Mitigating the effects of poverty

Material Information

Mitigating the effects of poverty a study of collective teacher efficacy, socioeconomic status, and student achievement in Colorado elementary schools
Pearce, Sharon Ann Roether
Publication Date:
Physical Description:
xii, 136 leaves : ; 28 cm


Subjects / Keywords:
Academic achievement -- Colorado ( lcsh )
Teacher effectiveness -- Colorado ( lcsh )
School children -- Social conditions -- Colorado ( lcsh )
Elementary schools -- Colorado ( lcsh )
Academic achievement ( fast )
Elementary schools ( fast )
School children -- Social conditions ( fast )
Teacher effectiveness ( fast )
Colorado ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 124-136).
General Note:
School of Education and Human Development
Statement of Responsibility:
by Sharon Ann Roether Pearce.

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:
181588630 ( OCLC )
LD1193.E3 2007d P33 ( lcc )

Full Text
Sharon Ann Roether Pearce
B.A., Eastern New Mexico University, 1965
M.A., University of Northern Colorado, 1974
A thesis submitted to the
University of Colorado at Denver and Health Sciences Center
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Educational Leadership and Innovation

2007 by Sharon Ann Roether Pearce
All rights reserved.

This thesis for the Doctor of Philosophy
degree by
Sharon Ann Roether Pearce
has been approved
Rodney Muth
Alan Davis
/- n-ioof
Stevie Quate

Pearce, Sharon A. (Ph.D., Educational Leadership and Innovation)
Mitigating the Effects of Poverty: A Study of Collective Teacher Efficacy,
Socioeconomic Status, and Student Achievement in Colorado Elementary
Thesis directed by Professor Rodney Muth
The purpose of this study was to determine correlational relationships
among teacher efficacy, collective teacher efficacy, socioeconomic status of
schools, and student achievement in Colorado elementary schools. The
theoretical basis for this study is Banduras social cognitive theory of self-
efficacy expanded to the collective school level. The unit of study for the
research was 25 elementary schools selected to represent high, average,
and low achieving schools. Schools in each achievement category
represented a range of socioeconomic levels, from high to low. Achievement
was measured by school student achievement data for third, fourth, and fifth
grade students on state reading, writing, and mathematics assessments.
Teacher information for teacher efficacy and perceived collective teacher
efficacy was obtained through a survey instrument.
A conceptual model was developed to represent hypothesized
relationships of teacher characteristics, teacher efficacy, collective teacher
efficacy, socioeconomic status of students in a school, and student
achievement for a school. Data were analyzed at both the teacher and
school levels. Results showed that collective teacher efficacy (CTE) is highly
related to elementary school student achievement on 3rd, 4th, and 5h grade
state assessments for math, reading, and writing. Student achievement and
CTE are correlated highly with SES. When student achievement was
controlled for SES, 3rdgrade reading significantly related to CTE,
independent of SES. The study also explored teacher characteristics that
might be sources of efficacy. Years of teaching experience accounted for
differences in personal teaching efficacy between beginning and mid-career
This abstract accurately represents the content of the candidates thesis,
recommend its publication.
Rodney Muth

This dissertation is dedicated to my family who supported me through this
My husband, Mike, who for more than forty years has been my chief
supporter, provided encouragement through many setbacks and challenges.
My children, Todd, Doug, and Denae, used humor to gently nudge me
My mother, Opal Copenbarger Roether, has always provided a compelling
vision for me to seek advanced education and at age 94 provides an
amazing model of how continuous learning enriches our lives and those of
others. My father, Lamar J. Roether, always stressed the importance of
formal learning and wanted each of his children to be well-educated. The
memory of his dreams for his children provided added motivation to
complete this dissertation.
Finally, my grandson, Benjamin, continually demonstrates joy of learning
that is contagious and, thus, provided an incentive for me to complete this
dissertation so I can play in the sandbox of life with him.

Sincere appreciation is expressed to the following people for their significant
contributions to my dissertation.
Dr. Rodney Muth, my advisor, held me to a standard of excellence, provided
support and expertise, read and responded to numerous drafts, and always
provided positive motivation.
Dr. Alan Davis, my methods advisor, provided expertise in methodology and
statistics and kept me focused on high quality research as a goal.
Dr. Sammye Wheeler-Clouse and Dr. Stevie Quate, committee members,
supported me through their recommendations and review of my work.
Dr. Nanci Avitabile, provided support for me in navigating the exciting world
of statistics.
Elementary school principals and teachers in my sample schools supported
this research by participation in survey completion.
Numerous friends and colleagues encouraged me during this process.

Figures ...................................................xii
Tables ....................................................xiii
1. INTRODUCTION.............................................1
Context of Problem....................................3
Purpose of the Study..................................6
Theoretical Framework.................................9
Overview of Methodology..............................13
Significance of the Study............................15
Limitations of the Study.............................18
2. LITERATURE REVIEW.......................................20
Teacher Efficacy.....................................23
Teacher Efficacy and Student Achievement.............30
Collective Efficacy..................................32
Collective Efficacy and Student Achievement..........36
Collective Efficacy and Socioeconomic Status.........38
Socioeconomic Status and Student Achievement.........40

Measurement of Efficacy..............................43
Survey Instruments...............................43
Methods for Analysis.............................45
3. METHODOLOGY.............................................51
Purpose and Research Questions.......................51
Methodological Framework.............................55
Independent Variables................................56
Dependent Variables..................................57
Survey Instrument....................................57
Teaching Experience..............................59
Years Teaching Current Grade/Subject.............59
Years in the School..............................59
Teachers Level of Education.....................60
CSAP Responsibility..............................60
School Position/Role.............................60
Sample Selection.....................................61
Survey Procedure.....................................65
Data Collection......................................67
Teacher Data.....................................67

School Data........................................68
Data Analysis...........................................69
Analysis Instrument................................71
Teacher Data.......................................71
School Data........................................72
4. FINDINGS...................................................73
Description of Teacher Sample......................73
Teaching Experience...........................74
Years Teaching Current Grade/Subject..........75
Years in the School...........................75
Teachers Level of Education..................76
CSAP Responsibility...........................76
School Position/Role..........................76
Description of School Sample.......................77
Factor Analysis for Teacher Efficacy....................79
Data Assumptions........................................80
Results for Question 1: Teacher Efficacy................82
Question 1.a.......................................83

Question 1.b
Question 1.c.......................................86
Question 1.d.......................................87
Question 1.e.......................................88
Results for Question 2: Collective Teacher Efficacy....88
Question 2.a.......................................89
Question 2.b.......................................90
Question 2.c.......................................91
Question 2.d.......................................94
5. DISCUSSION AND CONCLUSION.................................99
Summary of Findings....................................99
Discussion of Findings................................104
Measuring the Efficacy Constructs.................105
Sources of Efficacy Beliefs.......................106
Influence of Efficacy on Student Achievement......108
Efficacy, SES, and Student Achievement............111
Theoretical Implications..............................113
Practical Implications................................115
Future Research.......................................116

A. TEACHER SURVEY............................119
B. LETTER TO SUPERINTENDENTS.................121
C. LETTER TO PRINCIPALS......................122
D. CONSENT INFORMATION.......................123
REFERENCES .....................................124

1 A conceptual framework for teacher efficacy and collective
efficacy in relation to student achievement and SES.........14

1. Characteristics of Teacher Participants...........................74
2. Comparison of Percent of Students Identified
Proficient or Advanced for Study Schools (n = 25)
and State..........................................................78
3. Principal Component Analysis Using Varimax
Rotation with Kaiser Normalization................................81
4. Years of Experience Related to Teacher Efficacy...................84
5. Teacher Education Level Related to Teacher Efficacy...............86
6. Teachers Years in School Related to Collective Efficacy..........87
7. Relationships between Achievement, Teacher Efficacy,
Collective Efficacy, and SES at the School Level..................90
8 Correlations between SES, Teacher Efficacy, and
Collective Efficacy................................................91
9. Correlations between Student Math Achievement and
Teacher Efficacy Factors with and without Controlling
for SES at the School Level.............................................93
10 Correlations between Student Reading Achievement
and Teaching Efficacy Factors with and without
Controlling for SES at the School Level...........................93
11 Correlations between Student Writing Achievement
and Teaching Efficacy Factors with and without
Controlling for SES at the School Level............................94
12 Correlations between Collective Teacher Efficacy and
Student Achievement with and without Controlling for
SES at the School Level............................................95

Americas schools are engaged currently in meeting a challenge
that is unprecedented in history. That challenge is to educate one hundred
percent of the countrys young people to achieve high academic standards
and to graduate from high school. National legislation challenges every
school to show significant progress toward this ambitious goal by closing
the learning gap for subgroups of children who traditionally have not
shown high levels of achievement in public schools. One of the subgroups
that has been of concern for a long time is the population of students
labeled economically disadvantaged. Over three decades ago, Robert
Kennedy called this achievement gap between advantaged and
disadvantaged students a stain on our national honor (Brodsky, 2005).
So, while the concern has been present for a prolonged time, the national
focus on improving the school academic outcomes for these students has
significantly increased as a result of the No Child Left Behind Act (NCLB)
In Colorado, Colorado Student Assessment Program (CSAP)
results show that many schools are failing their students. In general, these

schools that consistently produce underachieving students share certain
characteristics. With few exceptions, these low-achieving schools reflect a
large percentage of students in poverty. Indeed, most schools in Colorado
show a substantial relationship between the number of students living in
poverty and the achievement of their students on state assessments. In
other words, in most cases when schools are sorted by percentage of
students eligible forfree/reduced lunch, the highest achievement levels
are seen in schools with low numbers of children receiving free/reduced
lunch. In the schools designated academically low or unsatisfactory, all
are impacted by higher numbers of students eligible for free/reduced
However, some schools despite their preponderance of students
with low socioeconomic status (SES) have shown significant student
achievement increases. Thus, while all of the Colorado elementary
schools in the lowest quintile of achievement are low SES schools, not all
low SES schools are in the lowest achievement quintile.
What accounts for this difference in student outcomes? How can at
risk schools positively affect academic achievement for disadvantaged
students? For America to be able to close the learning gap for students in
poverty, schools must identify the reasons for these disparate outcomes
and then change to ensure that all children succeed.

