ﬂo<
......cvm. :80. ......om _ . immo. 336.- tbs. :2 ﬂ. «ammo. :30. mucouéaoﬁou
mmEmCm>
o>=o< mason 58K 285m 9Q a; :32 ﬁzwcm 3:58:50 95.832 .=< 80:00
. . waged 3an96

 

902??» 0.8 038 £5 E 855 05.5 mix—£2. 24: E @8585 88: .5.“ x532 coca—2.80 Batatm mm<m ocom _.< 03¢

REFERENCES

American Federation of Teachers (1995). Principles for professional development: AFT ’s
guidelines for creating professional development programs that make a difference.
Washington, D. C.

Ball, D. L., & Cohen, D. K. (1999). Developing practice, developing practitioners:
Toward a practice-based theory of professional development. In L. Darling-
Hammond & G. Skyes (Eds.), Teaching as the learning professional: Handbook
of policy and practice (pp. 3-32). San Francisco: Jossey-Bass.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, N.J.: Prentice-Hall.

Bellanca, J. A. (1995). Designing professional development for change: A systemic
approach. Palatine, IL: IRI/Skylight Pub.

Berry, B., Turchi, L., Johnson, D., Hare, D., Owens, D., & Clements, S. (2003). The
impact of high-stakes accountability on teachers' professional development:
Evidence ﬁom the South. Chapel Hill, NC: Southeast Center for Teaching Quality,
Inc. Retrieved June 7, 2004, from
http://www.teachingquality.org/resources/pdfs/Spencer_FinalReport.pdf.

Bickel, R., & Howley, C. (2000). The inﬂuence of scale on student performance: A
multi-level extension of the Mattew Principle. Education Policy Analysis
Archives, 8 (22).

Borko, H., & Putnam, R. (1996). Leaming to teach. In D. C. Berliner & R. C. Calfee
(Eds.), Handbook of educational psychology (pp. 673-708). NY: Macmillan.

Bransford, J ., Brown, A. L., & Cocking, R. R. (1999). How people learn: Brain, mind,
experience, and school. Washington, DC: National Academy Press.

Bryk, A. S., & Driscoll, M. E. (1988). The school as community: Theoretical foundations,
contextual inﬂuences, and consequences for students and teachers. Chicago:
Center for School Improvement at the University of Chicago.

Bryk, A., Camburn, E., & Louis, K. S. (1999). Professional community in Chicago
elementary schools: Facilitating factors and organizational consequences.
Educational Administration Quarterly, 35, 751-781.

Bryk, A. S., & Raudenbush, S. W. (2002). Hierarchical linear models (2nd ed.).
Newbury Park: Sage.

Clement, M., & Vandenberghe, R. (2000). Teachers' professional development: A

170

solitary or collegial (ad)venture? Teaching and Teacher Education, 16, 81-101.

Cobb, P., & Bauersfeld, H. (1995). The emergence of mathematical meaning: Interaction
in classroom cultures. Hillsdale, N.J.: L. Erlbaum Associates.

Cohen, D. K., & Barnes, C. A. (1993). Pedagogy and policy. In D. K. Cohen, M. W.
McLaughlin & J. E. Talbert (Eds), Teaching for understanding: Challenges for
policy (pp. 207-239). San Francisco: Jossey-Bass Inc.

Cohen, D. K., & Hill, H. C. (2001). Learning Policy: when state education reform works.
New Haven: Yale University Press.

Confrey, J. (1990). A review of the research on student conceptions in mathematics,
science, and programming. In C. Cazden (Ed.), Review of Research in Education

(Vol. 16, pp. 3-56). Washington, DC: American Educational Research
Association.

Corcoran, T. C. (1995). Transforming professional development for teachers: A guide of
state policymakers. Washington, DC: National Governors' Association.

Corcoran, T. C., Passantino, C., & Gerry, G. (2000). Mapping professional development
opportunities: A pilot study of two subjects in three regions. Lexington:
Partnership for Kentucky Schools.

Cox, T. (1993). Cultural diversity in organizations: Theory, research, and practice (lst
ed.). San Francisco, Calif: Berrett-Koehler.

Darling-Hammond, L. (1995). Inequality and access to knowledge. In J. A. Banks & C. A.
Banks (Eds), Handbook of research on multicultural education (pp. 465-483).
NewYork: Macmillan.

