Citation
An investigation into the influences on mathematics teacher efficacy in elementary teachers

Material Information

Title:
An investigation into the influences on mathematics teacher efficacy in elementary teachers a mixed methods study
Creator:
Swackhamer, Lyn Ely
Publication Date:
Language:
English
Physical Description:
xix, 245 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Mathematics -- Study and teaching (Elementary) ( lcsh )
Mathematics teachers ( lcsh )
Teacher effectiveness ( lcsh )
Elementary school teachers ( lcsh )
Elementary school teachers ( fast )
Mathematics -- Study and teaching (Elementary) ( fast )
Mathematics teachers ( fast )
Teacher effectiveness ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 233-245).
General Note:
School of Education and Human Development
Statement of Responsibility:
by Lyn Ely Swackhamer.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
437259848 ( OCLC )
ocn437259848
Classification:
LD1193.E3 2009d S82 ( lcc )

Full Text
AN INVESTIGATION INTO THE INFLUENCES ON MATHEMATICS
TEACHER EFFICACY IN ELEMENTARY TEACHERS:
A MIXED METHODS STUDY
by
Lyn Ely Swackhamer
B.B.A., North Georgia College, Dahlonega, 1987
M.A., University of Colorado, Denver, 2005
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
School of Education and Human Development
2009


by Lyn Ely Swackhamer
All rights reserved.


This thesis for the Doctor of Philosophy
degree by
Lyn Ely Swackhamer
has been approved
by
Karen Koellner
Nancy L. Leech
3
A
ate


Swackhamer, Lyn Ely (Ph.D., Educational Leadership and Innovation)
An Investigation into the Influences on Mathematics Teacher Efficacy in Elementary
Teachers: A Mixed Methods Study
Thesis directed by Professor Karen Koellner
ABSTRACT
The purpose of the study was to investigate the influences on elementary
teachers levels of mathematics specific teacher efficacy in order to inform teacher
preparation and professional development programs. Eighty six inservice elementary
teachers participated in this mixed methods research study. All 86 participants took a
survey that consisted of a mathematics specific teacher efficacy instrument, a
collective teacher efficacy instrument, a mentor support instrument, and a math
anxiety instrument. From this pool of participants, 4 teachers with high levels of
mathematics teacher efficacy and four teachers with low levels of mathematics
teacher efficacy were selected for qualitative interviews.
The results indicate that mathematics content knowledge, years of teaching,
and teacher support directly influence levels of mathematics teacher efficacy. In the
current study, high levels of mathematics teacher efficacy produced the teacher
behaviors of innovative teaching and a high level of perceived effectiveness as a math
teacher, which in turn led to higher levels of differentiation of instruction and more
confidence in using any curriculum source. Low levels of mathematics teacher
efficacy produced the teacher behaviors of more teacher centered classrooms and


lower levels of perceived effectiveness as a math teacher. These behaviors led to
being uncomfortable answering student questions and a need for scripted curriculum
materials. The study demonstrates that teachers with high levels of mathematics
teacher efficacy have higher levels of mathematics content knowledge.
This abstract accurately represents the content of the candidates thesis. I recommend
it for publication.
Signed
aren fvoeilner


DEDICATION
I dedicate this dissertation to my husband, Bill. I will never be able to give to
you what you have given to me. Without your love, support, encouragement, and
belief in me, I would be a much lesser woman. I thank you so much for always being
there for me for the past 21 years and most especially the past 3. I also dedicate this
to my wonderful daughters, Emily and Katie. Thank you for your love and support,
but especially thank you for the sacrifices you had to endure in order for me to
accomplish this goal.


ACKNOWLEDGMENTS
The completion of this dissertation would not have been possible without the
encouragement and support of many people. I would not be where I am today
without these people, and I will be eternally grateful to each one of them.
I would like to thank Dr. Karen Koellner for her continued support and
friendship over the last 3 years. Karen, you have acted as my advisor, my committee
chair, my mentor, and my friend. Your support and belief in me has been invaluable,
and I thank you so much for continually pushing me to achieve more.
I would also like to thank my committee members, Dr. Carole Basile, Dr.
Nancy Leech, and Dr. Larry Linnen. Carole, you have been part of my educational
career for 9 years, and I thank you so much for your ongoing support and guidance
throughout those years. You have been a valuable teacher, advisor, and friend, and I
truly appreciate all of your help. Nancy, I would not have begun this journey and
surely could not have ended it without your continued presence. I value you as a
teacher, mentor, and friend and thank you so much for your continuing support and
guidance. Larry, although you joined my journey at the end, your presence during
my teaching career and your help during the dissertation process have been
invaluable. Thank you.
I cannot fully express the love and gratitude I have for my friends and family
who supported me along this journey. Thank you for all of your love and support


over the years. To all of my immediate and extended family, I would not have been
able to make or complete this journey without you.


TABLE OF CONTENTS
FIGURES......................................................xvii
TABLES.......................................................xviii
CHAPTER
1. INTRODUCTION.................................................1
Conceptual Framework......................................3
Overview of the Study.....................................5
Definition of Variables...................................8
2. REVIEW OF LITERATURE........................................11
Social Cognitive Theory and Self-Efficacy................11
History and Measurement of Teacher Efficacy..............15
Teacher Efficacy.....................................16
Adding subconstructs..............................17
Broadening the construct..........................18
New models of teacher efficacy....................19
Measurement of Teacher Efficacy......................20
Locus of control measures.........................21
Beginning teacher efficacy measures...............24
Social cognitive theory measures..................28
Summary...........................................31
IX


Research Studies in the Field of Teacher Efficacy
32
Variable Relationships with Teacher Efficacy...............33
Teacher behaviors......................................33
School specific variables..............................45
Student achievement....................................50
Summary................................................52
Measuring Changes in Teacher Efficacy......................53
Professional development...............................53
Teacher preparation and induction......................62
Score differences......................................68
Summary................................................70
Examining the Influences on Teacher Efficacy...............71
Exploring Banduras sources of teacher efficacy........72
Influence of personal variables........................77
School related variables...............................80
Effect of teacher efficacy on other variables..........84
Summary................................................86
Conclusion.....................................................87
3. METHODS...........................................................90
Rationale and Purpose for a Mixed Methods Inquiry..............91
Research Questions.............................................91
x


Research Design.................................................94
Setting and Participants........................................95
Sampling Scheme.................................................96
Measures........................................................97
Quantitative Measures......................................97
TSES....................................................97
AMAS....................................................98
CES.....................................................98
MEMT....................................................98
Qualitative Measure........................................99
Data Analysis..................................................100
Quantitative Analyses.....................................100
Qualitative Analyses......................................103
Reliability and Validity.......................................105
Quantitative Analysis.....................................105
Mixed Method Legitimation.................................105
Limitations and Delimitations..................................106
Limitations...............................................106
Delimitations.............................................107
Researchers Perspective.......................................107
xi


4. RESULTS
109
Quantitative Results........................................Ill
Reliability and Validity................................111
CES.....................................................Ill
AMAS....................................................113
MEMT....................................................114
TSES.....................................................114
Quantitative Research Questions la, 2a, and 2b..............117
Research Question la....................................118
Overall mathematics teacher efficacy................119
Efficacy for instructional strategies................119
Efficacy for classroom management....................122
Efficacy for student engagement......................123
Research Question 2a.....................................125
Research Question 2b.....................................127
Qualitative Results.........................................127
Constant Comparative Analysis...........................132
Elementary math teacher efficacyhigh group.........134
Elementary math teacher efficacylow group..........135
Comparison of math efficacy to other subjectshigh
group...........................................136
Comparison of math efficacy to other subjects-low group.. 136
xii


Personal learning efficacyhigh group................137
Personal learning efficacylow group.................138
Content knowledgehigh group.........................139
Content knowledgelow group..........................140
Professional development experiencehigh group.......141
Professional development experiencelow group........141
Years of teaching experiencehigh group..............142
Years of teaching experiencelow group...............143
Grade level teachinghigh group......................144
Grade level teachinglow group.......................144
Curriculumhigh group................................145
Curriculumlow group.................................145
Math anxietyhigh group..............................146
Math anxietylow group...............................147
Mentor/coach supporthigh group......................147
Mentor/coach supportlow group.......................148
Collective teacher efficacyhigh group...............149
Collective teacher efficacylow group................149
Other influenceshigh group..........................150
Other influenceslow group...........................150
Domain Analysis.........................................151
xiii


Taxonomic Analysis.......................................153
Elementaiy math teacher efficacyhigh group..........154
Elementary math teacher efficacylow group...........154
Comparison of math efficacy to other subjectshigh
group............................................155
Comparison of math efficacy to other subjectslow group.. 155
Personal learning efficacyhigh group................156
Personal learning efficacylow group.................156
Content knowledgehigh group.........................157
Content knowledgelow group..........................157
Professional development experiencehigh group.......158
Professional development experiencelow group........158
Years of teaching experience-high group..............158
Years of teaching experiencelow group...............159
Grade level teachinghigh group......................159
Grade level teachinglow group.......................159
Curriculumhigh group................................159
Curriculumlow group.................................160
Math anxietyhigh group..............................160
Math anxietylow group...............................160
Mentor/coach supporthigh group......................161
Mentor/coach supportlow group.......................161
xiv


Collective teacher efficacyhigh group
161
Collective teacher efficacylow group.................162
Other influenceshigh group...........................162
Other influenceslow group............................163
5. DISCUSSION.......................................................164
Reliability and Validity of the Data Produced from the TSES..165
Linking the Findings to the Existing Literature..............166
Acquiring Mathematics Content Knowledge..................168
Experience and Support...................................169
Internal Factors.........................................173
Teacher Behaviors........................................176
Implications of the Findings.................................178
Directions for Future Research...............................182
APPENDIX
A. SUMMARY OF RESEARCH ON THE HISTORY OF TEACHER
EFFICACY.....................................184
B. MEASURES BASED UPON A LOCUS OF CONTROL CONCEPT.... 186
C. EVALUATIONS OF RELIABILITY AND VALIDITY FOR
LOCUS OF CONTROL BASED INSTRUMENTS...........188
D. MEASURES BASED UPON BANDURAS THEORY OF
TEACHER EFFICACY.............................190
E. EVALUATIONS OF RELIABILITY AND VALIDITY FOR
BANDURA BASED INSTRUMENTS....................192
xv


F. STUDIES MAKING SUGGESTIONS TO IMPROVE
MEASUREMENT OF TEACHER EFFICACY............194
G. SUMMARY OF RESEARCH INTO VARIABLE RELATIONSHIPS
WITH TEACHER EFFICACY......................196
H. STUDIES MEASURING CHANGES IN TEACHER EFFICACY.204
I. INFLUENCES ON TEACHER EFFICACY STUDIES.......213
J. RESEARCH SURVEY..............................222
K. INTERVIEW PROTOCOL...........................230
REFERENCES..........................................233
xvi


LIST OF FIGURES
Figure
1.1 Conceptual framework for teacher efficacy for mathematics.................6
3.1 Teacher efficacy research study rationale and purpose.....................92
3.2 Explanatory design with follow-up expectations for teacher efficacy study.94
4.1 Taxonomic analysis of high efficacy group.................................155
5.1 Conceptual model of mathematics teacher efficacy..........................167
xvii


LIST OF TABLES
Table
1.1 Description of Variales Used in the Study..................................9
2.1 Comparison of Items from the RSA or TLC with Items from the TES............27
4.1 Factor Loadings for the CES Scale Rotated Factors..........................112
4.2 Factor Loadings for the AM AS Scale Rotated Factors........................113
4.3 Factor Loadings for the MEMT Scale Rotated Factors.........................115
4.4 Factor Loadings for the TSES Scale Rotated Factors.........................116
4.5 Summary of Results for Question la.........................................118
4.6 Means, Standard Deviations, and Inter-correlations for Math Teacher
Efficacy and Predictor Variables......................................120
4.7 Backward Multiple Regression Analysis Summary for Math Content
Knowledge, Mentor Effectiveness, Years of Teaching, Collective
Teacher Efficacy, and Perceived Effectiveness on Predicting
Math Teacher Efficacy.................................................120
4.8 Means, Standard Deviations, and Inter-correlations for Efficacy for
Instructional Strategies and Predictor Variables......................121
4.9 Backward Multiple Regression Analysis Summary for Math Content
Knowledge, Mentor Effectiveness, Years of Teaching, and
Perceived Effectiveness on Predicting Efficacy for
Instructional Strategies..............................................122
4.10 Means, Standard Deviations, and Inter-correlations for Efficacy for
Classroom Management and Predictor Variables..........................123
xviii


4.11 Backward Multiple Regression Analysis Summary for Math Content
Knowledge, Mentor Effectiveness, Years of Teaching, and
Collective Teacher Efficacy on Predicting Efficacy for
Classroom Management................................................123
4.12 Means, Standard Deviations, and Inter-correlations for Efficacy for
Student Engagement and Predictor Variables..........................124
4.13 Backward Multiple Regression Analysis Summary for Math Content
Knowledge and Collective Teacher Efficacy on Predicting
Efficacy for Student Engagement.....................................124
4.14 Summary of the Results for Question 2a.................................125
4.15 Means, Standard Deviations, and Inter-correlations for Teacher Efficacy
for Low Efficacy Participants and Predictor Variables Variables....126
4.16 Backward Multiple Regression Analysis Summary for Math Content
Knowledge, Perceived Effectiveness, and Math Anxiety on
Predicting Low Group Teacher Efficacy...............................126
4.17 Summary for Results of Question 2b......................................127
4.18 Transcript of Interview With 1 High and 1 Low Efficacy Participant.....129
4.19 Summary of Results of Questions lb......................................130
4.20 Summary for Results of Question 2c......................................132
4.21 Categories Used for Qualitative Data Analysis...........................133
4.22 Axial Coding Selection..................................................134
4.23 Domain Analysis Worksheet...............................................152
4.24 Examples of Domain Analysis for High Efficacy Group.....................153
xix