Context of Problem
Increasingly, researchers are looking for beat the odds schools
that appear to be closing their achievement gaps for students living in
poverty. While interest points to identifying high achieving schools that
have high poverty, high minority populations, currently no agreement
exists on what percent of schools are actually successful in meeting this
standard. Thus, it is no surprise that we do not have solid evidence to date
that shows a proven path to rapidly increase achievement levels for
disadvantaged students.
It is important to understand those factors that contribute to
success in low SES, high performing schools. One of these factors,
teacher expectations, has been shown through multiple studies (Eccles &
Jussim, 1992; Good, 1987; Lumsden, 1998; Rosenthal & Jacobson, 1968)
to be related highly to student achievement. These expectations are
influenced by teachers beliefs about the extent to which educators can
impact student achievement directly. Given that teachers beliefs are
important factors when considering student outcomes, previous studies
have explored the significance of individual teacher beliefs in relation to
school reform (Agne, Greenwood, & Miller, 1994; Armor et al., 1976;
Berman, McLaughlin, Bass, Pauly, & Zellman, 1977; Dembo & Gibson,

While numerous studies have explored individual teacher beliefs,
only recently has the impact of the collective beliefs of school
professionals in relation to the effectiveness of school reform become a
focus. Indeed, a basic assumption of the current NCLB Act (2002) is that
schools can expect the same high achievement outcomes for all students
despite differences in SES. However, this basic assumption is in direct
opposition to a previous study (Coleman et al., 1966) which concluded
that the single most reliable predictor of school success was the socio-
economic status of a students parents. While many acknowledge the
current relationship between SES and student outcomes, few want to
accept the notion that ones luck (or lack of luck) to be born to parents of a
particular SES predetermines ones educational destiny.
To ameliorate the current situation, NCLB (2002) mandates that
states receiving federal dollars must require specific actions of schools.
While the requirements are well-intentioned, some point out that these
remedial actions are in themselves discriminatory against schools serving
large numbers of students from disadvantaged backgrounds. For instance,
federal mandates require low functioning schools to initiate
comprehensive school improvement processes based on scientific
research-based practices. While schools acknowledge the need to
improve results for low achieving students, few high quality, scientific

research-based resources are available to these schools. Another
requirement aimed at improving the effectiveness of schools is that
schools must determine that all of their teachers are highly qualified,
based on nationally prescribed criteria. Currently, schools with the least
number of low SES students tend to attract the most experienced, most
highly trained teachers, while less experienced, less trained teachers are
generally assigned to the schools with highest number of low SES, low
achieving students. Increasingly school leaders are asking for clear
direction from researchers about what interventions will be effective and
realistic for students in low SES schools.
Because the school increasingly is recognized as the unit of
change, understanding the impact of educators beliefs in their collective
ability to raise achievement for all students is critical. The concept of
collective efficacy, or the collective belief of teachers within a school that
they can impact student outcomes positively, regardless of the challenges
that they meet, is beginning to receive the attention of researchers.
Schools are complex organizations and numerous factors have
direct and indirect affects on their effectiveness. Recent studies have
focused on identifying factors that impact student achievement. This study
examines some of the factors that have been suggested to be related to
teacher effectiveness and by association to school effectiveness.

Specifically, this research project examines individual and collective
teacher beliefs in relation to SES and student achievement. Information
about the relative strength of variables that affect teacher and student
behaviors may provide additional insights for school leaders and has the
potential to provide much needed direction for school improvement efforts.
Purpose of the Study
This study sought to answer the question, How are teacher
efficacy, collective teacher efficacy, school SES, and student achievement
related? The major hypothesis of the study is that elementary schools
with higher teacher efficacy and collective teacher efficacy have higher
student achievement, independent of school SES. Thus, this research
investigated teacher efficacy and collective teacher efficacy in Colorado
elementary schools and their relationships to student achievement and
SES. It was further hypothesized that collective teacher efficacy may be
as strong as SES in predicting student achievement. Specific research
questions and subquestions include the following:
Question 1: What are the relationships among selected teacher
characteristics, school characteristics, and efficacy for teacher level data?
a) What is the relationship between Teacher Efficacy and teachers
years of teaching experience?

b) What is the relationship between Teacher Efficacy and teachers
level of education?
c) What is the relationship between Collective Teacher Efficacy
and the teachers years in the school?
d) What is the relationship between SES and Teacher Efficacy?
e) What is the relationship between Teacher Efficacy and
Collective Teacher Efficacy?
Question 2: What relationships exist among collective efficacy, school
SES, and student achievement in Colorado elementary schools?
a) What are the bivariate relationships between student
achievement and Teacher Efficacy, Collective Teacher Efficacy,
and SES, at the school level?
b) What are the relationships between Teacher Efficacy, Collective
Teacher Efficacy, and SES?
c) What is the relationship between Teacher Efficacy and student
achievement controlling for SES?
d) What is the relationship between Collective Teacher Efficacy
and student achievement controlling for SES?

The following hypotheses guided this study.
1. Years of teaching experience are positively related to personal
teaching efficacy (PTE) and negatively related to parent trust
(PT) and teaching efficacy (TE).
2. Teachers with graduate education have higher teacher efficacy.
3. Teachers who have been at a school for more than three years
have higher collective teacher efficacy.
4. Teachers who work in low SES schools have lower teacher
efficacy than teacher working in high SES schools.
5. At the teacher level, teacher efficacy and collective teacher
efficacy have slight or no significant relationship to each other.
6. Schools with higher teacher efficacy have higher student
7. Schools with higher collective teacher efficacy have higher
student achievement.
8. Schools with lower SES have lower student achievement.
9. Teacher and collective efficacy have a relationship to SES.
10. When SES is controlled, teacher efficacy and student
achievement are positively related.

11. When SES is controlled, collective teacher efficacy has a
positive relationship to student achievement for reading and
12. Collective teacher efficacy is as strongly related to student
achievement as either teacher efficacy or socioeconomic status.
Theoretical Framework
Banduras (1977) social cognitive theory constitutes the framework
for this study. In relation to teaching, this theory specifies that teacher
perceptions of self and group capability influence actions, which in turn are
judged by the group norms established by collective efficacy beliefs
(Goddard & Goddard, 2001). Thus, social cognitive theory provides a
foundation for the theoretical analysis of the relationships between
personal and collective teacher efficacy and student achievement.
The first research project to apply Banduras (1977) theory to
schools came from two RAND Corporation studies of innovative
educational programs funded by the Federal Elementary and Secondary
Education Act (Armor et al., 1976; Berman et al., 1977). In these studies,
teaching efficacy and personal teaching efficacy were determined by
computing a total score from teacher responses to two, five-point Likert
scale items: (a) 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 and (b) If I try really hard, I can get
through to even the most difficult or unmotivated students. In the first
study, an evaluation of reading programs, student increases in reading
were reported to be strongly related to their teachers sense of efficacy
(Armor et al., 1976). The second RAND study concluded that teachers
attitudes toward their own professional competence have major affects on
learning outcomes (Berman et al., 1977).
Initial studies of efficacy in schools focused on the significance of
individual teacher efficacy beliefs. While these beliefs have been shown to
be strongly related to student achievement, more recent attention has
been given to collective teacher efficacy beliefs at the team or school
level. Bandura (1986, 1997) proposed that the concepts and assumptions
of social cognitive theory can be extended to organizations and are useful
in examining school outcomes. Thus, the relationships that exist between
individual teacher efficacy and student achievement also may be present
for collective teacher efficacy and collective school achievement.
Emerging information on collective school efficacy further suggests
that the interaction of teachers in a school is more than the sum of the
individual teachers efficacy (Bandura, 1982; Fuller, Wood, Rapoport, &
Dornbusch, 1982; Goddard, Hoy, & Woolfolk-Hoy, 2000; Lindsley, Brass,
and Thomas, 1995). Perceived collective efficacy is defined as a groups

shared belief in its conjoint capabilities to organize and execute the
courses of action required to produce given levels of attainment
(Bandura, 1997, p. 477).
The organization of schools is influenced by both internal and
external forces. External forces include community expectations for and
perceptions of the school. These external factors often are influenced by
cultural norms, as well as by perceived effectiveness of the school in
accomplishing its mission. The internal factors include social norms of the
people who work within the school and the schools culture. Because
schools are organizations where teachers work together in an interactive
social system (Bandura, 1993), the beliefs of the staff affect the
instructional activities and the organization structures. Peoples sense of
collective efficacy determines their well-being and what they accomplish
as a group (Bandura, 1997, p. 448).
Social cognitive theory relates teachers perception of self and
organizational influence to subsequent teacher actions. Thus, teachers
hold strong beliefs about their capacity to create change for students
within their organization. These beliefs are affected by experience, group
norms, and the culture of the organization. Simply, teacher beliefs
influence teacher behaviors, which in turn influence student achievement
(Goddard et al., 2000).

SES is another important variable to consider when examining
student achievement because it is a strong predictor of student success
(Barr, 2002; Bourke, 1998; Coleman et al, 1966). If teachers accept
Coleman et al.s (1966) research finding that most of the variation in
student achievement is due to home background, specifically the SES of
the parents, this external variable may influence their beliefs about the
capabilities of their students. This in turn may influence their beliefs about
their own ability to educate children from low SES backgrounds (Ashton &
Webb, 1986).
Collective teacher efficacy has been seen as a way to mitigate the
effects of poverty by focusing on internal motivators of behavior. Although
studies show a strong relationship between SES and student achievement
(Ashton & Webb, 1986; Ashton, Webb, & Doda, 1983; Bandura, 1997),
Bandura (1993) hypothesized that collective teacher efficacy may be as
predictive of student achievement as either socioeconomic status or
individual teacher efficacy. That hypothesis was supported by studies
(Bandura, 1993; Goddard et al., 2000) in which collective teacher efficacy
was positively associated with the differences in student achievement that
happened between schools. Further evidence of the power of collective
teacher efficacy was provided by Hoy, Sweetland, and Smith (2002). Their
study found a significant positive relationship between collective teacher

efficacy in a school and school achievement in mathematics, when SES
was controlled. Interestingly, Barr (2002) found that collective efficacy had
a positive effect on writing results for 8th grade students, independent of
SES, but that the relationship between collective efficacy for math and
reading were not independent of SES.
Based on the review of literature and the social cognitive theory
theoretical framework, I developed a conceptual framework to guide this
study. This conceptual framework includes three teacher characteristics
viewed as possible contributors to teacher efficacy or collective efficacy.
Student achievement is the outcome variable and teacher efficacy,
collective efficacy, and SES are the intervening variables. The conceptual
framework also represents possible relationships among several factors
that have the potential to influence student achievement at the elementary
school level. In addition to providing the theoretical framework for the
study, this conceptual model also helped prescribe the data to be
collected and analyzed: school-level and teacher-level data. Figure 1
shows this conceptual framework.
Overview of Methodology
Because this conceptual model combines individual teacher and
collective school data, it suggested two separate processes. The first
process involved determining relationships for components of the

Figure 1. A conceptual framework for teacher efficacy and collective
efficacy in relation to student achievement and SES.
framework that were represented by data that were collected and
analyzed at the individual teacher level. This included the teacher
characteristics (years of teaching experience, educational level, and years
at the current school), as well as teacher efficacy (including TE, PTE, and
PT) and collective teacher efficacy. The other process included data that
were collected, aggregated, and analyzed at the school level. These items
included teacher efficacy (including TE, PTE, and PT), collective efficacy,
student SES, and student achievement (for math, reading, and writing).