Darling-Hammond, L., & McLaughlin, M. (1995). Policies that support professional
development in an era of reform. Phi Delta Kappan, 76(8), 597-604.

Darling-Hammond, L., & Sykes, G. (1999). Teaching as the learning profession:
Handbook of policy and practice (1 st ed.). San Francisco: J ossey-Bass Publishers.

Darling-Hammond, L., Hightower, A. M., Husbands, J. L., LaFors, J. R., Young, V. M., &
Christopher, C. (2003). Building instructional quality: “Inside-out" and “outside-
in ” perspectives on San Diego Is school reform. University of Washington Center
for the Study of Teaching and Policy, document R-O3-3.

David, J. L., McDiarmid, G. W., & Corcoran, T. B. (2000). District leadership in

professional development: Exemplary cases. Lexington: Partnership for Kentucky
Schools.

171

David, J. L., & Shields, P. M. (2001). When theory hits reality: standards-based reform
in urban districts. Mnlopark, CA: SRI International.

Desimone, L., Porter, A. C., Birman, B. F., Garet, M. S., & Yoon, K. S. (2002). How do
district management and implementation strategies relate to the quality of the

professional development that districts provide to teachers? Teachers College
Record, 104(7), 1265-1312.

Elmore, R. (2002). Bridging the gap between standards and achievement: The imperative
for professional development in education. New York: Albert Shanker Institute.

Elmore, R., & Burney, D. (1999). Investing in teacher learning: Staff development and
instructional improvement. In L. Darling-Hammond & G. Sykes (Eds), Teaching
as the learning profession: Handbook of policy and practice (pp. 263-291). San
Francisco: J ossey Bass.

F enstermacher, G. D., & Berliner, D. C. (1985). A conceptual framework for evaluating
staff development. Elementary School Journal, 85, 281-314.

Fink, E., & Resnick, L. (2001). Developing principals as instructional leaders. Phi Delta
Kappan, 82(8), 598-606.

Firestone, W. A., Herriott, R. E., & Wilson, B. L. (1984). Explaining differences in
between elementary and secondary schools: Individual, organizational, and
institutional perspectives. Philadelphia, PA: Research for Better Schools, Inc.

Firestone, W. A., Mayrowetz, D., & Fairman, J. (1998). Performance-based assessment
and instructional change: The effects of testing in Maine and Maryland.
Educational Evaluation and Policy Analysis, 20(2), 95-113.

Firestone, W. A., Monfils, L. F ., & Schorr, R. Y. (2004). The ambiguity of teaching to the
test: Standards, assessment, and educational reform. Mahwah, NJ: L. Erlbaum
Associates Publishers.

Firestone, W. A., Mangin, M. M., Martinez, M. C., & Polovsky, T. (2005). Leading
coherent professional development: A comparison of three districts. Educational
Administration Quarterly, 41(3), 413-448.

Fisk, C. W. (1997, April) The costs and benefits of designing and implementing a
portfolio-based support and assessment system for beginning teachers. Paper
presented at the Annual Meeting of the American Educational Research
Association, Chicago.

Frank, K. A. (1998). Quantitative methods for studying social context in multilevels and
through interpersonal relations. Review of Research in Education, 23, 171-216.

172

 

Fullan, M. (1995). The limits and the potential of professional development. In T. R.
Guskey & M. Huberrnan (Eds), Professional development in education: New
paradigms and practices (pp. 253-267). New York: Teachers College Press.

Garet, M., Porter, A., Desimone, L., Birman, B., & Yoon, S. K. (2001). What makes
professional development effective? Analysis of a national sample of teachers.
American Education Research Journal, 38(4), 915-945.

Goodman, L. A. (1978). Analyzing qualitative/categorical data: Log-linear models and
latent-structure analysis. Cambridge, Mass: Abt Books.

Grodsky, E., & Gamoran, A. (2003). The relationship between professional development
and professional community in American schools. School eﬂectiveness and school
improvement, 14(1), 1-29.

Grossman, P. L., & Stodolsky, S. S. (1994). Considerations of Content and the
Circumstances of Secondary School Teaching. In L. Darling-Hammond (Ed.),
Review of Research in Education (V 01. 20, pp. 179-221). Washington, DC:
American Educational Research Association.