CHAPTER 1
INTRODUCTION
Mathematics education in the United States is under fire. Achievement scores
are down or stagnant across the country, and the 2006 PISA study on the mathematics
achievement of 15 year olds revealed that 23 other industrialized countries across the
world scored higher than American students (Baldi, Jin, Skemer, Green, & Herget,
2007). At the center of this dilemma is hiring and retaining quality mathematics
instructors. Ingersoll and Perda (2006) reported the supply of math and science
teachers is lower than the supply of teachers in other subjects. They also reported that
secondary schools have acknowledged great difficulty in staffing math and science
positions with quality teachers. Teacher quality has also been the subject of recent
reports from the National Mathematics Advisory Panel (2008) and the National
Council on Teacher Quality (2007). Both reports recognized that quality mathematics
teachers can significantly impact student achievement, and thus recommend
strengthening teacher preparation and professional development programs in order to
increase mathematics content knowledge and pedagogical content knowledge. While
the shortage in quality mathematics teachers occurs at all levels, effective elementary
teachers are an essential element in the production of high achieving mathematics
students (National Council on Teacher Quality). The National Council of Teachers of
Mathematics (NCTM) stresses that teachers must have a deep understanding of the
1


mathematics they are hired to teach (National Council of Teachers of Mathematics,
2000). Unfortunately, studies have shown that preservice elementary teachers scored
well below the average entering college student on 2007 mathematics SAT scores
(College Board, 2007), have demonstrated the highest level of mathematics anxiety of
any college major grouping (Hembree, 1990), and tend to take the lowest number of
college level mathematics credits (Harper & Daane, 1998). These findings indicate
that some elementary teachers may not be highly effective at preparing their students
for the rigorous study of higher level mathematics. However, a highly effective
mathematics teacher must possess more than content knowledge, he or she must also
understand how to teach all mathematical concepts to all students of all abilities. This
ideal combination of content knowledge and pedagogical content knowledge is
difficult to measure, thus in order to learn from highly effective teachers, we can
attempt to identify them in other ways.
One potential way of identifying highly effective teachers seems to be through
an individuals level of teacher efficacy. Teacher efficacy has been defined as how
competent the teacher feels in his or her ability to affect the performance of all
students, no matter how unmotivated or difficult (Tschannen-Moran, Woolfolk Hoy,
& Hoy, 1998). Studies have shown that highly effective teachers tend to have high
levels of teacher efficacy. In fact, a number of studies have concluded that teachers
with high levels of efficacy differ significantly from teachers with low levels of
efficacy. Specifically, high efficacy teachers demonstrate more effective teacher
2


behaviors that lead to higher student achievement (Ashton, Webb, & Doda, 1983;
Gibson & Dembo, 1984; Guskey, 1988). Teachers with low efficacy levels are more
likely to: (a) set incomplete goals, (b) lose motivation, (c) perceive their ability as
incompetent, and (d) lose interest in completing the task (Bandura, 1997).
Unfortunately, simply identifying high and low efficacy teachers will not provide
information on how to increase levels of efficacy to become more effective teachers
of mathematics. What is needed is a deep understanding of the influences on
mathematics teacher efficacy. Therefore, the purpose of this study is to investigate
the influences on elementary teachers levels of mathematics specific teacher efficacy
in order to inform teacher preparation and professional development programs on
how best to incorporate efficacy-building components into these programs.
Conceptual Framework
This study, exploring the influences on mathematics teaching efficacy, is
grounded in Banduras social cognitive theory and specifically his concept of self-
efficacy. This theory is explained in more detail in the following chapter, but briefly,
self-efficacy is the judgment of ones ability to produce desired results and forestall
detrimental ones through ones own actions (Bandura, 2001, p. 10). Self-efficacy
influences all aspects of decision making and outcome production through goal
setting, motivation, perceived ability, and interest. If a person believes he or she is
fully capable of completing a task and achieving the desired result, she will set
appropriate goals, be fully motivated, perceive her ability level to be adequate to the
3


task, and will be interested in the outcome. Two people may have the same skills and
knowledge to complete a task but may perform in entirely dissimilar manners if their
level of efficacy differs significantly (Bandura, 1993). It is also important to note that
efficacy levels are context and task specific, meaning an individual possesses efficacy
for any and all proposed tasks, and that the efficacy level for a proposed task may
change within differing contexts. For example, a persons level of efficacy for
teaching multiplication may be high in a classroom of high ability students and low in
a classroom of special education students.
Bandura (1997) proposed four sources of self-efficacy: mastery experiences,
vicarious experiences, social persuasion, and physiological and affective states.
Mastery experiences are the actual successful or unsuccessful outcomes of
performing the task. These experiences are considered the most powerful influence
on a persons level of self-efficacy. Vicarious experiences occur when a person sees
someone with perceived similar ability perform the task. Social persuasion refers to
having a more knowledgeable other encourage the person to perform the task, and
physiological and affective states refers to the physical and mental processes that may
interfere with the performance of the task, such as exhaustion or anxiety. These
influences are affected by four mental processes: cognitive, motivational, affective,
and selection (Bandura, 1997). Cognitive processes can enhance or inhibit the
performance on the task through the self-appraisal of ones capabilities. Motivation
can affect efficacy through the establishment of goals, while the ability to control
4


negative affective thoughts and enhance positive thoughts also has a strong impact on
efficacy beliefs. Finally, selection refers to the choices people make that are based
upon their efficacy beliefs.
Using this theory of self-efficacy, researchers have applied the ideas to the
task of teaching. Applying Banduras concepts to the field of mathematics teacher
efficacy requires a few changes in order to make the theory more applicable to
mathematics education. This proposed model is described in Figure 1 and takes
Banduras sources of self-efficacy and transforms them into more teaching specific
influences on efficacy. These influences include: personal teaching experiences,
comparable others, mentor/coach, and internal sources. Personal teaching
experiences consist not only of actual teaching experiences, but also the individuals
level of content knowledge and teaching experience. The influence of comparable
others refers to the individual teachers perceived confidence in the teaching ability of
his or her colleagues. The next influence is the experience of having a mentor or
coach work with the individual teacher. The final influence of internal sources refers
to the impact of stressors and anxiety on efficacy beliefs. The four influences on
mathematics teacher efficacy are then filtered through the individuals cognitive
processing before becoming a measurable level of efficacy.
Overview of the Study
The goal of this mixed methods research study was to gain a greater
understanding of the influences on mathematics specific teacher efficacy for
5


Figure 1.1. Conceptual framework for teacher efficacy for mathematics.
elementary teachers in order to inform teacher education and professional
development programs. Through this study, I attempted to predict the influences on
mathematics teaching efficacy, add to the current knowledge base, have a personal,
social and organizational impact upon the teaching profession, and gain a deeper
understanding of a complex phenomenon. The study randomly selected 600
elementary teachers from a large suburban school district and a large urban school
district in a western U.S. state. The main objectives of this mixed methods inquiry
was to: (a) build a model that significantly predicts the influences on math teacher
efficacy, and (b) qualitatively examine these influences on math teacher efficacy and
the individual cognitive processing by interviewing a random selection of teachers
6


who exhibit high and low levels of teacher efficacy. Therefore, the following
research questions were explored:
la. Do the independent variables of math anxiety, math content knowledge,
perceived collective efficacy, perceived mentor effectiveness, number of years
teaching experience, and perceived level of mathematics teaching effectiveness
significantly predict overall math teacher efficacy and/or the subscales of efficacy for
instructional strategies, efficacy for student engagement, and efficacy for classroom
management?
lb. How do teachers with high and low levels of math teaching efficacy
explain how the influences of personal teaching experiences, comparable others,
mentor/coach, and internal sources have affected their level of math teaching
efficacy?
In addition, several secondary research questions will also be examined:
2a. Do the independent variables of math anxiety, math content knowledge,
perceived collective efficacy, perceived mentor effectiveness, number of years
teaching experience, and perceived level of mathematics teaching effectiveness
significantly predict overall math teacher efficacy for the group of teachers with
efficacy levels below the mean score?
2b. How do teachers self-reported levels of math teaching effectiveness
significantly correspond to the calculated level of math teaching efficacy?
7


2c. Do the teachers with high levels of efficacy report different influences
than teachers with low levels of efficacy?
Definition of Variables
In this study, I explored the relationship between the independent variables of
personal teaching experience, mentor/coach experience, comparable others, and
internal states on a teachers sense of math efficacy. Table 1.1 provides the definition
for each of the variables used in the study, the variable type (independent or
dependent), a connection to the specific research question using the variable, and an
explanation of how the variable will be measured.
The selection of these variables provided needed information on the
influences on teacher efficacy. Teachers with higher levels of efficacy have been
shown to demonstrate characteristics of highly effective teachers and produce higher
achieving students (Ashton et al., 1983; Gibson & Dembo, 1984; Guskey, 1988).
Therefore, it is important to identify levels of efficacy in order to explore the factors
that have influenced those levels. By quantitatively exploring the variables of
personal teaching experience, mentor/coach experience, comparable others, internal
states, and mathematics teacher efficacy, a model of how these influences affect
teacher efficacy was built. This model was qualitatively explored in order to confirm
and expand upon the relationship between these influences and efficacy levels.
8


Table 1.1
Description of Variales Used in the Study
Research question Variable name Variable type Definition Measurement
la, lb, 2a, 2b, 2c Teacher efficacy Dependent How competent the teacher feels in his or her ability to affect the performance of all students. Teacher Sense of Efficacy Scale (TSES)
la, lb, 2a Math anxiety Independent Feeling of tension and anxiety that interfer with the manipulation of numbers and solving of mathematical problems (Richardson & Suinn, 1972, p. 551) Abbreviated Math Anxiety Scale (AMAS)
la, lb, 2a Math content knowledge Independent Perception of each participants level of mathematics knowledge. Rating scale (1- 5)
la, lb, 2a Perceived collective teaching efficacy Independent The perceived capability a group of teachers has towards impacting student achievement. Collective Teacher Efficacy Scale (CTE)
la, lb, 2a Mentor support Independent Level of support the participating teachers perceive and how the mentor/coach relationship has impacted their level of effectiveness. Mentoring for Effective Mathematics Teaching Scale (MEMT)
la, lb, 2a Year of teaching experience Independent Number of years the participant has been teaching Participant provide the number
la, lb, 2a Math teaching effectiveness Independent How effective the participant perceives his or her effectiveness in teching math Rating scale (1- 5)
This study provides important information for the field of teacher efficacy.
Identifying specific influences and determining the strength of their impact on levels
9


of mathematics teacher efficacy will allow teacher preparation and professional
development programs to develop strategies and activities that will help increase
efficacy levels. By examining elementary teachers in the context of teaching
mathematics, the study provides the education community with valuable information
towards developing effective mathematics teachers that are able to support our
youngest students in the attainment of deep conceptual and fluent mathematical
content knowledge that prepares them for the rigors of higher-level mathematics.
10


CHAPTER 2
REVIEW OF LITERATURE
The purpose of the study was to investigate the influences on elementary
teachers levels of mathematics specific teacher efficacy. First, this chapter will
describe the components of social cognitive theory and self-efficacy as developed by
Bandura (1997). Second, the concept of teacher efficacy will be explored through an
examination of the construct and the development of instrumentation. Finally,
relevant studies in teacher efficacy will be reviewed in the following three categories:
(a) examining variable relationships with teacher efficacy, (b) measuring changes in
teacher efficacy, and (c) investigating influences on teacher efficacy.
Social Cognitive Theory and Self-Efficacy
Social cognitive theory examines the human capacity to exercise control over
the nature and quality of ones life (Bandura, 2001, p.l). This exercise of control or
human agency allows individuals to make things happen through intentional actions
(Bandura, 2001). Bandura (2001) described human agency as being guided through
four cognitive outlets: intentionality, forethought, self-regulation, and self-
reflectiveness. Intentionality refers to the ability of humans to choose a course of
action. Forethought refers to setting goals, anticipating likely consequences, and
choosing actions that are more likely to produce desired outcomes instead of
undesirable outcomes. Self-regulation refers to the human ability to monitor the
11


chosen choice of action through goal setting and motivation. Self-reflectiveness
refers to the evaluation of the persons motivation, values, and meaning of the chosen
action. For example, a person may be faced with the task of driving a car over a
hundred miles to another city. This person, through intentionality and forethought,
may choose a driving path that avoids all major highways. Through self-regulation
the person continues to find the appropriate back roads that avoid the highways, while
self-reflecting on the motivation and meaning behind this choice of action. This self-
reflection aspect of human agency is considered most influential for a person
achieving the desired outcome as this is how people make judgments on the outcomes
and effects of their choice of action (Bandura, 2001).
Judging ones ability to produce desired results and forestall detrimental
ones through actions is defined as self-efficacy (Bandura, 2001, p. 10). Self-efficacy
influences all aspects of decision making and outcome production through goal
setting, motivation, perceived ability, and interest. If a person believes he or she is
fully capable of completing a task and achieving the desired result, she will set
appropriate goals, be fully motivated, perceive her ability level to be adequate to the
task, and will be interested in the outcome. A person with a low level of self-efficacy
for the task will set incomplete goals, lose motivation, perceive his ability as
incompetent, and lose interest in completing the task. Two people may have the same
skills and knowledge to complete a task but may perform in entirely dissimilar
manners if their level of efficacy differs significantly (Bandura, 1993). It is also
12


important to note that efficacy levels are context and task specific, meaning an
individual possesses efficacy for any and all proposed tasks, and that the efficacy
level for a proposed task may change within differing contexts. For example, a
persons level of efficacy for teaching multiplication may be high in a classroom of
high ability students and low in a classroom of special education students. In order to
distinguish where the levels of self-efficacy begin to differ, researchers must have a
great understanding of the levels of difficulty associated with the task under
investigation (Bandura, 2007).
Bandura (1997) proposed four sources of self-efficacy beliefs: mastery
experiences, vicarious experiences, social persuasion, and physiological and affective
states. Mastery experiences are the actual successful or unsuccessful outcomes of
performing the task. These experiences are considered the most powerful influence
on a persons level of self-efficacy. Vicarious experiences occur when a person sees
someone with perceived similar ability perform the task. Social persuasion refers to
having a more knowledgeable other encourage the person to perform the task, and
physiological and affective states refers to the physical and mental processes that may
interfere with the performance of the task, such as exhaustion or anxiety. These
influences are affected by four mental processes: cognitive, motivational, affective,
and selection (Bandura, 1997). Cognitive processes can enhance or inhibit the
performance on the task through the self-appraisal of ones capabilities. Motivation
can affect efficacy through the establishment of goals, while the ability to control
13