The schools selected for the study were identified following a
modified random sampling strategy. Although thirty schools participated in
the study, the final data analyses and results were based on data received
from 389 teachers in 25 elementary schools and school data provided by
the Colorado Department of Education. Schools with incomplete
demographic data or multiple missing survey responses were eliminated
from the study. Data were analyzed with various statistical procedures,
including Pearson Product Moment Correlations, One-way Analysis of
Variance, Partial Correlations, and Semipartial Correlations, using the
Statistical Package for the Social Sciences (SPSS for Windows, 2005).
Significance of the Study
Why is it important to study the relationships of teacher efficacy,
collective teacher efficacy, and socioeconomic status of students to
student achievement? If teacher efficacy is related to higher teacher
competency, then it is very important that low performing schools seeking
to improve outcomes for students develop a group of teachers who have
high efficacy. Currently, this is often not the case because teachers with
high efficacy may select to transfer to schools where they may feel higher
levels of success and greater support for their teaching efforts (National
Partnership for Teaching in At-Risk Schools, 2005). One study of New
York teachers by Lankford, Loeb, and Wyckoff (2001) found that teachers

who transferred to another district or left teaching altogether tended to
have better qualifications than their peers who remained. The present
reality of publishing school achievement results combined with school
report cards based on student achievement may lead to decreased
efficacy of students and teachers, and ultimately, may lead to a culture of
low expectations and learned helplessness. Glickman & Tamashiro (1982)
present evidence that those who leave teaching have significantly lower
scores on measures of teacher self-efficacy than teachers who remain in
teaching. Despite changes in both funding levels and people in leadership
roles, the outcomes for these schools may not be improved without
attention to the importance of individual and collective teacher beliefs.
Increased school accountability for student outcomes has placed
greater responsibility on local schools to demonstrate adequate yearly
progress (AYP) for all students, based on the assumption that all students
in all schools will reach high standards. Yet, the challenges to achieve this
goal are not uniform across school districts or even across individual
schools in districts. Major factors that contribute to these discrepancies of
opportunity relate to socioeconomic status (SES) and quality of teaching
staff. While pockets of excellence and some highly qualified teachers exist
in low SES schools, to a large extent, higher SES schools have the most
qualified and most experienced teachers. In contrast, teachers in schools

with high levels of poverty are often the youngest, least experienced
(National Partnership for Teaching in At-Risk Schools, 2005; Oakes,
Gamoran, & Page, 1992), and possibly the least efficacious teachers in a
district. According to the National Center for Education Statistics (2001),
twenty percent of teachers in high-poverty schools have three or fewer
years of teaching experience, compared with eleven percent of teachers in
low-poverty schools. Because high teacher efficacy has been shown to be
related to high teacher effectiveness and higher student achievement,
schools with the most qualified, highest efficacy teachers generally create
a culture of excellence in which student achievement is consistent across
time (Ashton & Webb, 1986; Fuller & Izu, 1986; Newmann, Rutter, &
Smith, 1989).
As schools are increasingly pressured to raise academic
achievement for all students and to close the gaps between outcomes for
disaggregated populations, the importance of understanding the impact of
collective efficacy on individual and group teaching behaviors becomes of
increasing importance for schools. For school leaders to develop effective
school improvement plans, for example, they need to know what factors
affect teacher behaviors and student outcomes. School leaders will benefit
greatly from having research-based evidence about factors that impact
student achievement regardless of the SES of the students in the school.

This study provides a contribution to professional literature by
presenting correlations that currently are not available. It has the potential
to provide a research base for school leaders who are responsible for
increasing student achievement in the presence of factors that have been
identified as barriers to school improvement. Although some of the
variables have been studied in other contexts, no previous studies have
investigated the relationships of teacher efficacy and collective teacher
efficacy to student achievement in low-, mid-, and high-SES elementary
Limitations of the Study
As with any study, this study has some limitations that need to be
considered. This study involves Colorado elementary schools and
Colorado state assessment data and, thus, the results may not generalize
to other states with other assessment measures. Since the schools
selected for study included some specific characteristics, it is not possible
to generalize the results from this study to schools that differ from the
sample schools. Specifically, these results may not apply to small
elementary schools of less than 100 students. It is also not possible to
generalize from this study to nontraditional elementary schools, such as
charter schools, magnet schools, or therapeutic schools, since these
schools were eliminated from the study. Additionally, only elementary

schools that had a principal who was not new to the school within the past
two years were selected. Therefore, results from this study may not be
applicable to schools with new principal leadership. Student achievement
was measured by school scores for math, reading, and writing on state
assessments. It cannot be assumed that these measures are equivalent to
state assessment measures in other states. This study did not attempt to
gather individual student or classroom achievement data to relate to
individual teacher efficacy beliefs. Finally, the small number of schools in
the study may have limited the statistical significance of some
relationships. Studies with more schools are needed to provide more
certainty about the statistical relationships among the variables.

This chapter presents a review of literature for the three primary
areas of emphasis for this study-teacher efficacy, collective teacher
efficacy, and socio-economic status-as they relate to student
achievement. Each section of this chapter discusses research on one of
these specific factors related to student achievement.
Efficacy is the belief that one can successfully accomplish a
specific task, such as teaching students to read at a proficient level.
Bandura (1986) defined self-efficacy as peoples judgments of their
capabilities to organize and execute courses of action required to attain
designated types of performances (p. 391). Wood and Bandura (1989)
expanded the definition of self-efficacy by adding that self-efficacy refers
to beliefs in ones capabilities to mobilize the motivation, cognitive
resources, and courses of action needed to meet situational demands (p.
408). Mitchell, Hopper, Daniels, George-Falvey, and James (1994)
concluded that self-efficacy clearly refers to what a person believes he or
she can do on a particular task (p. 506). Similarly, Gist and Mitchell

(1992) noted that efficacy judgments include motivational and integrative
aspects. Mitchell et al. (1994) concluded that capability, although based
heavily on ability, also reflects a forward-looking prediction of how hard
one will work and an integration of both of these factors (p. 506).
In addition to being task-specific, self-efficacy is context-specific.
Thus, some teachers may feel they are able to effectively help students
learn math in a middle-class suburban school, but may feel ineffective in
using these same instructional strategies in a high-poverty, urban school.
Beginning with such familiar ideas as self-fulfilling prophecy and the
Pygmalion effect, organizational researchers have focused on Banduras
(1977, 1982, 1986, 1991) theory of the relationship between self-efficacy
and performance (Gist & Mitchell, 1992). The conviction that one can
successfully execute the behavior required (Bandura, 1977, p. 193) has
been shown to have a positive effect on performance (Wood & Bandura,
Efficacy expectations have been shown to impact levels of
performance in a number of ways because they serve as a behavioral
predictor (Bandura, 1986). They affect goal setting, choice of activity,
amount of effort that will be expended, analytic strategies, and persistence
of coping behavior (Bandura, 1977; Wood & Bandura, 1989). Whereas
individuals avoid tasks perceived as exceeding their capabilities, they

undertake and perform successfully tasks that they perceive they are
capable of handling (Bandura, 1978). Thus, Wood and Bandura (1989)
concluded that individuals who demonstrate strong self-efficacy are more
likely to undertake challenging tasks, persist longer, and perform more
successfully than those with lower self-efficacy.
Banduras (1977, 1982, 1986, 1991, 1997) theory of self-efficacy
has been tested in varied settings and disciplines and has received
increasing support from diverse fields: health, mental health, vocational
studies, athletics, human resources, and education. It has been studied in
a variety of specific contexts: phobias (Bandura, 1983), depression (Davis
& Yates, 1982), stress, social skills, smoking behavior (Garcia, Schmitz, &
Doerfler, 1990), pain control, athletic performance, organizational
motivation, addiction (Marlatt, Baer, & Quigley, 1995), job satisfaction and
worker motivation (Bandura, 1997; Bandura, 2004; Pajares, 1996; Latham
& Pinder, 2005). Findings from these studies suggest that people who
have stronger perceptions of self-efficacy exhibit behaviors associated
with success. They set more challenging goals, exert more effort to attain
those goals, persist longer in their efforts, show greater resilience when
confronted with difficulties, maintain more positive attitudes, and respond
to stress more positively. As a result, these people attain higher
performance levels (Bandura, 1993).

Teacher Efficacy
Teacher efficacy, an extension of self-efficacy to education, has
received increasing attention in educational research during the past two
decades (Parker & Topping, 2006; Gibbs, 2003; Tschannen-Moran,
Woolfolk-Hoy & Hoy, 1998; Goddard, 2002a). The findings related to
teacher efficacy are similar to those found for self-efficacy. Teachers with
higher efficacy beliefs set more challenging goals, give greater effort to
their work, persist longer when faced with difficulties, show more
willingness to view non-confirmatory data, and show greater resilience. In
addition, they maintain more positive attitudes toward their work, their
students, their colleagues, their principal, consultants, and parents of their
students (DeForest & Hughes, 1992). Efficacy beliefs influence teachers
instructional practices, their decisions about referring students for
specialized help, their willingness to work with mainstreamed special
education students, and their willingness to adopt innovations (Podell &
Soodak, 1993).
Levels of teacher efficacy in schools have also been shown to be
correlated with the health and organizational climate of the school (Hoy &
Woolfolk, 1993), and with school morale of staff members (Gibson &
Dembo, 1984). School climate is increasingly being identified as a factor in
high achieving schools. As with education level, this connection between

teacher efficacy and school climate needs to be viewed cautiously. Hoy
and Woolfolk (1993) suggest that, student achievement may be the
fundamental factor affecting both teacher efficacy and school climate (p.
Teacher efficacy is the general term used to describe a teachers
self-efficacy beliefs. This includes both Teaching Efficacy and Personal
Teaching Efficacy, two dimensions of efficacy specific to education
(Gibson & Dembo, 1984; Hoy, 1993; Tschannen-Moran, Woolfolk-Hoy, &
Hoy, 1998). First, Teaching Efficacy (TE) is the belief of teachers in
general that they have the capability to significantly impact student
learning. Teachers with high TE believe that educators can affect student
learning through their teaching practices. How teachers perceive the
ability of educators to help all students achieve high academic standards
has a direct relationship to their effectiveness in teaching. The other type
of efficacy specific to education is Personal Teaching Efficacy (PTE), the
belief by a teacher that s/he can effectively teach in a way that will achieve
high results with all students. Teachers with high PTE not only believe in
the ability of teachers in general to make a difference for all students, but
they also believe that they personally have the skills and knowledge to
instruct all students effectively, regardless of other environmental or
student factors.