Guskey, T. R. (1997). Research needs to link professional development and student
learning. Retrieved 1 1-13-04 from http://www.nsdc.org/library/jsd/jsdgusk.html.

Hannaway, J. & Kimball, K. (1998). Big isn 't always bad: School district size, poverty,
and standards-based reform. Report prepared for the planning and evaluation
service of the US. Department of Education. The Urban Institute, Washington,
DC.

Hawley, W., & Valli, L. (1999). The essentials of effective professional development: A
new consensus. In L. Darling-Hammond & G. Sykes (Eds), Teaching as the
learning profession: Handbook of policy and practice (pp. 151-180). San
Francisco: J ossey Bass.

Hightower, A., Knapp, M. S., Marsh, J ., & McLaughlin, M. (2002). School districts and
instructional Renewal. New York: Teachers College Press.

Hox, J. (2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Erlbaum.

Ingersoll, R. M., & Alsalam, N. (1997). Teacher professionalization and teacher
commitment: A multilevel analysis (NCES 97-069). In Statistical analysis report.
Washington, D. C.: National Center for Education Statistics.

Ingvarson, L., Meiers, M., & Beavis, A. (2005). Factors affecting the impact of

professional development programs on teachers' knowledge, practice, student
outcomes & efﬁcacy. Education Policy Analysis Archives, 13(10).

173

Kennedy, M. (1998). Form and substance in in-service teacher education. Arlington,
VA: National Science Foundation.

King, M. B., & Newmann, F. M. (2000). Will teacher leaming advance school goals? Phi
Delta Kappan, 81, 576-580.

Klonsky, M. (2002). How smaller schools prevent school violence. Educational
Leadership, 59(5), 65-69.

Knapp, M. S. (2003). Professional development as a policy pathway. Review of Research
in Education, 27, 109-157.

Kreft, I. G. G., DeLeeuw, J ., & Aiken, L. S. (1995). The effect of different forms of
centering in hierarchical linear models. Multivariate Behavioral Research, 30(1),
1-2 1 .

Lee, V. E., Smith, J ., & Bryk, A. S. (1993). The organization of effective secondary
schools. In L. Darling-Hammond (Ed.), Review of Research in Education (V 01. 19,
pp. 171-267).

Lieberman, A. (1994). Teacher development: Commitment and challenge. In P. P.
Grimmett & J. Neufeld (Eds), Teacher development and the struggle for
authenticity: Professional growth and restructuring in the context of change (pp.
15-30). New York: Teachers College Press.

Lieberman, A. (1996). Practices that support teacher development: Transforming
conceptions of professional learning. In M. W. McLaughlin & I. Oberman (Eds),
Teacher learning: New policies, new practices (pp. 185-201). New York:
Teachers College Press.

Little, J. W. (1982). Norms of collegiality and experimentation: Workplace conditions of
school success. American Educational Research Journal, 19(3), 325-340.

Little, J. W. (1990a). The persistence of privacy: Autonomy and initiative in teachers'
professional relations. Teachers College Record, 91(4), 509-536.

Little, J. W. (1990b). Conditions of professional development in secondary schools. In M.
W. McLaughlin & J. E. Talbert (Eds), The contexts in question: The secondary
school workplace (pp. 187-223). New York: Teachers College Press.

Little, J. W. (1999a). Organizing schools for teacher learning. In L. Darling-Hammond &
G. Sykes (Eds), Teaching as the learning profession. Handbook of policy and
practice (pp. 233-262). San Francisco: Jossey Bass.

Little, J. W. (1999b). T eachers' professional development in the context of high school
reform: Findings from a three-year study of restructuring schools. Washington, D.

174

C.: National Partnership for Excellence and Accountability in Teaching.

Little, J. W. (2001). Professional development and the pursuit of reform. In A. L. a. L.
Miller (Ed.), Teachers caught in the action: Professional development that
matters (pp. 23-44). New York: Teachers College Press.

Lave, J ., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation.
Cambridge, England: Cambridge University Press.

Lord, B. (1994). Teachers' professional development: Critical colleagueship and the role
of professional communities. In N. Cobb (Ed.), The future of education:
Perspectives on national standards in education (pp. 175-204). New York:
College Entrance Examination Board.