negative affective thoughts and enhance positive thoughts also has a strong impact on
efficacy beliefs. Finally, selection refers to the choices people make that are based
upon their efficacy beliefs.
It is important to note that a persons level of self-efficacy may not be an issue
for routine tasks or for those tasks where skills have been well established but can be
highly important for learning tasks that require new skills (Bandura, 1997). Schunk
(1988) determined that self-efficacy may predict student motivation and learning of
new tasks. Through the use of task engagement variables such as instructional
context, instructional events, and performance feedback, students assess their efficacy
levels through cues such as outcome patterns, social comparisons, and attributions.
Careful monitoring is required in order to ensure all learners maintain engagement in
the task. Unfortunately, a person may have high levels of efficacy for a task, but still
become unmotivated to complete the task. Ability level also has an impact on
interest, task completion and self-efficacy. Lopez, Lent, Brown, and Gore (1997)
suggested that ability affects interest both directly and indirectly through self-
efficacy. The results of the study demonstrated the effect of ability on self-efficacy
was integrated into the other efficacy sources. Therefore, a person with high ability
may still have low efficacy through an incorrect analysis of his or her objective
ability. Efficacy beliefs can also influence aspirations and how strongly a person is
committed to the aspiration. Bandura, Barbaranelli, Caprara, and Pastorelli (2001)
studied childrens aspiration and career trajectories to determine the impact of self-
14


efficacy upon these future possibilities. They determined that levels of perceived
efficacy had a direct effect on the childrens selection of possible career choices.
With the advent of a global economy and multicultural societies, it is
important to understand how cultural differences can impact self-efficacy beliefs.
Bandura (2002) explored the aspect of culture on social cognitive theory and self-
efficacy and argues that self-efficacy is a universal concept that is valued culturally
by individualistic societies such as the United States and by collectively oriented
societies such as Japan. Possession of a high level of personal efficacy can increase
the likelihood of success whether it is achieved individually or within a group.
Bandura (2002) also stressed the differences within cultural groups can be as large as
between cultural groups and that with increased globalization, cultures are changing
to reflect the viewpoints of others. In the following section, we will explore how
these important tenets of self-efficacy have been, and are currently being, applied to
the teaching profession through the construct of teacher efficacy.
History and Measurement of Teacher Efficacy
After the introduction of social cognitive theory and self-efficacy, researchers
began exploring how this concept could be used in the study of teaching and learning.
This research began with identifying the construct of teacher efficacy and the
development of instrumentation to measure this construct. In the following two
sections, the construct of teacher efficacy will be explored followed by the history of
instrument development and implementation.
15


Teacher Efficacy
Teacher efficacy has been explored along two differing strands of thought.
The first definition uses the idea of locus of control as developed by Rotter (1966).
Locus of control involves the idea of attributing the consequences of actions to an
outside force or to an internal force, thus the prevailing definition of teacher efficacy
was the level of control a teacher believed they had over student achievement and
motivation (Tschannen-Moran et al., 1998). Teacher efficacy was evaluated through:
(a) a teachers perspective of his or her power over outside influences, and (b) his or
her confidence in teaching different strategies to assist student learning. The first
perspective was purported to measure general teaching efficacy while the second
perspective was labeled personal teaching efficacy (Tschannen-Moran et al.).
The second strand of thought regarding teacher efficacy was based upon the
work of Bandura and his concept of self-efficacy as described in the previous section.
Perceived self-efficacy and locus of control have very different theoretical
underpinnings. Locus of control focuses on whether certain actions can affect
outcomes, whereas perceived self-efficacy is a persons internal belief on whether he
or she can actually produce a desired outcome (Tschannen-Moran et al., 1998).
Therefore, the prevailing definition of teacher efficacy using Banduras (1997) work
is the level of belief a teacher has in his or her ability to affect student achievement.
Under the umbrella label of teacher efficacy, researchers have conducted a variety
of studies using one or the other of these two concepts. Unfortunately, the two
16


concepts are not empirically related, do not measure the same phenomenon, and differ
in prediction ability (Bandura, 1997). While perceived self-efficacy strongly predicts
behavior, locus of control has been shown to be a weak predictor (Bandura, 1997).
As Ross (1995) stated, locus of control differs from TE [teacher efficacy] in that
taking responsibility for outcomes does not mean that one anticipates that the
outcomes will be positive (p. 229). Over the last 20 years, researchers have explored
the concept of teacher efficacy and have made suggestions towards expanding the
construct. These suggestions vary from the addition of a new subconstruct, ideas on
how to broaden the construct, to totally restructuring the construct.
Adding subconstructs. With the development of the Teacher Efficacy Scale
(TES) by Gibson and Dembo (1984), teacher efficacy was described as having two
subconstructs: personal teaching efficacy (PTE) and general teaching efficacy
(GTE). PTE is generally defined as a teachers belief in his or her skills and abilities
to positively impact student achievement, while GTE has been defined as a teachers
belief that the educational system can work for all students, regardless of outside
influences such as socio-economic status and parental influence. Although the
instrument was designed to measure teacher efficacy using Banduras theory of self-
efficacy and efficacy expectations, the GTE scale has been found to reflect a locus of
control orientation (Henson, 2002). The instrument became the dominant measure of
teacher efficacy, although dissatisfaction continued with the definition and
measurement of the two subconstructs. Soodak and Podell (1996) argued that teacher
17


efficacy was actually a three dimensional construct consisting of personal efficacy,
outcome efficacy, and teaching efficacy. Using the TES, they added additional items
and through factor analysis, identified the three subconstructs. They defined personal
efficacy as a teachers belief that he or she has the necessary teaching skills.
Outcome efficacy was defined as the belief that when the teaching skills are used they
produce a desired student outcome. Teaching efficacy was defined as the belief that
teachers can overcome the effects of all outside influences on their students. Emmer
and Hickman (1991) also modified the TES to include the subconstruct of classroom
management to the two subconstructs of PTE and GTE. While these studies were
attempting to strengthen the construct of teacher efficacy, they were using a poorly
designed instrument that did not fully capture the complexity of Banduras theory of
self-efficacy. Therefore, researchers began the work of redefining and remodeling
the construct of teacher efficacy.
Broadening the construct. Wheatley (2002, 2005) suggested that teacher
efficacy research be combined with other research into teacher beliefs in order to
strengthen understanding and produce meaningful research that is relevant to teachers
and teacher educators. He argued that high levels of teacher efficacy have not been
empirically shown to be beneficial in all cases. He also stressed that teachers may
benefit from having doubt about their abilities as doubt can create disequilibrium,
foster reflection, support motivation to learn, and promote productive collaboration.
Labone (2004) also argued that the research into teacher efficacy should be broadened
18


in order to deepen understanding of the construct. She suggested further research into
the sources and processing of teacher efficacy and an extension of the research
beyond traditional teacher roles. Using a critical theory perspective, she argues that
teacher efficacy research should extend beyond the traditional instructional skills
focus to teaching skills related to community engagement and society enrichment.
Citing the increasing difference in diversity between the student population and the
current teaching population, she stressed the need to push teachers into looking
beyond the classroom tasks in order to actively engage in social reconstruction (p.
356). These researchers raised important points that should be considered in future
research. Teacher efficacy should be examined in light of other teacher beliefs in
order to more fully examine how belief systems can affect student achievement.
Additional research is also needed in order to determine optimum levels of teacher
efficacy. As Wheatley (2002) pointed out, a teacher who is overconfident in his or
her abilities may not recognize the need for additional learning, therefore future
research should focus on determining how closely a teachers self-reported level of
teacher efficacy matches his or her actual teaching effectiveness.
New models of teacher efficacy. Tschannen-Moran et al. (1998) introduced a
new conceptual model of teacher efficacy that demonstrates the cyclical and context
specific nature of this construct. The model begins with Banduras four sources of
efficacy information: mastery experiences, physiological and emotional states,
vicarious experiences, and social persuasion. These information sources are
19


cognitively processed and then lead to either an analysis of the teaching task or an
assessment of personal teaching competence. The result of either of these processes
lead to a level of teacher efficacy, then to the consequences of that level of efficacy.
This level of efficacy determines the teaching performance, which becomes a new
source of efficacy information. Friedman and Kass (2002) suggested a new model of
teacher efficacy that incorporates the teachers school context and his or her
relationships within the school. This model is based upon the following three ideas:
(a) a teacher works with the school itself and within the individual classroom, (b) a
teacher engages socially with students and with colleagues, and (c) a teacher must
balance between task goals and relational goals. They also proposed a new definition
of teacher efficacy:
a teachers perception of his or her ability to (a) perform required professional
tasks and to regulate relations involved in the process of teaching and
educating students (classroom efficacy), and (b) perform organizational tasks,
become part of the organization and its political and social processes
(organizational efficacy), (p. 684)
These two proposed models of teacher efficacy strive to extend the concept of teacher
efficacy beyond the locus of control roots. The Tschannen-Moran et al. (1998) model
seems to more fully embrace Banduras model of self-efficacy and appears to be a
better match to this complex concept. Additional research is needed to verify both
the underlying components of the model and the model itself. The table in Appendix
A provides a summary of the research presented on the history of teacher efficacy.
Measurement of Teacher Efficacy
20


Of primary importance within this field of study is the development of a
reliable and valid instrument that accurately measures teacher efficacy. The idea of
measuring teacher efficacy began with the inclusion of two items on a Rand study in
the seventies (Armor et al., 1976). Based upon Rotters (1966) concept of locus of
control, Rand researchers included items that explored the perceived ability of
teachers to control their actions and their environment. Item 1 states, When it comes
right down to it, a teacher really cant do much because most of a students
motivation and performance depends on his or her home environment. This item
was labeled general teaching efficacy and reflects agreement with an external locus of
control. Item 2 states, If I really try hard, I can get through to even the most difficult
or unmotivated students. This item has been labeled personal teaching efficacy and
reflects agreement with an internal locus of control. These two items became the
basis and inspiration for teacher efficacy instruments developed by various
researchers.
Locus of control measures. Ashton, Buhr, and Crocker (1984) used the two
Rand items and a 25 item vignette measure in a study of 65 inservice teachers to
attempt to determine if the construct of teacher efficacy is a self-referenced or a
norm-referenced construct. Specifically, they were attempting to determine if
teachers evaluate their effectiveness as an internal personal question, or do they judge
effectiveness as a comparison to other teachers. Using 25 vignettes that represent
different teaching situations, they constructed two forms using either a norm-
21


referenced or a self-referenced approach. The results indicated that the self-
referenced vignette scores were not correlated to the Rand items score, while the
norm-referenced vignette scores were significantly correlated (r = 0.36, p < 0.05)
with the Rand items score. Therefore, in this study it seems that teachers were more
likely to judge their performance against the performance of other teachers. Although
this research was produced early in this field of study, the instruments that have been
most widely used in teacher efficacy research have chosen to use a self-referent
format.
Guskey (1981) developed a 30 item instrument that measures responsibility
for student achievement (RSA). Teachers were asked to determine whether various
situations occurred because of factors within or outside of teacher control. For each
of the questions, the respondent is asked to weight the two choices according to their
preference using one hundred percentage points. For example, question one states: If
a student does well in your class, would it probably be (a) because that student had
the natural ability to do well, or (b) because of the encouragement you offered? The
scale consists of 15 positive situations and 15 negative situations which result in two
scores, an R+ score and an R- score. The measure was given to 215 inservice
teachers and the data demonstrated sufficient validity and reliability. Factor analysis
revealed two factors which explained 60.9% of the variance. Reliability coefficients
were 0.79 for the R+ scale and 0.88 for the R- scale.
22


At the same time, Rose and Medway (1981b) developed a 28 item forced-
choice instrument, the Teacher Locus of Control Scale (TLC), which measures a
teachers general expectancies for student success or failure in the classroom. In this
measure, teachers were asked to choose between two descriptions of student success
or failure by assigning control to internal or external factors. Fourteen of the items
reflect positive situations (When the grades of your students improve, it is more likely
(a) because you found ways to motivate the students, or (b) because the students were
trying harder to do well), and 14 of the items reflect negative situations (When some
of your students fail a math test, it is more likely (a) because they werent attending to
the lesson, or (b) because you didnt use enough examples to illustrate the concept).
Factor analysis confirmed the two factors of 1+ and I- although it appears that some of
the factor loadings were below or near the 0.3 cut off generally recognized. Kuder-
Richardson formula 20 reliabilities were reported of 0.81 for the I- scale and 0.71 for
the 1+ scale.
Henson, Kogan, and Vacha-Haase (2001) examined published reliability
estimates from studies using these two instruments. Only five studies were found that
used the RSA and three that used the TLC; therefore, it is difficult to accurately
estimate the reliability of the subscales for these two measures. Of the four subscales,
the I- scale seemed to reflect the lowest reliability estimates. Although these scales
were not widely used in teacher efficacy research, their impact on the development of
subsequent measures was strong.
23


Beginning teacher efficacy measures. With the introduction of Banduras
self-efficacy theory, as described above, researchers began modifying efficacy
instruments to more closely model these ideas. In 1984, Gibson and Dembo created
the most popular measure thus far, the Teacher Efficacy Scale (TES), which proposes
to capture Banduras two elements of efficacy, efficacy expectation and outcome
expectancy, by having participants respond to statements such as when a student
gets a better grade than he usually gets, it is usually because I found better ways of
teaching and even a teacher with good teaching abilities may not reach many
students. The measure uses a 5-point Likert scale ranging from 1 (strongly disagree)
to 5 (strongly agree). Factor analysis on the data produced from the study revealed
two factors which were labeled personal teaching efficacy and general teaching
efficacy. These two factors only explained 28.8% of the variance which is extremely
low. Cronbachs alpha coefficients were reported as 0.78 for PTE, 0.75 for GTE and
0.79 for the overall scale. The authors acknowledged the two subscales were
developed from the two Rand items which created an instrument that combined locus
of control and self-efficacy theory. This conceptual issue combined with the low
validity of the data should have led to further refinement and more validity and
reliability studies but instead led to this instrument becoming the standard in teacher
efficacy research for 20 years.
With the recognition that efficacy belief is context specific, researchers began
developing scales that focused on specific content areas, such as Riggs and Enochs
24