Researchers (Coladarci & Breton, 1997; Hoy & Woolfolk, 1993)
have noted the confusion between teacher efficacy, the overall construct,
and teaching efficacy, a dimension of that construct. For this reason, I
chose to use initials only for the dimensions of teacher efficacy, including
teaching efficacy (TE).
Accepted practice for examining efficacy among teachers is to use
a scale that measures both of its components (TE and PTE) and to check
for congruence between the scales. Thus, it is possible to have a teacher
who believes that educators, in general, can significantly impact student
learning (teaching efficacy) but to doubt that s/he can successfully impact
the learning of every student (personal teaching efficacy). It is generally
accepted that this type profile might be most true of a beginning teacher
who enters the profession with high commitment because of strong
teaching efficacy beliefs but quickly learns that teaching can be quite
difficult given the complexity of the challenges that are inherent within the
system. However, Hoy and Woolfolk (1990) identified this pattern of high
TE and low PTE among teachers who had been in the profession for
many years. He cautioned that it is not possible with current data to
determine cause and effect. Did these teachers always demonstrate high
TE and low PTE or did these teachers grow to have less PTE across time

as they encountered challenges and disappointments in producing
universally high achieving students?
It is conceivable, but less likely, that a teacher may have low
teaching efficacy and high personal teaching efficacy. In other words, a
teacher might not have a strong belief that educators in general within the
current system of education can truly produce high results for all students.
That same teacher might believe that s/he, due to extraordinary mission,
drive, skills, and other factors can successfully help all students rise to
ever higher levels of achievement. These teachers might be presumed to
take nontraditional career paths of seeking out new frontiers, perhaps the
most highly disadvantaged schools or charter schools. Thus, they may be
educational equivalents of Albert Schweitzer who went to Africa and built
his own villages, clinics, and schools and worked tirelessly in that setting
until his death. Despite this possibility, it is more likely that a person who
has little belief in the ability of educators to significantly impact learning for
all students would choose another profession.
Studies have consistently shown that these two dimensions of
efficacy are independent (Hoy & Woolfolk, 1990; Ashton & Webb, 1986;
Woolfolk, Rosoff, & Hoy, 1990; Woolfolk & Hoy, 1990). However, Hoy and
Woolfolk (1990) asserted that their results indicate that the dimension of
general teaching efficacy does not represent an outcome expectation as

defined by Bandura. Instead, it appears to reflect a general belief about
the power of teaching to reach difficult children and has more in common
with teachers conservative/liberal attitudes about education (p. 357).
Since Teaching Efficacy and Personal Teaching Efficacy are independent
but related constructs, it is important that research projects measure both
of them.
The concept of self-efficacy is closely related to Rotters (1966)
work in which he defined locus of control as a generalized expectancy
for internal or external control of reinforcements. The belief that an event
is contingent upon factors beyond personal control, such as random
chance or difficulty in the task, is termed external control. People with low
efficacy attribute success or failure to external factors. Thus, low efficacy
teachers may attribute the lack of student success to factors such as the
students capability or motivation, the students family or home situation,
lack of resources, central administration, wrong curriculum, or other
factors outside their control. Internal control refers to an individuals belief
that an event or outcome is contingent on her or his own behavior or on
relatively stable characteristics such as ability. A teacher who believes that
her/his teaching behaviors (internal locus of control) have high impact on
student achievement may experience more success with students than a
teacher who does not have that belief. Interestingly, Rotter defines internal

control in much the same way that Ashton and Webb (1986) describe
teacher efficacy. Bandura (1977), however, saw internal/external control
as different from efficacy and provided clarification for his reasoning. He
wrote that beliefs about whether one can produce certain actions (i.e.,
self-efficacy) are not the same as beliefs about whether actions affect
outcomes (i.e., locus of control). In fact, the data show that perceived
self-efficacy and locus of control bear little or no empirical relationship with
each other, and moreover, perceived self-efficacy is a strong predictor of
behavior, whereas locus of control is typically a weak predictor (Hoy,
1998, p. 155).
Because self-efficacy has such a powerful influence on individual
cognition, action, and outcomes, some researchers have sought to identify
how self-efficacy is created. While the sources of self-efficacy are varied
and complex, Bandura (1997) identified four broad categories of
experience that are central in the development of self-efficacy. First,
enactive mastery (i.e. prior personal attainment), has been confirmed to
have an effect on both student and teacher efficacy (Goddard, 2002a;
Goddard, LoGerfo, & Hoy, 2004). For example, teachers who work in
schools where students have historically achieved at high levels are more
likely to have high teacher efficacy beliefs that students will be successful
in the future and that they can successfully instruct those students. The

second influence on efficacy is vicarious experience (i.e., modeling). Thus,
in schools teachers learn from other teachers and perhaps through
coaching from another educator. The third experience that influences
efficacy is verbal or social persuasion which may take the form of
feedback and encouragement provided by effective school leaders who
influence teachers through development of and commitment to shared
missions, collaborative dialogues, coaching to support learning, and
collegial interactions. The fourth influence on efficacy is physiological and
affective states (i.e., anxiety and excitement) that are frequently referred to
in effective schools literature as school climate and school morale.
Schools currently engaged in developing professional learning
communities may be influencing teacher efficacy through a number of
these sources.
If teacher efficacy is so important, how do schools develop efficacy
in their teachers? Some studies have examined the factors that might
predict teacher efficacy. In Hoy and Woolfolks (1993) study, educational
level was the only personal variable that uniquely predicted teacher
efficacy. Teachers who went to graduate school for further education were
more likely to have a higher sense of personal teaching efficacy. On the
other hand, teaching experience was negatively related to general
teaching efficacy. Experience increased the likelihood that teachers would

believe that they could motivate difficult students and at the same time
promoted a sense of powerlessness to overcome the negative constraints
of the home. However, these findings do not indicate causality. This
apparent paradox might be the result of more highly motivated teachers
seeking additional education rather than as a result of advanced education
encouraging efficacy.
Teacher Efficacy and Student Achievement
Individual teacher efficacy beliefs and their relationship to elements
of schools and learning have been studied for almost three decades.
Considerable research has demonstrated strong links between individual
teacher efficacy beliefs, teacher behavior, and student achievement
(Ashton & Webb, 1986; Bandura, 1997; Gibson & Dembo, 1984; Pajares,
1996; Schunk, 1991; Tschannen-Moran, Woolfolk-Hoy, & Hoy, 1998;
Woolfolk & Hoy, 1990; Zimmerman, 1995). Numerous studies have shown
that teachers with high teacher efficacy take more risks, set higher
standards for themselves and their students, and have students who
consistently show academic gains (Ashton & Webb, 1986; Wood &
Bandura, 1989).
Teacher efficacy also has been shown to be related to many other
behaviors that have the potential to impact teacher effectiveness and,
therefore, student achievement. For instance, teacher efficacy has been

shown to be related to teachers adoption of innovations (Berman et al.,
1977; Guskey, 1988; Smylie, 1988), thus high efficacy teachers may be
more likely to accept newer scientific research-based instructional
practices and to persevere longer in becoming proficient in use of the
innovation. Additionally, teacher efficacy beliefs are strongly related to
their classroom management strategies (Ashton & Webb, 1986; Gibson &
Dembo, 1984) with higher efficacy teachers using strategies that preserve
student motivation and self-esteem, which can translate into a higher
likelihood of success for individual students. Teachers with high efficacy
feel they can be effective in managing student behavior, thus accepting
responsibility for establishing an orderly classroom environment, whereas
teachers with lower efficacy are more likely to assign blame for classroom
problems to others: the community, the parents, the students, or the
Some studies have identified the relationship between teacher
efficacy and a teachers decision making. Teachers with greater personal
efficacy are less likely to refer students with mild learning and behavior
problems to special education (Meijer & Foster, 1988; Soodak &
Podell,1993). Strong teacher efficacy beliefs serve to mediate teacher bias
when viewing student characteristics (Podell & Soodak, 1993).

Given the significance of the relationship between teacher beliefs
and student achievement an important question arises: Is the current
system failing students of poverty because of teacher beliefs about their
ability to help these students learn?
Collective Efficacy
Collective efficacy is an extension of Banduras (1977, 1982, 1986)
self-efficacy concepts. While self-efficacy refers to judgments that people
make about their personal or individual competency, group or collective
efficacy concerns judgments that people make about a groups level of
competency. Bandura (1997) defined collective efficacy as a groups
belief in its conjoint capabilities to organize and execute the course of
action required to produce given levels of attainments (p. 467). Simply
stated, it is the group members collective estimate of the groups
capability to perform a specific task (Gibson, 1999). Gibson, Randel, and
Earley (2000) suggest a distinction between collective efficacy and group
efficacy, defining collective efficacy as the aggregation of individual group
member perceptions of the efficacy of the group in contrast to group
efficacy, which they define as the consensus of the group regarding their
Bandura (1997) posited that the construct of self-efficacy might also
apply at the group level and that it might be even more potent than

individual efficacy. As work settings have come to require more
collaboration and interaction among workers, the concept of collective
efficacy has received more attention by researchers. Latham, Winters, and
Locke (1994) found higher levels of performance for groups that used
participative decision making versus individuals who set their own goals
and self-efficacy and task strategy were shown to be the mediating factors
on performance.
Empirical studies have examined the effects of collective efficacy in
diverse social systems, including athletic teams (Carron, 1984; Feltz &
Lirgg, 1998; Mullen & Cooper, 1994; Spink, 1990), combat teams (Jex &
Bliese, 1999), urban neighborhoods (Sampson, Raudenbush, & Earls,
1997), business organizations (Earley, 1994; Hodges & Carron, 1992),
and political systems (Pollock, 1983; Seligson, 1980). Regardless of the
system, collective efficacy is powerfully related to behavior that produces
desired outcomes.
Collective efficacy may be evident even when people are not
formally organized for a cause. Bandura (2004) points to the changes that
have occurred in America related to tobacco use as an example. Citizens
with shared beliefs in their collective efficacy to accomplish change played
key roles in recent policy and public health approaches to health
promotion and disease prevention. The prevalence of smoke-free

workplaces, restaurants, public buildings, and airliners, despite heavily
funded lobbying efforts to block these changes, reflects the power of
collective efficacy.
Only recently has research focused on the construct of collective
efficacy as it applies to schools. A study of collective and personal efficacy
of elementary school teachers found that the two measures of efficacy are
related but independent constructs (Parker, 1994). The same study found
that socioeconomic composition of a schools student body is a strong
predictor of teachers collective-efficacy (Parker, 1994). For example,
teachers who worked in schools with large numbers of students who
qualified for free and reduced lunch reported lower beliefs in the ability of
the school to raise student achievement than their peers in higher SES
Recent findings in the area of group efficacy also have established
strong links between group efficacy and group performance (Gibson,
1999; Gist & Mitchell, 1992; Silver & Bufiano, 1996). In addition, a groups
mission and commitment, the manner in which group members work
together, and the groups resilience in the face of difficulties are all
affected by group efficacy (Bandura, 1997).
Just as individuals tend to attribute outcomes to either internal or
external causes, organizations tend to attribute success to internal causes

and failures to external causes (Clapham & Schwenk, 1991). Thus, staff in
high achieving schools could be expected to attribute their success to their
effort and skills and to attribute any failures to factors external to them and
their school. When school staff members perceive that they have control
over student outcomes, self-correction is fostered (Lindsley et al., 1995).
Likewise, the perception that the causes of student, and thus school,
success are beyond the control of the staff may result in frustration,
anxiety, and feelings of helplessness (Mikulincer & Nizan, 1988). Feelings
of uncontrollability can stabilize over time, with debilitating results such as
learned helplessness (Peterson, Maier, & Seligman, 1993). In such a
case, individuals in the school will expend less effort and eventually
withdraw from activities (Kent & Gibbons, 1987). Hackman (1990) noted a
similar lack of effort and withdrawal by groups based on group attributions.
Thus, groups that perform poorly turn their attention outward, looking for
external factors to blame rather than focusing on ways to improve.
Bandura (1986) argued that the sources of collective efficacy are
similar to those for self-efficacy. One of the most powerful sources of
collective efficacy may be mastery experience, or past school
performance. In other words, schools that have performed well in the past
are more likely to have higher collective efficacy and schools that have
historically performed poorly are more likely to have lower collective