Loucks-Horsley, S., Hewson, P. W., Love, N., & Stiles, K. E. (1998). Designing
professional development for teachers of science and mathematics. Thousand
Oaks, Calif: Corwin Press.

Marks, H. M., & Louis, K. S. (1999). Teacher empowerment and the capacity for
organizational learning. Educational Administration Quarterly, 35, 707-750.

Massell, D. (1998). State strategies for building local capacity: Addressing the needs of
standards-based reform. Philadelphia: Consortium for Policy Research in
Education.

McCutchen, D., Abbott, R. D., Green, L. B., Beretvas, S. N., Cox, 8., Potter, N. S., et al.
(2002). Beginning literacy: Links among teacher knowledge, teacher practice, and
student learning. Journal of Learning Disabilities, 35, 69-86.

McLaughlin, M. W. (1994). Strategic sites for teachers' professional development. In P. P.
Grimmett & J. Neufeld (Eds), Teacher development and the struggle for
authenticity: Professional growth and restructuring in the context of change (pp.
31-51). New York: Teachers College Press.

McLaughlin, M. W., & Talbert, J. E. (1990). The contexts in question: The secondary
school workplace. In M. W. McLaughlin, J. E. Talbert & N. Bascia (Eds), The
contexts of teaching in secondary schools (pp. 1-14). New York: Teachers
College Press.

McLaughlin, M. W., & Talbert, J. E. (2001). Professional communities and the work of
high school teaching. Chicago: University of Chicago Press.

Metz, H. M. (1990). How social class differences shape teachers' work. In M. W.
McLaughlin, J. E. Talbert & N. Bascia (Eds), The contexts of teaching in
secondary schools (pp. 40-107). New York: Teachers College Press.

Miller, R. J. (2003). The sources and consequences of organic management in public

175

elementary and secondary schools. Unpublished Ph.D dissertation, The
University of Michigan, Michigan.

Murphy, J ., & Hallinger, P. (1988). Characteristics of instructionally effective school
districts. Journal of Educational Research, 81(3), 175-181.

Newmann, F. M. (1993). Beyond common sense in educational restructuring: The issues
of content and linkage. Educational Researcher, 22(2), 4-13.

Newmann, F. M. (1996). Center on organization and restructuring of schools: Activities
and accomplishments, [990-1996. Final report. Madison, WI: Center on
Organization and Restructuring of Schools.

North, D. (1992). Institutions, Institutional Change and Economic Performance.
Cambridge University Press.

Ostrom, E. (1990). Governing the Commons: Institutions for collective action.
Cambridge University Press

Parsad, B., Lewis, L., & Farris, E. (2001). Teacher preparation and professional
develOpment. Washington, DC: National Center for Education Statistics.

Perry, R. (1996). The role of teachers’ professional communities in the implementation of
California Mathematics reform. Unpublished Ph.D dissertation, Stanford
University, Califomia.

Poulson, L., & Avramidis, E. (2003). Pathways and possibilities in professional
development: Case studies of effective teachers of literacy. British Educational
Research Journal, 29(4), 543-560.

Putnam, R., & Borko, H. (2000). What do new views of knowledge and thinking have to
say about research on teacher learning? Educational Researcher, 29(1), 4-15.

Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and
data analysis methods (2nd ed.). Thousand Oaks: Sage Publications.

Rosenholtz, S. (1989). T eachers' workplace: The social organization of schools. New
York: Longrnan.

Rowan, B. (1990). Commitment and control: Alternative strategies for the organizational
design of schools. Review of Research in Education, I 6.

Schon, D. A. (1983). The reﬂective practitioner: How professionals think in action. New
York: Basic Books.

Scott, W. R. (1982). Organizations: Rational, Natural and Open Systems (4th Edition).

176

Englewood Cliffs, NJ: Prentice-Hall.

Scribner, J. (1999). Professional development: Untangling the inﬂuence of work context
on teacher learning. Educational Administration Quarterly, 35(2), 238-266.

Smith, M. T., & Rowley, J. K. (2005). Enhancing commitment or tightening controlzThe
function of teacher professional development in an era of accountability.
Educational Policy, 19(1), 126-154.