(1990) Science Teaching Efficacy Belief Instrument (STEBI, which was refined by
Enochs and Riggs [1990] into the STEB1-B), and Enochs, Smith, and Huinkers
(2000) Mathematics Teaching Efficacy Beliefs Instrument (MTEBI). These two
instruments rework the statements on the TES from a more general teaching focus to
a specific science or mathematics teaching focus. For example, I understand science
concepts well enough to be effective in teaching elementary science, and when a
student does better than usual in mathematics, it is often because the teacher exerted a
little extra effort. These scales retained the 5-point Likert scale scoring of the TES,
but also retained the low validity issues with data from both scales typically
explaining less than 30% of the variance in teacher efficacy.
Emmer and Hickman (1991) proposed a new dimension of teacher efficacy-
efficacy in classroom management and discipline. The authors asserted that teachers
spend a considerable amount of time managing student behaviors and therefore the
quality of this management may lead to feelings of high or low efficacy. Using the
TES, the researchers added items that reflected classroom management and discipline
situations. Using factor analysis, the final instrument data revealed three factors,
classroom management/discipline, external influences, and personal teaching
efficacy. The second factor consisted of many of the original general teaching
efficacy items from the TES, but also included some new items created by the
researchers which reflected the relative influence on student behavior of events or
25


characteristics beyond the teachers control (p. 759). Variance explained was not
presented, but Cronbachs alphas for the three scales ranged from 0.69 to 0.83.
Although these instruments purport to embrace Banduras theory of efficacy,
they still are based upon the Rand items and have been shown to correlate with an
intemal/extemal focus (Coladarci & Fink, 1995; Guskey & Passaro, 1993). Guskey
and Passaro noted that all of the items reflecting personal teaching efficacy began
with I and were positively worded with an internal focus. All of the general
teaching efficacy questions referred to teachers and were negatively worded with
an external focus. Deemer and Minke (1999) also noted the pattern of negative and
positive worded items could be creating false loadings. They observed the majority
of items on the general teaching scale reflect negative wording while the personal
efficacy scale reflects positive wording. It should also be noted that many of the
items on the TES are directly related to the RSA or the TLC scales which were
designed based upon the theory of locus of control. Table 2.1 compares the
similarities between several of the items on the RSA or TLC with items from the
TES.
These instruments also have issues with construct validity. Factor analysis
results from these three instruments revealed that several of the items tend to load on
both factors, and some of the items have problematic wording that can lead to
confusing results (Tschannen-Moran et al., 1998). Henson, Bennett, Sienty, and
Chambers (2000) determined the TES consists of three factors they labeled general
26


Table 2.1
Comparison of Items from the RSA or TLC with Items from the TES
Items from RSA/TLC
Items from TES
#4 TLC) When a student gets a better
grade on his report card than he usually
gets, it is
(a) because the student was putting
more effort into his schoolwork, or
(b) because you found better wasy of
teaching that student.
#1 TLC) When the grades of your
students improve, it is more likely
(a) because you found ways to motivate
the students, or
(b) because the students were trying
harder to do well.
#10 RSA) When your students seem to
learn something easily, it is usually
(a) because they were already interested
in it, or
(b) because you have helped them
organize the concepts.
#1) When a student does better than
usual, many times it is because I exerted
a little extra effort. (Strongly agree to
strongly disagree.)
#19) When the grades of my students
improve, it is usually because I found
more effective teaching approaches.
(Strongly agree to strongly disagree.)
#21) If a student masters a new math
concept quickly, this might be because I
knew the necessary steps in teaching
that concept. (Strongly agree to
strongly disagree.)
teaching efficacy (or external locus of control), past assessment of personal teaching
efficacy, and present assessment of personal teaching efficacy. Denzine, Cooney, and
McKenzie (2005) also determined that the TES had a three factor structure which
they labeled: (a) self-efficacy belief, (b) outcome expectancy, and (c) external locus
of control. They determined that the TES does not accurately fit Banduras model of
social learning theory. Although Bleicher (2004) determined that the data produced
from the STEBI-B was valid and reliable, it was only after rewording several of the
27


questions. Factor analyses of data using the TES and the STEBI-B have consistently
only explained around 30% of the variance and the general teaching subconstruct has
not shown a high degree of reliability (Henson et al., 2001). Bandura (1997) also
argued that these measures assume a stable level of efficacy across teaching tasks that
may not be accurate. Therefore, researchers began work on developing new
instruments with stronger construct validity.
Social cognitive theory measures. In response to the issues presented above in
the measurement of teacher efficacy, Bandura (1997) demonstrated that self-efficacy
and locus of control have little or no statistical relationship between each other.
Reacting to the generality of the above mentioned measures and to the lack of
correlation with locus of control, Bandura (n.d.) presented his own 30 item measure
that recognizes the differing tasks that teachers may be asked to perform. His seven
subscales include: efficacy to influence decision making, efficacy to influence school
resources, instructional efficacy, disciplinary efficacy, efficacy to enlist parental
involvement, efficacy to enlist community involvement, and efficacy to create a
positive school climate. The scale is measured on a 9 point Likert scale which details
the amount of influence a teacher believes they exert over the stated situation. The
amount of influences ranges from 1 (nothing) to 9 (great deal). Questions include:
(a) how much can you influence the decisions that are made in your school, (b) how
much can you do to get children to follow classroom rules, and (c) how much can you
do to get local colleges and universities involved in working with your school? While
28


this scale is more conceptually specific and task oriented, the teaching tasks which are
included have been criticized for not representing normal teaching activities
(Tschannen-Moran & Woolfolk Hoy, 2001). There is no reliability and validity
evidence available for this scale.
Roberts and Henson (2000) developed an efficacy instrument that attempts to
address the theoretical and methodological issues associated with the TES for use
with science education teachers. The researchers used the PTE scale from the TES
and added a knowledge efficacy scale in order to develop a new measure that would
more closely match Banduras theory. The knowledge efficacy construct is based
upon pedagogical content knowledge in that it attempts to measure the content
knowledge of the instructor and how that instructor is able to use the content
knowledge to advance student learning. This measure was called the Self-Efficacy
Teaching and Knowledge Instrument for Science Teachers (SETAKIST). Although
no additional studies have used this measure, it did contribute to the ongoing
discussion of what subconstructs should be included in a measure of teacher efficacy.
Using their conceptual model of teacher efficacy that was discussed above,
Tschannen-Moran and Woolfolk Hoy (2001) developed a new measure of teacher
efficacy, the Teachers Sense of Efficacy Scale (TSES). The instrument was
designed to measure both personal competence and an analysis of the task in terms
of the resources and constraints in particular teaching contexts (p. 795). The scale
attempts to clearly define normal teaching tasks and broaden the focus from
29


unmotivated students to include capable students. The sub-constructs of efficacy for
instructional strategies, efficacy for classroom management, and efficacy for student
engagement include questions such as: to what extent can you use a variety of
assessment strategies?, and how much can you do to control disruptive behavior in
the classroom? They presented the results of three studies that helped to refine the
final instrument. This procedure resulted in two versions of the instrument, a 24 item
long version and a 12 item short version. Factor analysis confirmed the existence of
the three factors that accounted for 54% of the variance in the long term and 65% in
the short form. Second-order analysis revealed the measures could conform to one
factor which explained 75% of the variance in the long form and 68% of the variance
in the short form. Item loadings were appropriate for the three subscales and the one
entire scale. Internal reliabilities ranged from 0.87 to 0.91.
This scale seems to have the potential to more fully capture the efficacy
expectation and outcome expectancy subconstructs that are at the heart of Banduras
self-efficacy theory, but additional research utilizing this instrument along with
confirmatory qualitative analysis is needed to ensure the validity of this measure
(Henson, 2002). Roberts and Henson (2001) used confirmatory factor analysis to
examine the initial version of the TSES and did not find sufficient support for the
subconstruct Efficacy for Student Engagement (ESE). They recommended the
construct either be strengthened or eliminated. Tschannen-Moran and Woolfolk Hoy
(2001) decided to strengthen the ESE subconstruct and the final version demonstrated
30


adequate reliability and validity. Heneman, Kimball, and Milanowski (2006)
reported on a validation study of the TSES. They reported that the data produced
from the TSES was valid and reliable and extended the pool of teachers studied to
include all level of teachers. They also determined that teacher scores on the TSES
significantly predicted teacher performance ratings at the end of the school year.
Summary. In reviewing the history of teacher efficacy, it becomes evident
that several issues need to be addressed before this research field can move forward in
a coherent manner. First and foremost, agreement between the construct of teacher
efficacy and the measurement of teacher efficacy must become more highly
correlated. While the field seems to be in agreement with Banduras concept of
teacher efficacy, the majority of research has been conducted using instrumentation
that reflects the theoretical underpinnings of locus of control. Agreement on an
instrument that has strong construct validity must occur soon. The TSES seems to
have the potential to become the standard measure of teacher efficacy, but more
reliability and validity studies need to be undertaken. A second issue with this field
of study is the lack of qualitative collaboration of the research findings. Efficacy
judgments appear to be the result of cognitive processing, thus determining the role
and nature of this cognitive processing would be most easily captured though the
addition of qualitative methodologies. Henson (2002) and Labone (2004) have both
urged researchers to consider adding a qualitative approach to the study of teacher
efficacy. A mixed methods approach to research in this field appears to be a
31


pragmatic and logical choice in order to more fully capture the essence of this elusive
construct. The tables in Appendices B through F summarize the research described
above in the development of teacher efficacy instrumentation.
Research Studies in the Field of Teacher Efficacy
In order to review the relevant research in the field of teacher efficacy, a
search of various databases (ERIC, Ebsco, Infotrac, etc.) using the terms teacher
efficacy and self-efficacy was conducted. Numerous hits were produced using the
terms, but careful analysis of each study revealed a limited number of appropriate
studies. Additional studies were found through analysis of reference lists of suitable
studies. Ninety total studies were found, 27 that have been discussed above regarding
the history and instrumentation of teacher efficacy and 63 studies exploring variable
relationships with teacher efficacy, measuring change in teacher efficacy, or
exploring the influences on teacher efficacy. The 63 studies were selected based
upon strength of research methodology. Of these 63 studies, 45 were quantitative in
nature, 15 employed a type of mixed methods, and 3 were purely qualitative. Thirty
five of the studies used the TES or the STEBI-B/MTEBI, 8 used the TSES, 14 used
some different type of instrument, and 3 used an interview protocol. Of the 16 who
used a different type of instrument, the majority developed their own survey. Twenty
one of the studies examined preservice teachers, 41 examined inservice teachers, and
one study used high-school and college students. In the following sections, the major
research findings for these studies will be explored with the three categories of
32


Variable Relationships with Teacher Efficacy, Measuring Changes in Teacher
Efficacy, and Examining the Influences on Teacher Efficacy.
Variable Relationships with Teacher Efficacy
In examining the 22 studies that explored variable relationships with teacher
efficacy, three categories of variables emerged: (a) teacher behaviors, (b) school
specific variables, and (c) student achievement. Teacher behaviors alone were
explored in 13 studies, school climate in seven studies, and student achievement in
three studies. One study explored teacher behaviors and school specific variables. In
the following sections, each of these categories of variables will be examined.
Teacher behaviors. Researchers have studied the relationship between teacher
efficacy and a variety of teacher behaviors that impact the classroom environment and
have made several interesting discoveries. All of the studies involved in teacher
behaviors have either used the Rand items, the TES, or the STEBI-B/MTEBI scales.
The issues raised earlier with the GTE portion of the TES have affected the results
and their interpretation, thus, in general, only the results relevant to the PTE scale will
be discussed. The main results of these studies indicate that teachers with high levels
of PTE are more likely to embrace instructional innovation, less likely to ability
group, more likely to have clear classroom expectations, have higher achieving
students, and more likely to maintain supportive, student-centered classrooms. High
efficacy teachers are also less likely to refer a problem student to special education,
more efficient in their teaching, and more likely to believe that all students can learn.
33


High efficacy teachers have higher levels of self-esteem and lower levels of
mathematics anxiety. Each of these results will be explored in more depth in the
following sections.
In exploring the relationship between teacher efficacy and implementation of
instructional innovation, Ghaith and Yaghi (1997) conducted a study of 25 middle
and high school Lebanese teachers. The researchers used a survey instrument that
contained the 16 item version of the TES and a measure of attitude toward
implementation of instructional innovation. The study utilized correlations and /-tests
to determine strength of relationships between the variables and the existence of any
group differences. The results indicated that levels of PTE were positively correlated
with positive attitudes toward instructional innovation. The /-test comparisons
between high and low efficacy groups supported the conclusion that high efficacy
teachers rated instructional innovation as being more important that the low efficacy
teachers. While the sample size of this study was small, it does raise an important
issue to consider for future researchers to confirm the finding that teachers with high
levels of efficacy are more likely to embrace new teaching methods and techniques.
High efficacy teachers have also been shown to be less concerned with student
misbehavior, less likely to ability group, have clear expectations, and have higher
achieving students. Ashton et al. (1983) conducted a two part study to ground their
conceptual framework of teacher efficacy and to investigate the relationship between
teacher efficacy levels and various teacher characteristics. In the first study, they
34


asked teachers from two middle schools to participate in the study by first completing
a questionnaire. One middle school had a traditional junior high organization, while
the other middle school utilized a modem middle school approach. The response rate
was approximately 50% from each school, with 29 of the modem middle school
teachers and 20 of the junior high teachers participating. The questionnaire included
the two Rand items as a measure of teacher efficacy. Two teachers scoring high in
efficacy and two teachers scoring low in efficacy were then selected for interviews
and observation. The researchers discovered that low-efficacy teachers were more
concerned with student disruption and misbehavior in the classroom than the high-
efficacy teachers. Low-efficacy teachers were more likely to ability group and focus
on high achieving students than the high-efficacy teachers. High efficacy teachers
were also more likely to have clear classroom expectations and routines and were
more likely to keep students and themselves on task. The second phase of this study
involved 48 basic skills high school teachers. These participants completed a
questionnaire that contained the two Rand items as a measure of teacher efficacy.
The researchers also examined student achievement and conducted classroom
observations. They discovered that high efficacy teachers had significantly higher
achieving students than low efficacy teachers. High efficacy teachers were also more
likely to have a supportive, interactive teaching style and a supportive, student-
centered classroom. These findings led researchers to further examine how levels of
teacher efficacy may affect a teachers classroom management beliefs.
35