efficacy. Teachers who work in schools that consistently have low
performing students may become part of the cycle of learned
helplessness: their students perform poorly, they feel ineffective, their
students perform even poorer, they feel worse, and a negative spiral
continues until they begin to feel powerless to change the outcomes for
students (Mikulincer & Nizan, 1988).
Although social cognitive theory proposes sources of efficacy, the
extant literature contains very little empirical information about sources of
collective efficacy in schools. Enactive mastery, or prior student
achievement for the school, has demonstrated a positive and systematic
link between a schools prior academic success and its level of perceived
collective efficacy (Goddard & Skrla, 2006, p. 218). A recent study
(Goddard & Skrla, 2006) in a large urban school district in the United
States found that teachers of color and those with more than 10 years
experience reported slightly higher levels of perceived collective efficacy.
The influence of the three other sources of collective efficacy (vicarious
experience, social persuasion, and affective states) proposed by Bandura
(1997) are currently largely theoretical.
Collective Efficacy and Student Achievement
While strong links between individual teacher efficacy beliefs,
teacher behavior, and student achievement have been demonstrated

through research (Ashton & Webb, 1986; Gibson & Dembo, 1984;
Woolfolk & Hoy, 1990; Tschannen-Moran et al., 1998), only recently have
studies of school effectiveness examined the relationship between
collective teacher efficacy and student achievement.
Researchers have examined the relationship between collective
efficacy and student achievement in elementary schools (Goddard et al.,
2000), middle school (Barr, 2002) and high school (Hoy, Sweetland, &
Smith, 2002). Those initial studies showed that collective efficacy beliefs
are related to student achievement but may be content specific. For
instance, in some studies there was a positive relationship of collective
efficacy to student achievement in mathematics, reading and writing
(Goddard et al., 2000; Hoy et al., 2002). However, in another study (Barr,
2002), while there was a positive relationship between collective efficacy
and math and reading achievement, it did not hold up when SES was
controlled. However, writing achievement was significantly related to
collective efficacy independent of SES. Goddard et al. (2004) studied the
link between collective efficacy and 12th grade student achievement on
state tests for science, math, social studies, writing, and reading. Their
findings showed that perceived collective efficacy was a positive and
statistically significant predictor of 12th grade student achievement in both
the verbal and mathematical domains, including all five subject areas for

which high schools were held accountable by the state (Goddard et al.,
2004, p. 415). That study also provided empirical evidence that previous
student achievement was a predictor of perceived collective efficacy.
The emerging evidence is that a positive relationship exists
between collective teacher efficacy and differences in student
achievement among schools (Bandura, 1997; Goddard et al., 2000).
Bandura (1993), for example, found that teachers beliefs about their
schools collective efficacy were equally predictive of school achievement
as teachers self-efficacy. Bandura (1993) suggested that collective school
efficacy may be even more influential than personal teacher efficacy.
Collective Efficacy and Socioeconomic Status
Because SES is considered to be a strong predictor of student
achievement, some studies of collective efficacy have controlled for SES
(Bandura, 1997; Goddard et al., 2000; Goddard & Goddard, 2001; Hoy et
al., 2002; Barr, 2002). Of the studies that have examined the relationship
between individual teacher efficacy beliefs and the socioeconomic status
of the schools in which they work, one study identified a strong link
between the SES of the school and teacher efficacy (Parker, 1994). In that
study, teachers in lower SES schools had low teaching efficacy, that is,
they believed that their ability to impact student achievement was
insufficient to overcome SES and other environmental factors. SES does

not determine individual efficacy beliefs but strongly influences the
development of beliefs in indirect ways. Socioeconomic advantage
provides the resources and access to opportunity structures for the
development and exercise of personal efficacy. Socioeconomic status
indeed fosters a sense of personal efficacy and aspirations (Fernandez-
Ballesteros, Diez-Nicolas, Caprara, Barbaranelli, & Bandura, 2002, p.
Few studies have focused directly on the link between collective
efficacy and SES. In one study, Bandura (1993) found that student
achievement summed at the school level was significantly and positively
related to collective teacher efficacy and collective teacher efficacy was
more effective in predicting student achievement than SES aggregated at
the school level. Another study (Goddard et al., 2004) reported the same
strong link between SES and student achievement. Hoy et al. (2002),
while controlling for SES, identified a significant positive relationship
between the collective teacher efficacy of the school and school
achievement in mathematics at the high school level. In Barrs (2002)
study, collective efficacy was significantly related to student achievement
for middle school students in writing, independent of SES. However, that
study also found that the relationship between collective efficacy and
student achievement on the grade 8 math and English Standards of

Learning tests were not independent of SES. These results raise the
question of whether the impact of SES on student achievement is
dependent upon the grade level and content measured.
With the current focus on student achievement that is part of the No
Child Left Behind Act, many are beginning to ask what factors influence
the ability of a school to increase academic achievement for students in
low SES schools. Collective efficacy may provide a construct that can
contribute to school success.
Socioeconomic Status and Student Achievement
The relationship between SES and academic competence has
accumulated for over seventy years. Numerous studies (Alexander,
Entwisle, & Dauber, 1993; Bloom, 1964; Bradley & Corwyn, 2002;
Duncan, Brooks-Gunn, & Klebanov, 1994; Escalona, 1982; Hess,
Holloway, Price, & Dickson, 1982; Pianta, Egeland, & Sroufe, 1990;
Walberg & Marjoribanks, 1976; Zill, Moore, Smith, Stief, & Coiro, 1995)
have documented the association between poverty and lower levels of
school achievement and IQ later in childhood. DeGarmo, Forgatch, and
Martinez (1999) found that SES was associated with better parenting,
which in turn affected school achievement.
Bradley and Corwyn (2002) report that, There is evidence that the
connection between SES and cognitive performance applies to many

societies (p. 376). Mpofu and Van de Vijver (2000) identified that among
Zimbabwean children social class predicted the level of cognitive
classification strategy a child would use. In a study by Bradley, Corwyn,
and Whiteside-Mansell (1996), SES indicators were strongly related to
cognitive development from infancy through middle-childhood. Evidence
suggests a particularly strong relation between SES and verbal skills
(Mercy & Steelman, 1982). Hart and Risley (1995) found major differences
in the language skills (particularly for vocabulary acquisition) of children
from high-SES and those from low-SES families. Their study reinforced
the earlier findings of Hoff-Ginsberg (1991) who found substantial SES
differences in language performance for children, beginning early in life.
Some evidence suggests that the relationship between SES and
academic attainment diminishes with age (White, 1982). Yet, Smith,
Brooks-Gunn, and Klebanov (1997) found the effects of family income on
achievement to be similar for 3-year olds and 7-year olds, suggesting that
if there is a decreased impact, it may be gradual, if at all. The results of a
study by Walberg and Marjoribanks (1976) suggested that there may be a
kind of accumulated value to family environment and SES as they relate to
academic achievement. Thus, without any mitigating factors in the life of a
child from a low-SES family, the effects may actually become more
pronounced across time. In contrast, with the presence of some

intervening factors, such as effective schooling or adults who provide
mentoring, the relationship between the childs SES and school attainment
may weaken over time. However, Bradley and Corwyn (2002) caution that,
to date, little empirical data exist to support this proposition.
The relation between SES and cognitive attainment may
be quite complex, with different components of SES contributing
to the development of particular cognitive skills in different ways
and with some components of SES serving to moderate the
effects of other components. (Bradley & Corwyn, 2002, p. 376)
Through meta-analyses, White (1982) found that family income
accounted for the greatest amount of variance among traditional
measures of SES, and SES measures that combined two or more
indicators accounted for more variance than single indicators. This finding
points to the complexity of identifying a direct impact of SES on student
achievement. Social scientists point to the number of factors that are
related to SES that impact success in school: health care, nutrition,
crowded living conditions, lack of stimulating books and learning
experiences, past negative experiences of parents with school, noise and
violence in the neighborhood and other such factors. Despite the
complexity of the SES and school achievement relationship, only a few
researchers (Brooks-Gunn & Duncan, 1997; McLoyd, 1998) have
attempted to unpack the effects of socioeconomic indicators.

People who work in schools impacted by poverty have long noticed
that the challenges to attaining high levels of student academic
achievement in these schools are great. The Coleman et al. (1966) study
concluded that the socioeconomic status of a students parent(s) is the
most significant predictor of school success. Four decades later, SES
remains one of the most consistent predictors of early high school drop-
out, with evidence suggesting that it is connected to low parental
expectations (Battin-Pearson, Newcomb, Abbott, Hill, Catalano, &
Hawkins, 2000).
Measurement of Efficacy
A review of literature provides information about the historical
development of ways to measure and analyze the constructs of efficacy in
education. This section presents an overview of currently accepted survey
instruments and methods for analysis of survey data.
Survey Instruments
The original teacher efficacy scale was developed by Gibson and
Dembo in 1984, and was used in many of the initial studies (Allinder,
1994; Anderson, Green, & Loewen, 1988; Gibson & Dembo, 1984;
Guskey & Passaro, 1994; Ross, 1992). It contained items for both
teaching efficacy (TE) and personal teaching efficacy (PTE). In the
intervening two decades since it was developed, the items have

undergone revisions and refinement. As researchers established the
strength of individual survey items and balanced items measuring
teaching efficacy (TE) and personal teaching efficacy (PTE), some items
were eliminated. An improved instrument, the Ohio State Teacher Efficacy
Scale (Tschannen-Moran et al., 1998), has emerged as a widely accepted
instrument for measuring teacher efficacy beliefs. Hoy and Woolfolk
(1993) developed and validated a short form comprised of 10 items that
are balanced between TE and PTE. Their study found the short form to
have internal consistency (r= .94) and to be comparable to the long form
in construct validity (p = < .001).
Measurement of the construct of collective teacher efficacy has
also been the subject of study resulting in refinements across time. Some
initial studies used teacher efficacy measures and determined collective
efficacy by calculating a mean score for the school. More recently, it has
become accepted to use a survey tool with items worded to address a
groups beliefs. Using the collective efficacy model of Tschannen-Moran,
Woolfolk Hoy, and Hoy (1998) as a basis, Goddard et al. (2000)
developed a 21-item Collective Efficacy Scale that improved previous
scales. All items on their scale were directed at the group, not the
individual level, and analysis used school-level aggregates of individual-
level responses to the scale items. A subsequently developed short form