Smylie, M. A. (1988). The enhancement function of staff development: Organizational
and psychological antecedents to individual teacher change. American
Educational Research Journal, 25(1), 1-30.

Smylie, M. A. (1994). Redesigning teachers' work: Connections to the classroom. Review
of Research in Education, 20, 129-177.

Smylie, M. A. (1995). Teacher learning in the workplace: implications for school reform.
In T. R. Guskey & M. Huberman (Eds), Professional development in education:
New paradigms and practices (pp. 92-113). New York: Teachers College Press.

Spillane, J ., & Thompson, C. (1997). Reconstructing Conceptions of Local Capacity: The
Local Education Agency's Capacity For Ambitious Instructional Reform.
Educational Evaluation and Policy Analysis, 19(2), 185 - 203.

Spillane, J ., & Zeuli, J. (1999). Reform and teaching: Exploring patterns of practice in the
context of national and state mathematics reforms. Educational Evaluation and
Policy Analysis, 21(1), 1-27.

Spillane, J. P. (2004). Standards deviation: How schools misunderstand education policy.
Cambridge, Mass: Harvard University Press.

Stein, M. K., D'Amico, L., & Israel, N. (1998). Observations, conversations, and
negotiations: Administrator support of literacy practices. Pittsburgh: Learning
Research and Development Center, University of Pittsburgh.

Sykes, G. (2002). Professional development for teachers: Principles, practices, and
context. Washington, D. C.: Learning First Alliance.

Sykes, G. (2005). An inquiry concerning professional development for teachers: NEA.
Unpublished manuscript.

Talbert, J. E., & McLaughlin, W. M. (1994). Teacher professionalism in local school
contexts. American Journal of Education, 102, 123-153.

Talbert, E. J ., & McLaughlin, W. M. (1999). Assessing the school environment:
Embedded contexts and bottom-up research strategies. In L. S. W. Friedman, D. T.

177

(Ed.), Measuring Environment Across the Life Span (pp. 197-227). Washington,
DC: American Psychological Association.

Thompson, C. L., & Zeuli, J. S. (1999). The frame and the tapestry: Standards-based
reform and professional development. In L. Darling-Hammond & G.Sykes (Eds),
Teaching as the learning profession: Handbook of policy and practice (pp. 341-
375). San Francisco: Jossey-Bass.

Togneri, W., & Anderson, S. E. (2003). Beyond islands of excellence: What districts can
do to improve instruction and achievement in all schools. Washington, DC:
Learning FirstAlliance.

Tourkin, C. S., Pugh, K. W., Fondelier, S. E., Partner, R. J., Cole, C., Jackson, B., Warner,
T., Weant, G, Walter, E., Gruber, K., Zhao, L. (2004). 1999-2000 Schools and
Staﬂing Survey (SASS) Data File User 3' Manual. Washington, DC: National
Center for Education Statistics.

von Glasersfeld, E. (1984). An introduction to radical constructivism. In P. Watzlawick
(Ed.), The Invented Reality (pp. 17-40). New York: W.W. Norton & Company.

Vygotsky, L. S., & Cole, M. (1978). Mind in society: The development of higher
psychological processes. Cambridge: Harvard University Press.

Wiley, D., & Yoon, B. (1995). Teacher reports of opportunity to learn: Analyses of the
1993 California Learning Assessment System. Educational Evaluation and Policy
Analysis, 17(3), 355-370.

Wilson, S. M., & Beme, J. (1999). Teacher learning and the acquisition of professional
knowledge: An examination of research on contemporary professional
development. Review of Research in Education, 24, 173-209.

Wilson, S. M., Darling-Hammond, L., & Berry, B. (2001). A case of successful teaching
policy Connecticut's long-term eﬂ'orts to improve teaching and learning. Seattle:
University of Washington, Center for the Study of Teaching and Policy.

Youngs, R, & King, M. B. (2002). Principal leadership for professional development to
build school capacity. Educational Administration Quarterly, 38(5), 643-670.

Zheng, H. Y. (1996). School contexts, principal characteristics, and instructional

leadership effectiveness: A statistical analysis. Paper presented at the Annual
Meeting of the American Educational Research Association, New York, NY.

178

   

Il'ﬂlﬂlllﬂilllﬂllﬁﬂl[1111:1111