Several studies have been conducted relating teacher efficacy with pupil
control ideology. Pupil control ideology is rated along a continuum from custodial to
humanistic, or from highly controlled to student-centered and supportive (Woolfolk,
Rosoff, & Hoy, 1990). These studies have shown that highly efficacious teachers fall
into the humanistic part of the pupil control ideology continuum. Woolfolk et al.
conducted a study of 55 Hebrew language school teachers in order to examine the
relationship between teacher efficacy and teacher belief regarding classroom control,
management, and motivation. The teachers were asked to complete the TES, a pupil
control ideology survey, problems in school inventory, and a teacher perception of
student motivation survey. The researchers conducted a correlational analysis of the
variables and a multiple regression analysis using pupil control ideology as the
dependent variable. The researchers discovered that PTE levels were negatively
correlated with attitudes about pupil control, which indicated that high personal
efficacy teachers were less likely to maintain a highly controlled teaching
environment. The results of the multiple regression using pupil ideology as the
dependent variable and GTE, PTE, beliefs about student satisfaction, and years of
teaching experience as the independent variables was significant and accounted for
29% of the variance. Again, high levels of PTE predicted low levels of pupil control
ideology. This study contributed important information in the study of teacher
efficacy by suggesting that teachers with high levels of PTE are confident enough in
their classroom management approaches to allow more student autonomy.
36


In a related study, Woolfolk and Hoy (1990) also examined levels of teacher
efficacy, pupil control ideology, motivational orientation, and bureaucratic orientation
in a study of 182 preservice teachers. Again, using the TES as the measure of teacher
efficacy, the researchers explored the relationships between these variables using
canonical correlations and multiple regression. They began the analysis by
conducting a factor analysis of the TES. The results indicated that PTE could be
separated into two subconstructs related to responsibility for positive student
outcomes and responsibility for negative student outcomes. The results of this study
supported the previous study in that high levels of PTE were negatively correlated
with pupil control ideology. Essentially, preservice teachers with high levels of
teacher efficacy were more likely to endorse a supportive, student centered classroom
management approach. The multiple regression results indicated that the interaction
between GTE and PTE contributed to the prediction of pupil control ideology and
explained 24% of the variance. In a further examination of the results, the researchers
determined the group of teachers with high levels of GTE was more humanistic in
their pupil control ideology than teachers with low levels of GTE, but only within the
group of teachers that also possess high levels of PTE.
Malow-Iroff, OConnor, and Bisland (2004) also looked at the relationship
between teacher efficacy and pupil control ideology, but added the variables of socio-
economic status, perceived administrative support, perceived peer support, and long
term goals. In a study of 68 alternatively certified teachers, the researchers conducted
37


correlations and multiple regression to examine the relationship between the
variables. The researchers used the TES and the pupil control ideology measure for
their instrumentation. In this study, PTE was negatively correlated will pupil control
ideology which indicates that teachers with high levels of PTE would have less
controlled classrooms.
Witcher et al. (2002) explored teacher beliefs regarding classroom control and
levels of teacher efficacy. In a study of 70 preservice teachers, the researchers used
the TES and the Witcher-Travers Survey of Educational Beliefs to examine the
correlation between the two. The Witcher-Travers Survey of Educational Beliefs
measures preservice teachers preferences towards either a progressive teaching
approach or a transmissive teaching approach. The progressive teaching approach
refers to a more student-centered, inquiry-based method of teaching, while the
transmissive approach refers to a lecture-driven, teacher-dominant method of
teaching. The results of the study indicated that this sample of preservice teachers
had significantly higher levels of PTE than inservice teachers in comparable studies.
The results also indicated that teachers with lower levels of GTE preferred a
transmissive view of teaching, while teachers with higher levels of GTE preferred the
progressive method of teaching. PTE was not significantly correlated with
educational beliefs. Preservice teachers high level of personal efficacy reported in
this study is an area of concern and could be contributing to the teacher retention
problems currently facing public schools nationwide. If potential teachers are
38


entering the profession with a false impression of high levels of efficacy, they may
become disillusioned when the realities of teaching become evident in the first few
years.
Two studies explored the relationship between teacher efficacy and classroom
management and determined that high efficacy teachers tend to have student-
centered, supportive classrooms. Gordon (2001) analyzed data from 301 inservice
elementary teachers in order to identify a pool of high (N= 96) and low (N 93)
efficacy teachers based upon the results of a survey that included the TES. The
survey also examined the variables of student behaviors, teacher attribution and
affect, intervention strategies, and the pupil control ideology scale. Using a series of
ANOVAs, the researcher determined that low efficacy teachers tended to perceive
student problems as being more chronic than the high efficacy teachers. High
efficacy teachers were also less likely to become angry with or embarrassed by their
problem students. The study also confirmed earlier findings that high efficacy
teachers have lower levels of pupil control ideology, thus creating a more supportive,
student-centered classroom environment. Finally, the results indicated that low
efficacy teachers were more likely to use negative consequences and severe
punishments than high efficacy teachers. In a related study, Henson (2001b)
examined the teacher efficacy beliefs and attitudes and beliefs on classroom control
of 126 preservice teachers. Using a modified version of the TES, the researcher
reworded items on the GTE scale to more fully reflect the external locus of control
39


aspect of this scale. She renamed the GTE subscale the external attributions for
student failure (EASF) scale. Classroom management was measured using the
Attitudes and Beliefs on Classroom Control Inventory (ABCC) which has three
subscales: instructional, people, and behavior management. People are classified
onto a continuum ranging from interventionist to interactionalist to non-
interventionist. Interventionists are described as people needing the most control over
their environment. Factor analysis of the ABCC scale resulted in the removal of
items with a final result of two factors: instructional and people. Results of the
canonical correlation revealed that highly efficacious teachers were less
interventionist for instruction and people than low efficacy teachers. These studies
support the contention that high efficacy teachers are more likely to maintain a less
structured classroom that reflects the needs of the students.
Studies have also discovered that high efficacy teachers are less likely to refer
problem students to special education, more efficient in their teaching, and more
likely to believe that all students can learn. Soodak and Podell (1993) examined the
relationship between levels of teacher efficacy and special education placement
decisions. In a quantitative study of 192 inservice teachers (96 regular education and
96 special education), the participants were asked to complete the TES and to
examine one of three case studies. The case studies were of a male, second grade
student with either a learning problem, a behavior problem, or both. The teachers
were asked to state the degree to which they agreed with: (a) the student being
40


referred to special education, and (b) the student being placed in a regular classroom.
The results indicated that among teachers with high levels of PTE, regular education
teachers considered placement in a regular classroom as more appropriate than
special education teachers. The situation was the opposite for the group of teachers
with low levels of PTE. When examining the interaction between PTE and GTE, the
researchers determined that within the group of high PTE teachers, those who had
low levels of GTE considered placement in a regular classroom less appropriate than
the teachers with high levels of GTE. These results indicate that regular education
teachers with high levels of efficacy are more willing to work with students who have
problems than teachers with low levels of efficacy.
Gibson and Dembo (1984) also looked at the relationship between teacher
efficacy and classroom behaviors of teachers in their TES creation study. From the
pool of 90 participating teachers, they selected eight teachers with high and low levels
of efficacy for observation. Utilizing a use-of-time protocol, the researchers
quantitatively analyzed the teacher observations and determined that high efficacy
teachers spent less time in small group instruction, more time monitoring student
work, and more time in preparation for teaching than low efficacy teachers. When
answering students who provide an incorrect answer, low efficacy teachers
sometimes responded with criticism while high efficacy teachers never responded
with criticism. Low efficacy teachers were also less persistent with struggling
students than the high efficacy teachers.
41


Toumaki and Podell (2005) examined the relationship between teacher
predictions of student success and teacher efficacy. Using a sample of 384 teachers,
each participant was presented with one of 32 case studies that described various
student situations. The cases varied by gender, reading achievement, attentiveness,
and behavior. After reading the case, the participant was given a 9 item survey that
measured the teachers perception of academic and social success for the student in
the case study. The TES was used as the measure of teacher efficacy and reliability
and validity evidence was provided for the data set. The results of the quantitative
analysis revealed that high efficacy teachers made more positive predictions for
student academic success than low efficacy teachers.
Yilmaz-Tuzun and Topcu (2008) explored the relationship between science
teaching efficacy and the teachers epistemological world views. In this quantitative
study of 429 preservice teachers in Turkey, the researchers administered the STEBI-
B, the Schommer Epistemological Questionnaire (SEQ), and the Epistemological
World View (EWV) scale. The SEQ revealed four factors: Innate ability, Certain
knowledge, Simple knowledge, and Omniscient authority. The EWV identifies three
different world views: realist, contextualist, and relativist. The researchers ran
multiple regression statistics for each of the four factors from the SEQ and using
PSTE, STOE, and the EWV factors as the independent variables. Results revealed
that teachers with lower scores on Innate ability had high PSTE levels; high STOE
levels, and tended to be relativist in their world view. These results indicate that high
42


efficacy teachers were more likely to believe that student learning ability was not
fixed.
Research has also been conducted that explores the relationship between
teacher efficacy and other internal constructs such as self-esteem and mathematics
anxiety. This research has concluded that high efficacy teachers also have high levels
of self-esteem and low levels of mathematics anxiety. Huang, Liu, and Shiomi
(2007) explored the relationship between teacher efficacy, self-esteem, and
orientation to seek help in 218 Japanese teachers. The study had 151 preservice
teachers and 67 inservice teachers. Using the PTE scale from a Japanese version of
the TES, the participants were also asked to complete the Teachers Self-esteem Scale
(TSE) and the Orientation to Seek Help scale (OSH). In this quantitative study, the
researchers conducted correlations for the variables and one-way ANOVAs to
examine group differences. The results indicated that PTE was significantly,
positively correlated with TSE, but PTE was not correlated to OSH. When
examining the differences between preservice and inservice teachers, the researchers
determined that experienced teachers had significantly higher levels of PTE and TSE
than preservice teachers.
Swars, Daane, and Giesen (2006) examined the relationship between
mathematics anxiety and mathematics teacher efficacy. In this mixed methods study
of 28 elementary preservice teachers, the participants were asked to complete the
MTEBI and the Mathematics Anxiety Rating Scale (MARS). Quantitatively, the
43


researchers conducted Pearson product-moment correlations between the two scale
measures. Qualitatively, the researchers interviewed four participants, two with the
highest level of math anxiety and two with the lowest level of math anxiety. Results
indicated a significant negative relationship between mathematics anxiety and
mathematics teacher efficacy. Those participants with high levels of anxiety had low
levels of mathematics teacher efficacy. The results of the interviews indicated that
high anxiety respondents had negative prior classroom experiences with mathematics
while the low anxiety participants reported positive prior classroom experiences with
mathematics. Both high and low anxiety participants believed in their ability to teach
mathematics. The low anxiety preservice teachers felt confident in their strong
content knowledge, while the high anxiety preservice teachers felt they could relate to
struggling students.
These studies described above provide evidence that high efficacy teachers
have desirable characteristics that create a more supportive environment for students.
These teachers are also higher in self-esteem and lower in mathematics anxiety.
Although each of these studies used either the Rand items or a TES based instrument,
the results discussed are primarily based upon the PTE scale, which has shown a
stronger correlation to Banduras theory of self-efficacy than the GTE scale. With the
TES limitations noted, it is important for future researchers to attempt to duplicate
these teacher behavior relationships with a more suitable measure of teacher efficacy,
such as the TSES.
44


School specific variables. Several studies have explored the relationship
between school specific variables and teacher efficacy. These studies have concluded
that principal influences, perceived support, level of collaboration, and job
satisfaction with extra role duties can all impact levels of teacher efficacy. Of these
eight studies described below, seven used the TES and one used items from a survey
proposed by Riggs (1989). Due to the use of the TES, only the results from the PTE
portion of the study will be discussed.
The influence of principal behaviors on teacher efficacy was explored in four
studies. Hoy and Woolfolk (1993) examined the relationship between the school
climate factors (institutional integrity, principal influence, consideration, resource
support, morale, and academic emphasis) and teacher efficacy. Using the TES and
the Organizational Health Inventory, 179 elementary teachers were studied. In this
quantitative study, correlational analysis and multiple regression statistics were
computed. Principal influence, academic emphasis, teaching experience, and
educational level were significantly positively related to PTE. Principal influence,
academic emphasis, and educational level significantly contributed to the model
explaining PTE. Principal influence was measured as the level of perception the
teachers had that their principals exerted influence on their behalf. Academic
emphasis referred to the perception that the teachers school was academically
oriented. Educational level was measured according to extra graduate work. These
results were supported in a study by Hipp (1996) that explored the relationship
45


between principal leadership behaviors and teacher efficacy. In this mixed methods
study, 280 middle school teachers completed surveys, and 10 middle school
principals also participated through surveys and interviews. The school with the
highest level of GTE, the school with the highest level of PTE, and the school with
the lowest overall level of teacher efficacy were also selected for additional
interviews. The TES, a behavior of principals scale, and a personal data sheet
composed the survey portion of the study. The behavior of principals scale consisted
of five factors: models behavior, inspires group performance, provides contingent
rewards, holds high performance expectations, and provides support. The results
indicated that PTE had a significant positive correlation with total leadership
behaviors and with the individual factors of models behavior and provides contingent
rewards. The interview data supported these results and also showed that principals
that empowered their teachers and allowed shared decision making also contributed
to higher levels of teacher efficacy.
In response to these findings, Nir and Kranot (2006) conducted a study to
confirm that transformational leadership behaviors of principals were significantly
related to levels of teacher efficacy. They argued that the lack of control variables in
the Hipp (1996) study could lead to incorrect interpretation of the results. This study,
which had 755 inservice teacher participants from Israel, used the TES, the
Multifactor Leadership Questionnaire (MLQ), and several job-related teacher
variables. The MLQ had four factors: transformational, passive-avoidance,
46