consisting of 12 items was determined to be more theoretically pure than
the earlier 21-item scale (Goddard, 2002b). The correlation between the
two scales (r= .983) suggested the 12-item scale was strongly related to
the original scale. The multilevel tests of predictive validity indicated that
the short form is a positive predictor of between-school variability in
student mathematics achievement (Goddard, 2002b, p. 108). In addition,
the 12-item scale yielded scores with high internal consistency (alpha =
.94). The Collective Efficacy Scale-Short Form is currently the most
reliable and valid measure for collective efficacy in schools.
Methods for Analysis
While agreement exists about appropriate instruments for collecting
teacher perceptions of group efficacy, less agreement exists about the
most appropriate method for aggregating and analyzing data. Gist (1987)
suggested three methods for assessing group efficacy. The first approach,
the aggregation of perceptions of individual self-efficacy, fails to account
for dynamic social and organizational processes that occur within groups.
Theoretically, it fails to acknowledge the group, or organization, as an
entity (Lindsley et al., 1995, p. 648). Supporters of this first approach
argue that homogeneity of individual measures justifies aggregation
(George & James, 1993; James, Joyce, & Slocum, 1988). Although
Bandura (1982) noted that collective efficacy is rooted in self-efficacy, he

emphasized the social and organizational processes that result in an
emergent, collective sense of efficacy and concluded that perceptions of
group as well as personal efficacy warrant examination (p. 143). Just as
an outstanding player on a poor team may have high self-efficacy, such a
player may also rate the teams collective efficacy as low. Lindsley et al.
(1995), however, surmise that situations may exist in which the sum of
individual self-efficacy is an adequate predictor of group performance. For
example, when extremely low task interdependence and low interaction
among group members exist, an aggregate may suffice. While these
conditions may be present in some elementary schools, aggregating
perceptions of self-efficacy is inconsistent with Banduras (1982)
recommendation. Because some disagreement remains about how
individual teaching efficacy beliefs aggregated at the group level relate to
collective efficacy, this study chose to use this method for one of its
analysis measures.
The second strategy for measuring group efficacy involves the
averaging of individuals perceptions of collective efficacy. As defined by
Earley (1993), This procedure reflects individuals perceptions of their
groups capability rather than a groups efficacy per se (p. 329). Using this
approach, respondents are asked their perception of collective efficacy.
For example, respondents indicate agreement/ disagreement with the

statement, I feel my group can accomplish this task. This measurement
strategy is useful if the assumption is made that individuals may not be
knowledgeable of the groups collective perception.
Gists (1987) third measurement possibility is the use of group
responses to a single questionnaire. This method avoids aggregation
problems and theoretically treats the group as an entity, consistent with
Banduras writings. Users of this strategy also consider the possibility of
differentiated collective mind (Weick & Roberts, 1993, p. 358) in that it
allows group members to combine parts into a consensual whole. Thus,
teachers would be asked to get together as a team or school and come to
consensus on a single response to each item. Such an approach may be
difficult either outside the laboratory or when focusing on large groups or
organizations, such as elementary schools.
Lindsley et al. (1995) suggest a fourth approach based upon
Earleys (1993) work. They endorse a method of using individuals as
informants to estimate the groups or organizations collective belief that it
can perform a specific task. This approach may be appropriate when the
researcher can reasonably assume that individuals have access to the
collective cognitions of the group (Lindsley et al., 1995, p. 649). Although
somewhat related to method two, focusing on the groups beliefs, rather
than the individuals beliefs, avoids many of the debates and pitfalls of

multilevel analysis (Glick, 1985; Klein, Dansereau, & Hall, 1994;
Rousseau, 1985). Weick and Roberts (1993) note that through social
interaction people, often discover higher-order themes, generalizations,
and ideas (p. 358) that transcend the differentiated parts.
A review of literature provides information essential for
understanding factors related to student achievement in schools. This
chapter reviewed studies relating to the general construct of self-efficacy
and to its specific application to teaching. Teacher efficacy has received
research attention for almost three decades and as a result is recognized
to contribute to student achievement through its influence on teacher
behaviors related to high effectiveness.
Collective efficacy has been studied in many fields other than
education and consistently demonstrates the powerful influence of the
beliefs of people to collectively achieve positive outcomes. More recently
researchers have begun to study efficacy at the collective level in schools.
While the construct of collective teacher efficacy has less of a research
history, the evidence from these recent studies is confirming that collective
teacher efficacy for a school is related to many of the same outcomes as
teacher efficacy. It appears that due to the power of collective beliefs that
influence collective action, collective teacher efficacy is an even more

powerful predictor of student achievement for a school than teacher
efficacy at the individual level.
This chapter also reviewed literature related to other factors
accepted to be related to student achievement. Socioeconomic status is
recognized widely as a strong influence on student achievement in a
school. While the prevalent view for the past three decades has been that
it is the single most powerful predictor of student and school success,
most researchers recognize that socioeconomic status does not need to
dictate destiny. Emerging evidence suggests that schools and
communities are able to intercede in the lives of students in such a way
that even students from the most disadvantaged backgrounds are beating
the odds in many American schools. High collective efficacy beliefs have
been shown in a few studies to be related to student achievement,
independent of SES. However, the evidence is sparse and currently is
specific to certain limited academic subjects.
Only recently have researchers begun to design studies to measure
the relationships between teacher efficacy, collective student efficacy, and
socioeconomic status of students in a school and their relationships to
student achievement for a school. The few studies currently available for
review have produced mixed results depending on the age of the students
and the specific achievement measures used.

Finally, this chapter reviewed literature related to accepted
research methodologies for measuring the constructs of teacher efficacy
and collective teacher efficacy. Researchers have developed survey
instruments with high reliability and high validity for both teacher efficacy
and collective teacher efficacy. Four methods for analyzing collective
efficacy data have been identified in professional literature: a) aggregation
of perceptions of individual self-efficacy, b) averaging of individuals
perceptions of collective efficacy, c) use of group responses to a single
questionnaire, and d) using individuals as informants to estimate the
groups or organizations collective belief that it can perform a specific
task. Each of these methods was reviewed and their advantages and
disadvantages were presented.

This chapter presents information about the methodology used in
this study. It describes the purpose, research questions, hypotheses and
conceptual framework. It also provides detailed information about the
research design, including independent and dependent variables, sample
selection, survey instrument and procedure, data collection, and data
Purpose and Research Questions
The purpose of this study is to determine what relationships exist
among factors that have been suggested to be related to student
achievement: teacher efficacy, collective teacher efficacy, and
socioeconomic status. To accomplish this purpose, two global research
questions guide the study.
The first question is, What are the relationships among selected
teacher characteristics, school characteristics, and efficacy for teacher
level data? This question relates to individual teacher efficacy beliefs and
the relationship of those beliefs to some factors identified in previous
studies to be related to teacher efficacy: formal education level (Hoy &

Woolfolk, 1993), years of professional teaching experience (Hoy &
Woolfolk, 1993; Goddard & Skrla, 2006), and the SES of the students in
the school in which a teacher works (Bandura, 1993; Parker, 1994;
Goddard, 2002a; Goddard et al., 2004). An additional factor that has not
been reported in the literature also is of interest, specifically, the
relationship between the length of time that a teacher has been in a school
and that teachers collective teacher efficacy. The final factor pertains to
the teacher efficacy of individuals in a school related to collective efficacy
for those teachers. These factors pertain to individual responses for all
teachers in the study. To answer this first question, the study explores
answers to the following subquestions:
a) What is the relationship between Teacher Efficacy and teachers
years of teaching experience?
b) What is the relationship between Teacher Efficacy and teachers
level of education?
c) What is the relationship between Collective Teacher Efficacy
and the teachers years in the school?
d) What is the relationship between SES and Teacher Efficacy?
e) What is the relationship between Teacher Efficacy and
Collective Teacher Efficacy?

The second global research question is, What relationships exist
among collective efficacy, school SES, and student achievement in
Colorado elementary schools? This question pertains to collective
efficacy of teachers within a school in relation to student achievement for
the school and the SES of students in the school. Thus, all of these factors
are examined at the school level. Subquestions of this question include:
a) What are the bivariate relationships between student
achievement and Teacher Efficacy, Collective Teacher Efficacy,
and SES, at the school level?
b) What are the relationships between Teacher Efficacy, Collective
Teacher Efficacy, and SES?
c) What is the relationship between Teacher Efficacy and student
achievement controlling for SES?
d) What is the relationship between Collective Teacher Efficacy
and student achievement controlling for SES?
Based on previous research, each question and subquestion has a
hypothesis. Thus, the following hypotheses guide this study.
1. Years of teaching experience are positively related to personal
teaching efficacy (PTE) and negatively related to parent trust
(PT) and teaching efficacy (TE).

2. Teachers with graduate education have higher teacher
3. Teachers who have been at a school for more than three
years have higher collective teacher efficacy.
4. Teachers who work in low SES schools have lower teacher
efficacy than teachers working in high SES schools.
5. At the teacher level, teacher efficacy and collective teacher
efficacy have slight or no significant relationship to each other.
6. Schools with higher teacher efficacy have higher student
7. Schools with higher collective teacher efficacy have higher
student achievement.
8. Schools with lower SES have lower student achievement.
9. Teacher and collective efficacy have a relationship to SES.
10. When SES is controlled, teacher efficacy and student
achievement are positively related.
11. When SES is controlled, collective teacher efficacy has a
positive relationship to student achievement for reading and

12. Collective teacher efficacy is as strongly related to student
achievement as either teacher efficacy or socioeconomic
Methodological Framework
The conceptual model suggests the methodology for this study.
Thus, research design includes examination of teacher-level and school-
level data to determine how the different constructs of teacher efficacy and
collective teacher efficacy might relate to each other, to SES, and to
student achievement for a school. Although the model reflects a
relationship between TE and student achievement, the design of this study
only allows for establishing this link for aggregated school achievement.
The research questions guiding the study address the relationships
between the various components of the conceptual model. Teacher-level
data are used for analyses related to Question 1 and a combination of
teacher data and school data at the school level are used for Question 2.
A representative school sample was obtained using a modified stratified
random sampling technique and data for the teachers in those schools
was collected employing a survey methodology. Other school data were
collected from the state department of education. The data were analyzed
through a variety of statistical methods to determine relationships between
the various components of the conceptual model.