individualized consideration, and active management by exceptions. Using the
MANOVA, ANOVA, and ANCOVA statistics, the researchers determined that
transformational leadership did have a strong effect on PTE. When the job-related
teacher variables were entered as covariates in the ANCOVA, the results indicated
there was actually an indirect relationship between transformational leadership and
PTE. The relationship was mediated by the positive levels of job experiences. In
fact, the variable of job satisfaction positively correlated with transformational
leadership. While the study does not dispute principal behaviors impact on teacher
efficacy, it does indicate the relationship may be more complicated than first thought.
This conclusion was confirmed in a study by Lee, Dedrick, and Smith (1991).
In this study, Hierarchical Linear Modeling (HLM) was conducted using a
combination of teacher efficacy and job satisfaction as the dependent variable. The
teacher level independent variable of control over classroom conditions and various
school-level independent variables were part of the model. The sample consisted of
data from 8488 teachers who had participated in a previous study. The sample
included Catholic school teachers and public school teachers. The measure of teacher
efficacy consisted of several efficacy items and job satisfaction items compiled from
the original survey. The results indicated that Catholic school teachers had higher
levels of efficacy than public school teachers. Through the HLM model, these
differences were attributed to strong principal leadership, high SES, school size, and
community involvement. The finding that strong principal leadership contributed to
47


the model was mediated by the finding that as leadership increased level of classroom
control also increased. This indicates that although a strong leader may contribute to
higher efficacy levels, it also can contribute to a teachers need for strong control of
the classroom environment. In an additional study exploring the perceived support of
administration on levels of teacher efficacy, Malow-Iroff et al. (2004) used a sample
of 68 alternatively certified teachers who completed the TES along with additional
survey questions. The researchers conducted correlations to examine the relationship
between the variables. The results indicated that the perception of support from the
administration was positively correlated with PTE. The results of these five studies
reveal some of the complications that arise in the study of teacher efficacy. Although
it appears that strong principal leadership behaviors positively correlate with levels of
teacher efficacy, more information is needed on how these behaviors impact efficacy.
Several studies have also explored the relationship between job satisfaction,
collaboration, and teacher efficacy. The findings from these studies indicate that job
satisfaction, perceived utility of supervision, and collaboration are all positively
correlated with teacher efficacy. Coladarci and Breton (1997) explored the impact of
supervision on levels of teacher efficacy for special education teachers. In this study
of 378 special education teachers, the researchers used the TES and a rating of the
frequency of and the perceived utility of administrative supervision to quantitatively
explore their relationship. Using multiple regression, the independent variables of
frequency of supervision, perceived utility of supervision, gender, age, job
48


experience, and job satisfaction significantly predicted the dependent variable of PTE.
Perceived utility of supervision and job satisfaction significantly contributed to the
equation which explained 10% of the variance. Although this is a low effect, the
findings are interesting. Job satisfaction again seems to have a role in levels of
teacher efficacy and frequency of supervision was not as important as the perceived
utility of the supervision. Somech and Drach-Zahavy (2000) also examined the
variables of job satisfaction and teacher efficacy. They explored the relationship
between these variables and extra-role behavior of teachers on a sample of 251
elementary teachers in Israel. The researchers used 11 items from a teacher efficacy
scale constructed by Riggs (1989) as a measure of teacher efficacy. Using three
factors of extra-role behavior (for the team, for the organization, and for the student),
the researchers conducted correlations that revealed significant positive relationships
between job satisfaction and all three extra-role behavior factors. Teacher efficacy
was significantly positively correlated with the extra-role behaviors for the team and
for the organization. Unfortunately, the researchers did not explore the relationship
between job satisfaction and teacher efficacy, thus that valuable information was lost.
In a study examining the relationship between teacher efficacy and
organizational commitment, Reames and Spencer (1998) utilized a mixed methods
approach with a sample of 280 middle school teachers. The teachers completed a
survey that included the TES, a school work culture profile, and the Organizational
Commitment Questionnaire (OCQ). Using canonical correlations and focus group
49


interviews, the researchers determined that collaboration had a direct positive effect
on levels of PTE. Through the focus group sessions, the researchers also determined
that providing teachers with learning opportunities in the form of staff development
and school goals and planning also contribute to higher levels of PTE. These results
indicate that teachers feel more efficacious when provided opportunities to extend
their teaching practices with the help of other teachers and through professional
development.
The results of these eight studies reveal the school specific variables of
principal behaviors, principal support, job satisfaction, collaboration with other
teachers, and learning opportunities all seem to have a positive relationship with
teacher efficacy. Supporting teachers through a transformational leadership style that
encourages collaboration and increasing teaching skills seems to relate to higher
levels of personal teaching efficacy. Additional research should be conducted using
these variables and the TSES in order to more fully develop how these factors can
impact teacher efficacy.
Student achievement. Three studies explored the relationship between teacher
efficacy and student achievement. These studies all support the conclusion that
teachers with higher levels of efficacy seem to produce higher achieving students.
Rose and Medway (1981a) used the TLC scale with a sample of 17 teachers and their
students. The TLC measures efficacy through an internal versus external orientation.
High internal teachers attribute consequences to their own actions, and high external
50


teachers attribute consequences to factors outside their control. Using scores from the
TLC measure, teacher observations, and student achievement data, the researchers
determined that teachers with high levels of internal attribution did have higher
achieving students. Ashton et al. (1983) also conducted a study to investigate the
relationship between teacher efficacy levels and student achievement. Using a
sample of 48 basic skills high school teachers, the researchers explored the
relationship using student achievement data, teacher efficacy scores from the two
Rand items, and classroom observations. The results of the study indicated that levels
of teaching efficacy were significantly positively correlated to student achievement in
mathematics and language arts. Ross (1992) also found a positive relationship
between student achievement and teacher efficacy. In a study exploring the effect of
coaching on teacher efficacy and student achievement, the researcher used 18 history
teachers and 6 instructional coaches and tracked student achievement as they
implemented new curriculum. Using the TES, student tests, a use of coaching survey,
and interviews, the results indicated that student achievement increased significantly
from pretest to posttest. Student achievement was significantly positively correlated
to teacher efficacy. In a regression analysis, PTE and self-report of coaching
contributed significantly to the equation predicting student achievement. The
equation predicted 57% of the variance in student achievement. The results of these
three studies all demonstrate a strong positive relationship between student
achievement and teacher efficacy, but more research is needed that not only uses the
51


TSES, but also more deeply explores how high levels of teacher efficacy impact the
achievement of students.
Summary. A summary of these studies is provided in in Appendix G. These
23 studies exploring the relationship between differing variables and teacher efficacy
provide valuable information on teacher characteristics, school related variables, and
student achievement. These studies reveal that high efficacy teachers have desirable
characteristics that create a more supportive environment for students, have higher
levels of self-esteem, and lower levels of mathematics anxiety. High personal
teaching efficacy was also related to school specific variables of principal behaviors,
principal support, job satisfaction, collaboration with other teachers, and learning
opportunities. Specifically, the transformational leadership style, which encourages
collaboration, and provides opportunities for teachers to work on their teaching skills
seem to relate to higher levels of personal teaching efficacy. Some of these studies
also revealed a strong positive relationship between student achievement and teacher
efficacy, with high efficacy teachers contributing to higher student achievement.
Since all of these studies used the TES, a variation of the TES, or a locus of control
measure, it is important for future researchers to attempt to duplicate these teacher
behavior relationships with a more suitable measure of teacher efficacy, such as the
TSES, in order to more fully determine the nature of the relationship between these
variables and teacher efficacy. If high efficacy teachers have a positive impact on
student achievement, then it is essential for teacher education programs and
52


professional development programs to understand how to influence levels of efficacy
in teachers.
Measuring Changes in Teacher Efficacy
In examining the 22 studies that measured changes in levels of teacher
efficacy, three categories of variables emerged: (a) professional development, (b)
teacher preparation and induction, and (c) score differences. The impact of
professional development programs on levels of efficacy were explored in 10 studies,
the impact of teacher preparation and induction programs on levels of efficacy were
explored in nine studies, and an examination of score differences were explored in
three studies. In the following sections, each of these categories of variables will be
examined.
Professional development. Several researchers have attempted to measure
teacher efficacy levels for inservice teachers before and after participating in a
professional development program. Although prior thought on affecting change in
efficacy levels has been that experienced teachers levels of efficacy should be less
malleable than novice teachers levels of efficacy (Bandura, 1997; Tschannen-Moran
et al., 1998), the results of these studies indicate that, in some cases, experienced
teachers can achieve higher levels of efficacy through teacher training. Ross and
Bruce (2007) conducted a professional development program designed to increase
mathematics teachers levels of efficacy. Participants included 106 sixth grade
teachers in Canada who attended a full day session and then three 2 hour sessions
53


conducted at the end of a school day. The professional development program was
specifically designed with Banduras (1997) four sources of efficacy: mastery
experiences, vicarious experiences, social persuasion, and affective thoughts. Using a
pretest/posttest experimental design, the participants were placed in treatment and
control groups. The participants were asked to complete the TSES before and after
the program. The control group was given the treatment after the initial treatment
group was finished. The participants were broken up into three groups in order to
facilitate learning. Instructors for the sessions worked to model standards-based
teaching practices, which the participants were then asked to apply in their own
classroom. In the next session, the teachers would discuss their experiences. The
instructors also worked with the teachers on redefining success by examining the
conceptual understanding obtained by their students and each students ability to
communicate mathematically (Ross & Bruce, 2007, p. 54). Using the MANCOVA
statistic with the posttest scores on the three dimensions of the TSES (efficacy for
instructional practices, efficacy for student engagement, and efficacy for classroom
management) as the dependent variables, the pretest scores as covariates, and the
treatment condition as the independent variable, the results indicated a non-significant
effect for the treatment. Since the number of participants was small, the univariate
effects were then examined. These results indicated that teachers scored higher from
pretest to posttest, but only significantly for efficacy for classroom management.
Unfortunately, this is the only study that has utilized the TSES for examining changes
54


in efficacy due to professional development. Additional research is warranted in
order to compare results and to examine causality.
Puchner and Taylor (2006) used a case study approach to determine the
impact on efficacy levels for 4 elementary teachers as they learned to implement the
Japanese model of lesson study. In this model of collaborative teaching, a group of
teachers come together to create a series of teaching lessons around one central
learning goal. The teachers observe each other teaching the lesson and then discuss
strategies for increasing student learning. The 4 participating teachers in this case
study were relatively new teachers who came together to enhance a mathematics
lesson on problem solving. Through interviews and observations of the meetings, the
researchers concluded that the lesson study process had a significant impact on levels
of teacher efficacy. Each of the teachers expressed feelings of increased competence
and the recognition that hard work paid off with student learning. The limitations of
this study include the lack of a quantifiable pre and post measure of efficacy levels,
and the number of participants. The rigor of the qualitative analysis offsets the
limitations somewhat and thus the study does contribute to the idea that efficacy
levels can be increased through teacher training.
Fritz, Miller-Heyl, Kreutzer, and MacPhee (1995) conducted a study to
determine the effect of teacher training on teacher efficacy. Utilizing a pretest/
posttest/delayed posttest experimental design, the researchers used the TES, the Self-
Perceptions of the Teacher Role (SPTR) survey, and a rating of the teacher use of
55


inservice activities, to measure the impact of the training and amount of use on levels
of efficacy. The study had 241 total participants, 130 in the treatment group and 111
in the control group. The training program was based upon the DARE to Be You
(DTBY) program that focuses on increasing the self-esteem, internal locus of control,
and communication skills of teachers. The training was offered for 20 to 24 hours
over a short period of time. The participating teachers were provided with activities,
teaching materials and teaching curriculum for K-8. The researchers conducted a
repeated-measures MANOVA using the pretest and posttest as the dependent
variables, along with follow-up ANOVAs to examine the impact of the independent
variable of use of materials on teacher efficacy. The results indicated treatment
teachers who had a high usage of the DTBY materials had significantly higher levels
of efficacy than low usage participating teachers or the control teachers. These
results indicate that investment in the teacher training may be an essential ingredient
in increasing levels of teacher efficacy.
Swackhamer, Koellner, Basile, and Kimbrough (in press) examined the
impact of increasing content knowledge in mathematics or science on the teacher
efficacy levels of inservice teachers. The 88 participating teachers had taken
anywhere from one course to nine courses over a period of 2 years. The courses were
offered primarily to middle school teachers who needed additional coursework to
achieve highly qualified status in mathematics and science, but teachers from all
levels were also encouraged to enroll. The courses were designed to not only
56


increase content knowledge, but also increase pedagogical content knowledge, and to
provide the teachers with hands-on activities and inquiry based teaching techniques.
Using a modified version of the STEBI-B to include mathematics and science, the
participants were asked to complete the survey at one point in time. Independent
samples /-tests were conducted to compare the mean scores on the PTE and GTE
scales with the group of teachers who had taken one to three courses and the group of
teachers who had taken four or more courses. The results indicated the participants
all had high levels of PTE, but GTE levels were significantly higher in the four or
more course group than the one to three course group. This higher level of GTE
seems to indicate that as the teachers gained additional content knowledge, they
became more confident in teaching students of all backgrounds and ability levels.
This study adds to the conclusion that increasing content knowledge can increase
efficacy as long as participant involvement is high.
Two additional studies were conducted with the intent of determining whether
professional development can impact efficacy beliefs. The main researcher for both
studies utilized his own survey instrument and did not discuss the reliability and
validity of the data. The results were also reported as a series of /-tests without
mention of any Bonferroni correction. Therefore, the results should be interpreted
with caution, but are included here due to the construction of the professional
development programs. Sottile, Carter, and Watson (2001) conducted a half-day
professional development program for 42 middle school teachers. The professional
57


development consisted of a demonstration of a hands-on science lesson to both
students and teachers. Sottile, Carter, and Murphy (2002) examined 44 teachers as
they participated in one of two courses. The first course was Physics with Toys and
met every week day for 3 weeks. The second course was Integrated Science and
Math Methods and met every Saturday for 8 weeks. Both courses were designed to
increase levels of content knowledge and confidence to teach by using hands-on
methods. The results of both studies indicated an increase in teacher efficacy when
comparing pretest scores to posttest scores. Although the rigor of the analysis and of
the data collection measure should have been stronger, when compared with the other
studies, the results do tend to confirm the conclusion that professional development
that focuses on increasing content knowledge or teaching skills can lead to increases
in efficacy levels.
Henson (2001a) conducted a year-long teacher research professional
development study with 8 teachers and 3 instructional assistants. The purpose of the
professional development was to teach the participants how to conduct participatory
teacher research in order to facilitate feelings of empowerment and efficacy. Using
the TES, and other instruments measuring school related variables, empowerment,
level of implementation, and level of collaboration, the researcher employed a mixed
methods approach. A repeated measures ANOVA was computed along with analysis
of interviews and field notes. With the limited number of participants, the
quantitative results alone would not have sufficient power to determine significant
58


increases in efficacy levels, but did reveal significant differences in PTE and GTE
from pretest to posttest. The interview data supported these results, but also indicated
that less experienced teachers were more impacted by the professional development
than experienced teachers. While these results support previous conclusions that
impacting efficacy levels in experienced teachers is difficult (Tschannen-Moran et al.,
1998) the small number of participants limits the generalizability of the results.
Additional research is warranted in order to explore the effects of professional
development on experienced teachers using the TSES as a measure of efficacy.
Two studies have demonstrated non-significant differences in levels of teacher
efficacy after participation in a professional development program. Ross (1994)
conducted a professional development program designed to increase knowledge and
skills related to cooperative learning techniques. The program was also initiated to
help teachers transition their teaching after a government mandated order to eliminate
ability grouping of students. The professional development occurred over one school
year and consisted of three 2-1/2 hour sessions and then monthly sessions with each
school district after the school day had ended. In this mixed methods study, 50
middle school teachers completed the TES and several other surveys designed to
measure levels of cooperative learning. Field notes from each of the meetings were
also analyzed qualitatively. The results of the repeated measures ANOVA indicated
that PTE increased over time but not at a statistically significant level. The
researchers noted that this professional development program did not provide any
59


hands-on activities or opportunities for the teachers to practice and reflect upon their
teaching practices. Those types of activities may have helped impact teacher efficacy
levels. Additional research is needed on the exact components of a professional
development program in order to increase the chance for attaining higher efficacy
levels.
Tucker et al. (2005) also determined that levels of teacher efficacy did not
change significantly after a professional development program. Their program was
designed to increase awareness of teaching skills and methods for working with
culturally diverse students. The program consisted of a 6 hour workshop that
emphasized teaching methods and strategies that have been shown to increase student
engagement and achievement in this population of students. From a population of 62
elementary teachers, 37 were in the treatment group and 25 were in the control group.
The participants were asked to complete a survey (pretest and posttest) containing
teacher efficacy items and culturally sensitive self-efficacy items. The researchers
conducted an ANCOVA using the posttest scores as the dependent variable and the
pretest scores as the covariate. The results indicated that the participating teachers
did not have significantly higher levels of teacher efficacy than the control group, but
did have significantly higher levels of culturally sensitive self-efficacy. These results
are not surprising since the focus of the professional development program was not to
increase teacher efficacy, but to increase cultural awareness in teachers. These two
60


studies demonstrate that if the goal of professional development is to increase teacher
efficacy, efficacy building components need to be instilled throughout the program.
Roberts, Henson, Tharp, and Moreno (2001) examined whether length of time
of a professional development program contributed to changes in teacher efficacy.
The researchers examined data from over 330 participants from seven separate
federally funded professional development programs that varied in length of time
from 2 weeks to 6 weeks. Using the PSTE scale from the STEBI-B, the researchers
explored the differences in the groups. They discovered little differences in the
participants scoring high in levels of PSTE, thus they explored the group of
participants who scored below the mode on the PSTE. When analyzing this group of
participants, no significant differences were found between the 2 week participants
and the 3 week participants, or between the 4 week participants and the 6 week
participants. There were significant differences between the 2 to 3 week participants
and the 4 to 6 week participants. The low PSTE group that participated in a 4 or 6
week program had significantly higher levels of PSTE than the participants in a 2 or 3
week program. These results indicate that a 4 week professional development
program may be the optimal length to increase low efficacy participants. This study
provided an important point to consider as teachers with initial high levels of efficacy
can face a ceiling effect which can limit the amount of increase possible. It may be
more beneficial for the field of teacher efficacy to focus on the population of teachers
with initial low levels of efficacy and track their changes.
61


The results of these ten studies provide information that levels of teacher
efficacy may be increased through the thoughtful creation of a professional
development program. While additional research is needed in order to corroborate
these results using the TSES, these initial results seem to indicate that a professional
development program that emphasizes content knowledge, hands-on activities, and
reflective feedback may help increase levels of efficacy for the participating teachers.
In examining length of time of the program, longer seems to be more effective than
shorter, but the construct of program would seem to be the primary factor.
Teacher preparation and induction. Researchers have also explored the
impact of teacher preparation programs and novice teacher induction on levels of
teacher efficacy. These studies have explored the impact of student teaching on
levels of efficacy, the durability and changes from methods classes to student
teaching to first year of teaching, and instruction method of teacher preparation. The
results of these studies indicate that teacher efficacy levels generally increase during
methods courses and student teaching, although a few studies have found some
variability in the pattern.
Volkman, Scheffler, and Dana (1992) examined differences in two groups of
preservice elementary teachers during their student teaching experiences. The first
group of 12 preservice teachers was placed in an environment that encouraged
reflective practices. They were given individualized meeting time with a supervisory
teacher and met biweekly with the other students in the treatment group for more
62


reflective practice activities. The remaining twelve preservice teachers were place in
one of two control schools with no emphasis on reflective practices. Each of the 24
participants was also asked to complete a reflective journal of their experiences
during this time period. The participants were also asked to complete the TES prior
to student teaching and after student teaching. Using the ANCOVA statistic with
posttest TES scores as the dependent variable, pretest TES scores as the covariate,
and treatment group as the independent variable, the results indicated the treatment
group had significantly higher levels of teacher efficacy than the control group. The
qualitative analysis of the journals supported the results and indicated the treatment
groups entries were more reflective and insightful in general.
Fortman and Pontius (2000) also explored the impact of student teaching on
100 preservice teachers. Although the teachers were placed in different settings, no
treatment or control groups were established. The students were asked to complete a
modified version of the TES prior to student teaching, at the mid point of student
teaching, and at the end of student teaching. Using a paired /-test, PTE posttest scores
were significantly higher than pretest scores. Scores on GTE were lower from pretest
to posttest, but not significantly. The results also indicated that females scored
significantly higher than males on both pretest and posttest PTE scores. The results
of these two studies indicate that student teaching had a positive impact on levels of
efficacy. Several other studies have explored the trend of efficacy levels from
63


methods courses to student teaching and beyond and have found somewhat
conflicting results.
Utley, Bryant, and Moseley (2005) explored differences in preservice
teachers levels of efficacy after completing a math and science methods course and
completing student teaching. Using the STEBI-B and the MTEBI, the 51
participating students were asked to complete the surveys prior to the methods course,
after taking the methods course, and after student teaching. The results of the trend
analysis revealed a statistically significant quadratic trend for PSTE and PMTE.
Scores increased from the beginning of the methods course until the end of the
methods course, but then decreased after student teaching. In a related study,
Woolfolk Hoy and Spero (2005) tracked levels of teacher efficacy in 53 preservice
teachers from the beginning of their teacher preparation, the end of student teaching,
and then to the end of their first year of teaching. The participants were asked to
complete three different measures of teacher efficacy, the TES, Banduras scale, and
a newly developed Ohio State Teaching Confidence Scale. Using repeated measures
ANOVA, the results indicated that levels of teacher efficacy increased from the
beginning of teacher preparation to the end of student teaching on all three measures.
From the end of student teaching until the end of the first year of teaching, efficacy
levels significantly decreased on Banduras scale and the GTE scale. On the PTE
scale levels decreased and approached significance. The OSU scale demonstrated
relatively stable scores.
64


Palmer (2006) examined efficacy levels of 55 preservice teachers from the
beginning of a science methods course, the end of the course, and at the end of the
program. The researcher used the STEBI-B and interview data from 18 of the
participants. A repeated-measures ANOVA was computed and revealed that teacher
efficacy increased significantly from the beginning of the methods course until the
end of the methods course. There were no significant differences between the
posttest scores and the delayed posttest scores administered at the end of the program,
thus the gains in efficacy remained stable across time and student teaching. The
interviews supported the results and indicated that all of the participants felt positive
about teaching science.
Fives, Hamman, and Olivarez (2007) examined the trend in teacher efficacy
during student teaching and whether teacher burnout characteristics could be revealed
as early as student teaching. Using the TSES and other surveys measuring teacher
burnout and learning questionnaires, the 49 participants were asked to complete the
questionnaires before and after student teaching. Using a MANOVA with the three
factors from the TSES (efficacy for instructional practices, efficacy for student
engagement, and efficacy for classroom management) as the dependent variables, the
results indicated that all three factors showed significant increases from pretest to
posttest.
Chester and Beaudin (1996) examined a group of 173 newly hired teachers in
urban schools in order to explore how efficacy levels are impacted by various school
65


and personal factors. This group of teachers included novice teachers and teachers
newly hired to a school. The researchers used the TES and administered a pretest at
the beginning of the school year and a posttest at the end of the school year. They
explored independent variables of teacher placement, opportunity to collaborate,
attention supervisors pay to performance, and the availability and quality of
resources. The regression results indicate that teacher efficacy increased in older
novice teachers but decreased in younger novice teachers. Efficacy levels tended to
decline for all experienced newly hired teachers. Teacher efficacy increased in all
teachers when they reported high levels of collaboration and when they reported a
high number of observations.
These studies reveal conflicting results regarding levels of efficacy and the
impact of teacher preparation and the first year of teaching. These results
demonstrate the variability inherent in a self-reported construct and reveal the need
for additional research into how efficacy levels are attained and maintained
throughout teacher preparation and into the teaching career. Variables such as
gender, age, collaboration, and teacher support were demonstrated to affect efficacy
levels, but more research is needed to determine the degree of impact.
Chambers (2003) examined whether length of time of student teaching can
produce changes in teacher efficacy. Using 28 preservice teachers enrolled in a two
semester student teaching program and 27 preservice teachers enrolled in a one
semester student teaching program, she examined group differences on the TES. A t-
66


test statistic and multiple regression statistic were computed, but revealed no
differences between the two groups on level of efficacy. These results indicate length
of time in student teaching may not create any differences in levels of teacher
efficacy.
Finally, Gorrell and Capron (1987) examined a group of 86 preservice
teachers with low to moderate levels of teacher efficacy to determine if different
instructional methods could create different levels of efficacy. Preservice teachers
were originally given the Potential Teachers Attitude Questionnaire (PTAQ) as a
measure of efficacy. Low scoring students and moderate scoring students were
placed into two groups. The first group received direct instruction of a teaching task,
followed by a student teacher demonstrating how to employ the task with a student
and discussing the direct instructions. The second group received cognitive modeling
instruction for completing the teaching task, followed by a student teacher
demonstrating the task with a student, but employing cognitive modeling thought
processes. The preservice teachers were then given the PTAQ again and scores were
compared. The results indicated the low efficacy group/direct instruction group had
higher levels of efficacy than the low efficacy/cognitive modeling group. The
moderate efficacy/cognitive modeling group had higher levels of efficacy than the
moderate efficacy/direct instruction group. This study indicates that in order to raise
efficacy levels in low efficacy teachers, a direct instruction approach might be more
effective.
67


These studies indicate that the pattern of efficacy seems to increase during
teacher preparation, but in some cases it may decrease during student teaching. Some
studies have also shown that efficacy levels may decrease during the first year of
teaching, but other factors may help mediate that decline. Much additional research
is needed that explores these patterns of efficacy development in order to more fully
understand the conditions that heighten efficacy levels and the conditions that
dampen efficacy levels. Researchers should also consider examining changes in
efficacy scores for initial low to moderate scoring teachers, as the initial high scoring
teachers should not have as much variability. The studies also give some guidance in
the length of student teaching and teaching methods for low efficacy student teachers,
but again, additional research is needed using a common instrument to confirm or
dispute these findings.
Score differences. Three research studies have explored group differences in
efficacy scores. One study examines score differences between a typical pretest and a
retrospective pretest, and two studies explored cultural differences between different
countries on efficacy scores. Cantrell (2003) examined whether traditional pretest
scores may differ from a retrospective pretest score on levels of teacher efficacy. The
study consisted of a group of 37 preservice teachers: 24 from the treatment group of
an elementary science course, and 13 in a treatment group enrolled in a combination
science methods and practicum course. The two groups were given the STEBI-B
before beginning coursework and then again at the end of the coursework. They were
68


then asked to complete the STEBI-B as a retrospective pretest (answer as they now
think they should when they entered the course). Eight students were then selected to
interview to explore differences between the two pretest scores. Paired samples t-
tests were conducted comparing the traditional pretest to the retrospective pretest, and
the two pretests to the posttest. Results indicated that PSTE scores were significantly
higher for the posttest than either pretest. The traditional pretest was also
significantly higher than the retrospective pretest. For the STOE scale the
retrospective pretest was significantly lower than the posttest. The interviews
supported the finding that students rated themselves higher on the traditional pretest
than the retrospective pretest. The students consistently commented on how little
they actually knew compared to how much they thought they knew. This study is a
very important finding for teacher efficacy research. Much of the research that has
been conducted on preservice teachers or novice teachers requires them to make an
efficacy judgment that may be based upon faulty information. Much of the variation
in efficacy levels may actually be attributed to an inaccurate valuation of their
beginning level of efficacy. More researchers should consider the use of a
retrospective pretest when evaluating preservice or novice teachers.
Two studies were conducted that compared the factor structure and levels of
efficacy between teachers in different countries. Cakiroglu, Cakiroglu, and Boone
(2005) compared levels of teacher efficacy between 100 Turkish preservice teachers
and 79 American preservice teachers. Using the STEBI-B, the researchers found that
69


the two groups had similar scoring patterns and levels, but American preservice
teachers had significantly higher levels of PSTE than the Turkish teachers. Although
the effect size was small, it again raises the issue of whether American preservice
teachers as a group are overconfident in their personal teaching efficacy. The second
study was conducted by Ho and Hau (2004) and examined the cross-cultural
similarities and differences between the factor structure of teacher efficacy for 273
Chinese teachers and 208 Australian teachers. The researchers used a revised version
of the TES with three factors: external influences, personal instruction, and personal
guidance. Goodness of fit indexes showed no significant differences between the
factor structures of the two countries, but a few small differences were noted. In
Chinese teachers personal instruction and personal guidance were highly correlated,
but only moderately so in Australian teachers. In Chinese teachers external
influences and personal guidance were negatively correlated, but were not correlated
at all in Australian teachers. Australian teachers also had significantly higher levels
of PTE (for both instruction and guidance) than the Chinese teachers. There were no
significant differences in external influences. The similarities between teachers in
western cultures have been recognized, thus the high levels of personal efficacy may
be a result of cultural norms.
Summary. A summary of these studies is provided in the table in Appendix H.
The results of these 22 studies provide information that levels of teacher efficacy may
be increased through professional development programs and through teacher
70