The following sections describe in more detail the identification of
schools to survey, the design of the survey instrument, the procedure for
collecting the data, and the statistical analysis methodology.
Independent Variables
Independent variables for this study operationalize the constructs of
teacher efficacy, collective teacher efficacy, and socioeconomic status of
students in elementary schools. Teacher efficacy is measured by the
Teacher Efficacy Scale-Short Form (Hoy & Woolfolk, 1993) and collective
teacher efficacy is measured using the Collective Efficacy Scale-Short
Form (Goddard, 2002b). These instruments are described in detail in the
survey instrument section.
For this study, socioeconomic status is represented by the
percentage of students in the school identified as eligible for free/reduced
lunch in the same year that the efficacy measures were obtained. This
measure of SES is inversely related to actual SES. Thus, a school with a
high proportion of students receiving free/reduced lunch is a low SES
school. While socioeconomic status is a complex variable, for purposes of
studies involving schools, the common practice is to use the free and
reduced lunch data because it is viewed as a good indicator of family SES
factors. While other measures such as household-level data covering
income, educational level of parents, occupation, and home ownership are

sometimes used to compute SES, they are not commonly used for school
studies because of the impracticality of collecting that type data in schools.
Free and reduced lunch data was obtained from current data reported to
the Colorado Department of Education and posted on their website.
Dependent Variables
The dependent variables for this study include student achievement
for reading, writing, and mathematics for each of the sample schools.
Student achievement for the school used scores on the Colorado Student
Assessment Program (CSAP) for reading, writing, and mathematics for
3rd, 4th, and 5th grades. The CSAP data were obtained from the Colorado
Department of Education for the same year in which the efficacy
measures were administered. These subjects were selected because at
least one study found the influence of SES and collective teacher efficacy
to differ for subjects such as reading, writing and mathematics (Barr,
2002). Grades three through five were selected for study since most state
elementary schools include those grades. Although a few of the schools in
the study included sixth grade the majority did not with that grade
assigned to middle school.
Survey Instrument
Teacher data for this study were gathered through survey
methodology. This survey was developed following a literature review to

identify currently accepted instruments for measuring teacher efficacy and
collective teacher efficacy and currently accepted best methodology. The
survey items selected to measure individual teacher efficacy were the 10
items from the Teacher Efficacy Scale-Short Form (Hoy & Woolfolk, 1993)
and the items selected to measure collective teacher efficacy were the 12
items from the Collective Efficacy Scale-Short Form (Goddard, 2002b).
Thus, the survey instrument included 22 items with the first 10 designed to
elicit responses about an individual teachers efficacy beliefs and the last
12 items worded to elicit responses about that individuals collective
efficacy beliefs for the school. To ensure congruence between the two
efficacy measures, the order of the scale for the first ten items was
reversed so that the 6 point Likert scale moved from Strongly Disagree to
Strongly Agree. The last twelve items of the survey were the 12 items from
the CES-S, preserving the original 6 point Likert scale order of Strongly
Disagree to Strongly Agree.
The final part of the survey instrument included six demographic
items related to teachers: (a) years teaching experience, (b) years
teaching current subject/grade, (c) years at current school, (d) educational
level, (e) responsibility for teaching a CSAP subject/grade, and (f) current
teaching position. These items were included to give information about

some teacher characteristics that may be influences on teacher efficacy or
collective efficacy. Each of these items is addressed in the next section.
Teaching Experience
The first item of demographic information requested was total
number of years of teaching experience. It was hypothesized that teachers
with more years of teaching experience would have higher teacher
efficacy and, conversely, that teachers new to the profession would have
lower teacher efficacy.
Years Teaching Current Grade/Subject
The second item that teachers provided was the number of years
teaching their current grade and/or subject. This item was included
because of the possibility that teachers might have been in the profession
for a significant length of time but recently have assumed new teaching
assignments that might influence their efficacy beliefs.
Years in the School
The third item for demographic information related to the number of
years a teacher had been at the current school. It was hypothesized that
collective efficacy might be influenced by the length of time teachers had
been at the school. Thus, teachers who have been at a school for a long
period of time might have higher efficacy beliefs about the ability of
teachers in their school to make a difference for students.

Teachers Level of Education
The fourth demographic item on the survey instrument asked
respondents to provide information about their educational level by
checking one of several choices: BA, BA+, MA/MS, MA+/MS+, or
Ph.D/Ed.D. The hypothesis was that teachers with more education would
have higher teacher efficacy because the agency required to pursue
graduate education would relate to higher efficacy. It was hypothesized
that a teacher with low efficacy would not make this investment of time
and effort.
CSAP Responsibility
The fifth demographic item requested was a response to the
question, Do you teach a grade/subject tested by CSAP? Respondents
were asked to check yes or no. Although this study was not designed to
analyze data based on teaching assignment, I had an interest in whether
teachers with primary responsibility for student achievement as measured
by the CSAP would respond differently from teachers without that
School Position/Role
The final demographic item asked teachers to check one of four
choices that most described their current position. These choices were:
classroom teacher, special education, electives/specials teacher, and

other. Respondents who checked other often provided additional
information, such as psychologist, reading specialist/coach, or
Speech/Language Pathologist. Although this study was not designed to
analyze data based on different educator roles, I felt this information might
provide valuable information about the participants and might provide the
basis for a future study. Appendix A contains the survey instrument used
in the study.
Sample Selection
The sample for the study was chosen from traditional public
elementary schools in Colorado. Charter schools and non-traditional
schools, such as magnet schools, were not included in the study because,
by their nature, they often do not represent a heterogeneous student
population. The schools to be sampled were selected to represent high,
middle, and low socioeconomic status and high, middle, and low student
achievement levels. The socioeconomic status was based on percentage
of students in a school who qualified for free/reduced lunch based on data
posted on the Colorado Department of Education website.
To obtain a representative sample, all public elementary schools in
the state, except for magnet schools, charter schools, and other special
schools, were sorted by Colorado School Accountability Report (SAR)
rankings, from highest achievement to lowest achievement. For the

School Accountability Report, schools are identified as excellent, high,
average, low, and unsatisfactory based on ten criteria, with strong
weighting on student performance on the Colorado Student Assessment
Program. For the purposes of this study, schools rated excellent or high
were included in the high category and schools rated low or unsatisfactory
were included in the low category. Thus, all public elementary schools in
Colorado were sorted into one of three achievement categories: high,
average, or low.
To obtain a sample with equal distribution of high, average, and low
socioeconomic status (SES) schools, each list of schools previously
sorted by performance was resorted by the percentage of students who
qualified forfree/reduced lunch. Thus, schools in the high achievement
category were listed from highest percent to lowest percent of students on
free/reduced lunch. Each list then was divided into three equal parts to
represent the high (top one-third), average (mid one-third), and low
(bottom one-third) portion of the list.
The final refinement of the lists narrowed the number of schools for
selection according to criteria related to the following practical
a) traditional neighborhood public elementary school (excluding
magnet, charter, and special population schools),

b) student population of 100 or more students,
c) located within four hours drive from Denver metropolitan area,
d) had been in existence for at least five years, and
e) principal had two or more years experience in the school.
These criteria were used for purposes of sample consistency on
selected characteristics while permitting differences in school location (i.e.
rural, urban, and suburban) and school size (with the exception of the
minimum number of students). Schools with small enrollments (under 100
students) were eliminated since it was expected that they would not have
sufficient numbers of classroom teachers (no less than five) to provide a
valid collective score. In addition, schools with at least one hundred
students had a higher likelihood of having scores reported on the CDE
web-site, since scores are not reported for any groups of less than fifteen
Another delimiting factor was distance from the Denver
metropolitan area. Because Colorado is a large state, some schools might
require up to ten hours of travel by car. I decided for practical purposes
only to include schools within four hours of travel.
The final factor used to define the target schools was the time the
principal had been at the school. Numerous studies have identified the
elementary school principal as a powerful influence on school culture,

climate, and student achievement. Thus, schools selected for inclusion in
the sample had a principal who had been at the school two years or more.
Information about the principals tenure at the school was determined
during the first personal contact, in most cases. In some instances, the
initial contact with the district superintendent provided that information.
At least six schools in the top third, middle third, and bottom third
for each achievement group were identified. Thus, from the original sort,
fifty-four schools were identified as potential participants. Four schools in
each section, for a total of 36 schools, were randomly selected to be
contacted initially. The sorted lists were retained in case replacement
schools needed to be chosen for any reason.
Administrators in each of the initially selected school districts were
contacted to request permission to conduct the survey with their teachers.
In most cases, the initial contact was with a superintendent or a building
principal. In some cases, permission for research needed to be obtained
through a central office person other than the superintendent and thus,
contacts were with that person. In some cases, this involved formal
application with documentation of university permission to conduct the
study. Once permission was granted to contact principals, contacts were
made with building principals to explain the study and to request
permission to collect data in their schools. In all cases, permission to

collect data within a school adhered to policies and procedures
established by UCD, school districts, and individual schools. When
requested, follow-up letters explaining the study and a copy of the
research prospectus were sent to school officials responsible for granting
permission for research studies. Appendix B contains a copy of a sample
letter to Superintendents and Appendix C contains a copy of a sample
letter to a principal.
The primary incentive for building principals to encourage their
teachers to participate in the survey was the offer to receive the schools
data for use in school improvement planning processes. Many principals
viewed that as sufficient to take the time for staff to complete the survey.
In a few cases, the building principal had sympathy for a doctoral student
based upon his/her personal experience in completing a dissertation.
Following principal permission to proceed, I arranged a time to administer
Survey Procedure
The method for administering individual teacher surveys depended
upon the preference of the building principal. In most cases, I personally
administered the surveys during a faculty meeting and collected the
completed surveys immediately. In other cases, I explained the survey
process at a staff meeting but teachers completed them outside of the

meeting and I returned to collect them later. An unanticipated method for
getting teacher participation arose toward the end of my visits to schools.
This method, suggested by a principal, turned out to be a very pleasant,
although time intensive, process. The principal would notify teachers that
they had an opportunity to gather information for school improvement
planning and that a doctoral student would be in the staff lounge during
lunch periods on a specified date and that they were encouraged to
participate in the survey. This proved to be highly effective in gaining
participation by the staff.
During the administration of the survey, I explained the purpose of
the study, assured confidentiality, and asked that teachers complete the
survey as candidly as possible by circling the number that best indicated
their degree of agreement or disagreement to the statements. All
participants were promised confidentiality related to their responses. In
addition, they were assured that all data collected would be reported as
aggregated data. Participants were informed that schools would be
notified of the completion of the study and given information on where they
could obtain a copy of the completed study. Participation in the study was
strictly voluntary. Appendix D contains the Informed Consent Form that
was signed by teachers and submitted in a separate envelope from the

Different incentives were used to encourage teachers to participate
in the survey. The most popular was warm, home-baked sweet breads
that I provided. In one case, I provided lunch for the entire staff on a staff
work day. The sanctioning by their building principal was also an important
incentive for teacher participation.
Data Collection
Data collected for this study included two types: teacher data
(participant responses on the survey) and school data. Separate SPSS
databases were developed for the two types of data.
Teacher Data
Teacher data, including both responses to items related to efficacy
and personal demographic information was supplied by teacher
participants on the survey instrument. Teachers who did not complete the
demographic items were not included for purposes of data analysis.
Following completion of surveys for a school, individual survey
responses were entered into a SPSS database. Surveys were not
constructed to personally identify individual participants, however, in small
schools the demographic information might allow individuals to be
identified. Thus, although this research analyzed collective teacher
responses for a school, rather than individual responses, an additional
step was taken to protect the identities of individual teacher respondents.

All data analysis was completed using this anonymous identifier for
schools. In addition, each individual survey was given a number that was
used for data entry. Responses from each survey were entered into the
SPSS database, coded by school number and survey number. Only
surveys with no more than two missing responses were included in the
final database. All entries were double checked for accuracy.
School Data
School data included school size (number of enrolled students),
SES indicator (percentage of students receiving free and reduced lunch),
and achievement data (percentage of students reported as proficient or
advanced on the CSAP for math, reading, and writing for 3rd-5th grades).
The school data were obtained from public records available on the
Colorado Department of Education website.
School data were entered into SPSS using an identifier number,
rather than school names. For each participant school, CSAP data for
each grade and each subject were entered into separate cells (i.e., 3rd
grade math, 3rd grade writing, 3rd grade reading, etc.). Because the
schools in the High achievement group skewed toward high or average
SES and the schools in the Low/Unsatisfactory achievement group
skewed toward lower SES, actual percentages of free/reduced lunch for a
school were used for purposes of data entry and analysis.