preparation. While additional research is needed in order to corroborate these results
using a standard measure, such as the TSES, and through more qualitative research,
these initial results seem to indicate that a professional development program that
emphasizes content knowledge, hands-on activities, and reflective feedback may help
increase levels of efficacy for the participating teachers. These studies also reveal
that much more research is needed into how efficacy is developed and maintained in
preservice and novice teachers. Variables such as gender, age, collaboration, and
teacher support have shown a relationship to teacher efficacy, but more information is
needed to fully understand all of the complexities. Research has also shown that
teacher efficacy seems to be a global construct, with a few differences between
Eastern and Western cultures. It appears that Western cultures may have higher
levels of personal efficacy than other countries, but more research is needed to
determine if this is true measure or a cultural over-evaluation. The research on the
use of a retrospective pretest also raises important issues for the field of teacher
efficacy. More studies using a retrospective and traditional pretest would provide
valuable information on how preservice teachers evaluate their efficacy beliefs.
Examining the Influences on Teacher Efficacy
In examining the 18 studies that explored various influences on teacher
efficacy, four types of studies emerged: exploring Banduras sources of teacher
efficacy, personal variables, school related variables, and the impact of teacher
efficacy on other variables. Banduras sources of efficacy were explored in six
71


studies, personal variables were explored in four studies, school related variables
were explored in five studies, and the effect of teacher efficacy on other variables was
explored in three studies. In the following sections, each of these categories of
studies will be explored in depth.
Exploring Banduras sources of teacher efficacy. These six studies explored
the relationship between Banduras (1993) four sources of self-efficacy (mastery
experiences, vicarious experiences, social persuasion, and affective states) and levels
of efficacy. Lent, Lopez, Brown, and Gore (1996) developed the Perceived Sources
of Math Self-efficacy Inventory (PSMSI) in order to explore the latent structure of
these sources on mathematics self-efficacy. In a study of 295 university students and
481 high school students, the researchers used a mathematics self-efficacy survey and
the PSMSI to explore how the four sources work together to impact levels of
mathematics self-efficacy. Reliability and validity evidence for the surveys was
reported and Banduras four sources from the PSMSI were supported through
Confirmatory Factor Analysis. A path analysis revealed a hierarchical structure to the
model of mathematics self-efficacy with a few changes in Banduras sources. Direct
experiences with mathematics appeared to have a direct impact on personal
performance, social persuasion, and emotional arousal. While this study did not
examine teacher efficacy, the results give researchers some important information
regarding mathematics self-efficacy, specifically that direct experiences with
mathematics may influence all other sources of efficacy.
72


This finding was supported in a study by Swars (2005) and a study by
Charalambous, Philippou, and Kyriakides (2008). Swars had a group of students
taking an elementary mathematics methods course complete the MTEBI. The 2
students scoring the highest and the 2 students scoring the lowest were selected to
interview in order to explore the differences between the two groups. From the
qualitative analysis past experiences with mathematics appeared to have a direct
influence on the students perceptions of mathematics teaching effectiveness. The 2
high efficacy students reported positive past experiences with mathematics, while the
2 low efficacy students reported negative past experiences with mathematics. These
past experiences influenced their perceptions of mathematics teaching effectiveness
with the low efficacy students acknowledging that they would have to work longer
and harder to be effective mathematics instructors and the high efficacy students
stressing their strong mathematics content knowledge as an asset to becoming an
effective mathematics instructor.
Charalambous et al. (2008) adapted the TSES to reflect mathematics teacher
efficacy in a study of 89 preservice teachers in Cyprus. Exploratory factor analysis
revealed a two factor solution that explained 60% of the variance in mathematics
teaching efficacy. The two factors were labeled efficacy for mathematics instruction
and efficacy for mathematics classroom management. Results indicated that levels of
mathematics teaching efficacy increased over the period of student teaching, but a
MANOVA statistic revealed that levels of teacher efficacy did not increase
73


uniformly. The researchers then selected eight students scoring high and low on
levels of efficacy for a series of three interviews. These interviews revealed that
efficacy for mathematics instruction was influenced highly by prior experiences with
mathematics and by mastery experiences during student teaching. Levels of efficacy
were also impacted by mentor support and by the feedback provided by the mentors.
These two studies support the conclusion that for mathematics efficacy, past
experiences with mathematics may have the strongest impact of any of the influences.
Two additional studies determined that mastery experiences and social
persuasion were the strongest of Banduras influences, while one determined that
other sources may actually have a stronger influence. Poulou (2007) developed a
Teaching Efficacy Sources Inventory (TESI) that expanded on Banduras four
sources. Factor analysis revealed seven factors for the instrument: motivation,
personality characteristics, capabilities and skills, teacher training, mastery
experiences with social persuasion, vicarious experiences, and physical and affective
states. It could be argued the four new sources are actually subsources of Banduras
original four sources. This inventory, along with the TSES, was given to 198
preservice teachers in Greece. Reliability and validity evidence for both surveys was
provided in the study. The results revealed that motivation, personality
characteristics, and mastery experiences with social persuasion all received high
mean scores from the students. Vicarious experiences and physical/affective states
received the lowest mean scores. Multiple regression revealed that personality
74


characteristics and capabilities/skills were significant predictors for all three factors of
the TSES (efficacy for instructional strategies, efficacy for classroom management,
and efficacy for student engagement). Motivation was also a significant predictor for
efficacy for student engagement. Computing another multiple regression using only
Banduras sources revealed that mastery experiences with social persuasion
significantly predicted all three factors of the TSES. This study reveals the need for
future research into the influences on teacher efficacy in order to more fully
understand all of the properties and components of each influence.
Mulholland and Wallace (2001) conducted a case study of one Australian
elementary teacher as she transitioned from a preservice teacher into an inservice
teacher in order to explore possible influences on her level of science teaching
efficacy. Through interviews and observations, the researchers determined that
mastery experiences, in the form of successful and unsuccessful lessons, had the
greatest impact on her perceived high level of science teaching efficacy. Social
persuasion from other teachers and from her students also impacted teacher efficacy
levels. Vicarious experiences appeared to have a slightly negative impact on her level
of science teaching efficacy due to the low status of science and the inexperience in
teaching science of her fellow teachers. No evidence of an impact from affective
states was found. This study reinforces the need for additional research into the
strength of influence on teacher efficacy for the proposed sources of vicarious
experiences and affective states.
75


Philippou and Charalambous (2005) explored the influence of social
persuasion on teaching efficacy in the form of mentor relationships with preservice
teachers. Using the same pool of preservice teachers described in Charalambous et
al. (2008), the researchers determined through a qualitative analysis of interview data
that mentors have a strong impact on efficacy levels through three avenues: teaching
style, feedback, and latent messages. Mentors who provided a good model of
organization and execution of the teaching task helped the student teachers develop
higher levels of teaching efficacy. Feedback that was supportive and constructive
also seemed to contribute to higher levels of efficacy. Finally, the mentor teachers
also influenced efficacy levels through latent messages. A sometimes subtle
disapproval of the student teachers decision making led to feelings of doubt
regarding teaching ability. This study reveals the need for appropriate mentor
training in order to provide these experienced teachers with the necessary skills to
help support efficacy building in novice teachers.
These studies highlight the need for additional research into Banduras
sources of efficacy, and the identification of other possible sources that may have an
even greater impact on efficacy building. This future research should combine
quantitative and qualitative methodologies in order to more effectively identify how
levels of teaching efficacy are developed and how they continue to be impacted in
experienced teachers. It is also evident that mathematics specific teaching efficacy
may be strongly impacted by prior experiences with mathematics, thus more research
76


is needed on how to help teachers with negative past experiences develop higher
levels of mathematics teacher efficacy.
Influence of personal variables. The four studies that explored personal
variables determined that teacher perception of student engagement, avoiding
negative thoughts, emotional exhaustion, and extraversion significantly predicted
levels of teacher efficacy. Ross, Cousins, and Gadalla (1996) examined the
variability of teacher efficacy for within teacher behaviors and between teacher
behaviors. Using a sample of 52 Canadian secondary teachers, the researchers
explored the within teacher variables of feelings of past success, feelings of being
well prepared, and perception of student engagement. The between teacher variables
of subject discipline, teaching experience, educational level, and gender were also
explored. Using a survey that contained one of the Rand items as a measure of
teacher efficacy, the researchers conducted a hierarchical linear model (HLM) that
attempted to predict teacher efficacy. Only teacher perception of student engagement
contributed significantly to the model. Student engagement is one of the factors of
teacher efficacy as identified in the TSES, thus it is not surprising that this would
contribute significantly to levels of teacher efficacy. The researchers choice of HLM
for the statistical analysis was probably not appropriate since they only had 52
participants. HLM generally requires large sample sizes for adequate power. Thus,
the results of the study may not be entirely accurate.
77


Ross, McKeiver, and Hogaboam-Gray (1997) conducted a case study of 4
mathematics teachers during the first year of a policy change that ended the practice
of ability grouping in Canada. The researchers examined changes in efficacy and the
factors that contributed to these changes. The 5 teachers were interviewed five times
over the course of the year and participated in a 2 hour focus group. The researchers
also conducted three teaching observations for each participant. Results of the
qualitative analysis revealed that levels of teacher efficacy decreased dramatically at
the beginning of the change year. These levels began to increase as the teachers
began to gather evidence that all students were achieving. The results indicated that
collaborating with other teachers and avoiding negative thoughts were the most
significant influences affecting the increase in efficacy levels.
Brouwers and Tomic (2000) studied the influence of teacher burnout on levels
of teacher efficacy. Using the Emmer and Hickman (1991) Self-efficacy for
Classroom Management and Discipline scale as a measure of teacher efficacy and the
Maslach Burnout Inventory (MBI) as a measure of teacher burnout, the researchers
used a sample of 243 secondary teachers in the Netherlands. The MBI contains 20
items and revealed three subscales: Emotional exhaustion, Depersonalization, and
Personal accomplishment. Using Structural Equation Modeling, the researchers
tested various models of these three factors and teacher efficacy. The best fit model
revealed that emotional exhaustion had a direct effect on teacher efficacy and teacher
efficacy in turn had a direct effect on personal accomplishment and a direct
78


longitudinal effect on depersonalization. These results indicate that levels of
emotional exhaustion have a direct influence on teacher efficacy.
Henson and Chambers (2002) explored whether personality type can be an
influence on levels of teacher efficacy. In a sample of 120 preservice secondary
teachers, the researchers used the TES, the Myers-Briggs Type Indicator (MBTI), and
the Attitudes and Beliefs on Classroom Control Inventory (ABCC) to test the level of
influence. The MBTI has four continuums of personality types: extraversion-
introversion, sensing-intuition, thinking-feeling, and judging-perceiving. Factor
analysis of the TES and the ABCC revealed one factor of personal teaching efficacy
for the TES and two factors of instructional management and people management for
the ABCC. Using canonical correlation analysis, the results indicated that the
personality type of extraversion-introversion was a weak influence on teacher
efficacy. Analysis revealed that extraversion was positively related to levels of PTE.
The results of these four studies show that teacher perception of student
engagement, avoiding negative thoughts, emotional exhaustion, and extraversion
have varying levels of influence on teacher efficacy. These four sources appear to be
a component of Banduras source of physiological and affective states. Additional
research into these and other internal, personal factors of teachers needs to be
conducted in order to gain a deeper understanding of the influence these factors have
on levels of teacher efficacy. Again, a combination of quantitative and qualitative
methods would ensure this deeper level of understanding is obtained.
79


School related variables. The five studies that explored the influence of
school related variables on levels of teacher efficacy determined that decision-
making, collaboration, teacher support, and teacher resources all had significant
influences on teacher efficacy. Moore and Esselman (1992) conducted a study of
1,802 teachers to explore the influence of teacher empowerment and school climate
on teacher efficacy. They also examined student achievement data in math and
reading for the students of the participating teachers. The researchers constructed
their own survey that consisted of seven factors: teacher efficacy, personal efficacy,
school decision making, classroom decision making, school atmosphere, lack of
impediment to effective teaching, and teacher collegiality. Results of the
correlational analysis indicated that high personal efficacy correlated with greater
influence in school-based decision making and with a positive school atmosphere. A
positive school atmosphere was related to lack of impediments to effective instruction
and teacher collegiality. High teaching efficacy was correlated with higher levels of
classroom decision making. Math achievement differed between grades two and five
for teacher efficacy. Teachers with higher levels of teaching efficacy had higher
achieving students than teachers with low levels of efficacy. Using this same
instrument, Moore and Esselman (1994) followed three groups of elementary teachers
(N >350 per year) over a 3 year period to examine the influence of school contexts
on levels of teacher efficacy and student achievement. Path analysis revealed that
historical reading achievement had a significant indirect effect on teacher efficacy.
80


This effect was mediated by the variables of positive school atmosphere and lack of
impediments to effective instruction. Teacher efficacy was directly influenced by the
teachers perception of the lack of impediments to effective instruction. Personal
efficacy was shown to have a direct effect on teaching efficacy and math achievement
was a significant predictor of teacher efficacy, but was mediated by classroom
decision making and personal efficacy. A high pupil-teacher ratio also had a direct
effect on teacher efficacy. Finally, school atmosphere was more positive when the
school experienced fewer years below the national norm on 1TBS reading scores.
These two studies demonstrate that a variety of school conditions can influence levels
of teacher efficacy and student achievement. Having more influence in school based
decision making and classroom based decision making have a direct effect on
personal and teaching efficacy. Schools with a positive atmosphere also influence
teacher efficacy levels. Additional research is needed to confirm that current
conceptions of the construct of teacher efficacy as measured by the TSES also
correlated with these school variables.
Tschannen-Moran and Woolfolk Hoy (2002; 2007) examined the impact of
teacher resources and support on levels of teacher efficacy. Using the same pool of
255 inservice teachers with varying degrees of experience, the researchers looked at
different support variables, job satisfaction, and teacher setting. The 2002 study
explored the differences between novice teachers (< 5 years teaching experience) and
career teachers (5+ years teaching experience) on these different variables. The
81