Accuracy was assured by a double check of all of the information.
The school data were printed and examined for any possible incorrectly
entered or missing data.
Data Analysis
Once all data were collected, entered into a database, and double
checked for accuracy, I began to analyze the data. This section describes
the process that I used for analysis.
A first step was to obtain frequency distributions for each item to
ensure that there was no problem with how people responded to individual
items. This process alerted me to any missing data or possibly incorrectly
entered data. A few changes were made to correct the data entered.
Previous studies (Ashton & Webb, 1986; Hoy, 1993) consistently
identified that teacher efficacy is comprised of two independent
dimensions: teaching efficacy and personal teaching efficacy. Hence, the
next step in the data analysis process was to run a confirmatory factor
analysis of the items that composed the teacher efficacy items to
determine if the instrument loaded on the same factors for the specific
population studied. The decision was made to perform the same analysis
for the CTE scale, although the developers of the instrument did not
specify that step. Once the factor analyses were completed, reliability

analyses were then done for the factors identified. Because the results of
these analyses suggested that the two factor solution might not represent
all the underlying constructs in this data set, a subsequent exploratory
factor analysis was conducted. One item, survey item #5, If parents would
do more for their children, I could do more did not seem to factor cleanly
into either TE or PTE. Upon examination, item #5 seemed to represent a
separate construct. Thus, it was decided to consider that single item
separately. This factor was subsequently labeled Parent Trust (PT) to
indicate it may relate to a basic belief in whether a teacher feels s/he can
trust that parents are supportive of education and the teacher.
All survey items were analyzed at both the individual teacher level
and collective school level. The average item score for each of the items
for a school was computed to obtain an average school score for each of
the items. TES-S represents independent factors, thus, the data for the
first ten items of the survey were aggregated by factor rather than as a
single aggregated item while the items for the CES-S were aggregated
into a single score. From these totals, the means for each factor (TE, PTE,
PT, and CTE) were computed to be used in analysis of the school data.

Analysis Instrument
Data were analyzed using the Statistical Package for the Social
Sciences (SPSS, 2005) software. Detailed information that resulted from
the statistical analyses is reported in detail in the next chapter.
Teacher Data
The first part of the conceptual model includes the elements related
to teacher efficacy for all teachers in the study. It provides the framework
for data analysis of items related to teacher efficacy and collective efficacy
for individual teachers for the schools sampled. Specifically, bivariate
correlations were calculated for each of the factors in relation to each of
the other factors. The order suggested by the framework was to start with
the demographic items on the left side of the model and to work toward
student achievement on the right side. It is important to note that all of the
items were analyzed as an aggregate of individual teacher responses with
the exception of school SES and student achievement which were based
on data related to the school as a whole.
Given the purpose of this study, and the advantages and
disadvantages of proposed approaches for computing collective efficacy,
this study used the first and fourth approaches identified by Gist (1987).
The first approach represented the sum and, therefore, the average
teacher efficacy at the individual level aggregated to a collective level.

This method was used for analyzing data for question one. The fourth
approach resulted in individual estimates of the schools collective belief
that it can impact student achievement for all students. The TES-S by
design lends itself to use with the first approach and the CES-S by design
conforms to the fourth approach.
School Data
School data were studied in much the same way as the teacher
data based on the conceptual model. To answer question 1 and each of
its subquestions, the variables of teacher efficacy at the school level,
collective efficacy for the school, student achievement, and SES were
analyzed using bivariate correlations to establish relationships of the
variables. To answer question #2, semipartial correlations were performed
to determine relationships among teacher efficacy, collective efficacy, and
student achievement when SES was controlled. All of these factors
involved school-level data, in contrast to the factors used for Questionl
that examined teacher-level data. Results for these analyses are
presented in Chapter 4.
Following the statistical analysis of the data, possible conclusions
were made and unusual patterns were identified for further study. These
conclusions and recommendations for further study are presented in the
final chapter.

This chapter describes the results for this study. It is organized by
the sequence in which the data were analyzed: descriptive data for the
participating teachers and schools, factor analysis for the survey
instrument, results for teacher efficacy, and results for collective efficacy.
This section presents the descriptive information for the study
sample. First, information is presented for the 389 teachers who
completed the survey. Second, descriptive information based on
aggregated data is given for the sample consisting of the 25 schools in
which the teachers worked.
Description of Teacher Sample
Teachers supplied information about themselves that helped
identify characteristics of the sample participants. Table 1 presents the
descriptive information based on the responses of 389 teachers who
completed the survey instrument used in this study.

Table 1
Characteristics of Teacher Participants
Factor no. %
BA 135 34.8
BA+ 1 0.3
MA 100 25.8
MA+ 151 38.9
Ph.D. 1 0.3
CSAP Grade Teacher
No 168 43.2
Yes 221 56.8
School Position/Role
Classroom Teacher 276 71.3
Special Education 36 9.4
Elective/Specials Teacher 43 11.1
Other 31 8.4
Teaching Experience
Beginning (1-5 years) 106 27.2
Tenured (6-10 years) 87 22.4
Mid-career (11-15) 69 17.7
Veteran (16+ years) 127 32.6
Years at current grade/subject
1 -5 years 225 57.8
6-10 years 93 23.9
11-15 years 35 9.0
16+ years 36 9.3
Years at current school
0-3 years 160 41.1
4-7 years 103 26.5
8-15 years 84 21.6
16+ years 42 10.8
Teaching Experience
For the total sample, the mean number of years of teaching
experience was 12.44 years ( 9.02) ranging from .25 to 41 years.
Collapsing the data into career level categories indicated that

approximately one-third of the teachers were veterans (16+ years) and a
little over one-quarter were beginning teachers (1-5 years). Almost all
schools in the study had at least one teacher with more than 25 years
experience. These schools appeared to be representative of the variance
among schools in Colorado. When I examined individual school data, it
appeared that districts adding classrooms due to growth in student
population had a larger proportion of beginning teachers.
Years Teaching Current Grade/Subject
The information teachers provided about the number of years they
had been teaching their current grade and/or subject showed that on
average they had been doing so for 6.52 years ( 6.73). It is interesting to
note the extreme range (.25 to 41 years) and that some teachers had
spent their entire careers teaching at the same grade level.
Years in the School
Teachers supplied information on the survey about the number of
years they had been at the current school. For the 389 teachers, the
responses ranged from .25 to 31 years with a mean of 6.73 years ( 6.37).
The sample was heavily skewed toward teachers who had been at the
school ten years or fewer. An examination of data at the school level
suggested that this may have been influenced by the age of the school.
While some schools have existed for a very long time, some for almost a

century, some schools in rapidly growing districts were fairly new schools
(5-10 years). However, due to the selection process, none of the schools
had been in existence for fewer than 5 years.
Teachers Level of Education
The majority (65%) of the teachers in the sample had advanced
degrees or education beyond the BA/BS level.
CSAP Responsibility
For the sample schools, 221 (56.8%) teachers indicated that they
are responsible for teaching students who will be taking the CSAP.
However, when the information provided for this item was compared with
the information about their school role (classroom teacher, electives
teacher, special education teacher, etc.), it appeared that some teachers
who taught students in multiple grades/subjects, such as special
education teachers also indicated they were responsible for CSAP
preparation. This may reflect the growing ownership of all certified staff in
a school for helping all students perform proficiently on state
School Position/Role
The majority (71.3%) of those who responded to this item were
classroom teachers. The next largest group (11.1%) was
electives/specials teachers (music, art, physical education, technology).

This distribution of teaching roles appeared to match a typical distribution
for elementary schools.
Description of School Sample
The schools selected for the sample were distributed throughout
regions representative of the states geographic areas and included
schools that were rural, suburban, and urban. They were located in
farming communities, mountain areas, moderate size towns, cities, and
metropolitan areas. The smallest schools had student populations of 1 GO-
175 students (or from 15 to 27 students at each grade level, depending on
the configuration of the school) while the two largest schools had multiple
classes at each grade level and had 663 students. These largest schools
were on a year-round school schedule with one-fourth of the students and
teachers on vacation during each track. The socioeconomic status of the
schools, by design, also represented the demographics of the state as a
whole. The lowest SES school selected for the study had a free and
reduced lunch eligibility of 91.41% and the highest SES school had only
1.89% of its students eligible for free or reduced lunch. Almost half of the
schools in the sample had 50% or more of their students eligible for free or
reduced lunch.
The data used to represent student achievement consisted of the
percentage of students at grades 3-5 who were proficient or advanced in

math, reading, and writing. This is a more precise way of representing
student achievement than the School Accountability Report descriptor
(Excellent/High, Average, or Low/Unsatisfactory), which was used as the
achievement measure for initial selection of sample schools. Table 2
presents the summary of the achievement scores for the sample schools
and the state means for each grade and subject measure.
Table 2
Comparison of Percent of Students Identified Proficient or Advanced for
Study Schools (n = 25) and State____________________________
Subject/grade Study State t
Range Mean S. D. Mean P
3rd grade 45-91 72.60 13.44 68.00 1.71 .10
4th grade 23-91 69.84 15.53 66.00 1.24 .23
5th grade 37-87 66.16 15.20 63.00 1.04 .31
3 grade 52-93 76.68 10.14 71.00 2.80 .01*
4th grade 24-91 67.20 17.76 64.00 0.90 .38
5th grade 43-90 70.52 14.14 69.00 0.54 .60
3 grade 25-79 54.28 15.16 56.00 -0.57 .58
4th grade 9-79 51.08 19.90 52.00 -0.23 .82
5th grade 25-85 56.40 16.14 57.00 -0.19 .85
*p < .05.
The variance between grades and subjects suggested that it was more
informative to view the data by subject and grade level than to collapse
the data into a single measurement for each subject. Because previous

studies have suggested that efficacy is context specific, I did not collapse
data for reading, writing, and mathematics into a single school
achievement rating. The range of scores within each grade and subject
reflects the design of the study to obtain a sample that represented low to
high achievement levels.
Factor Analysis for Teacher Efficacy
Confirmatory factor analysis was run on the data collected from the
389 teacher respondents. The Kaiser-Meyer-Olkin (KMO) measure of
sampling adequacy for this dataset was .74, which is good, indicating that
the results should be informative. Principal Components with Varimax with
Kaiser Normalization rotation found that all items loaded as expected on
the two factors, TE and PTE, using a .40 cutoff.
Cronbachs Alpha for the five TE items was .70. Examination of the
item-total statistics on the SPSS output showed that if Question 5 were
deleted the Alpha would increase to .74. This suggested that Question 5
may not have as strong a relationship to the TE construct as the other
items. The reliability coefficient for the five items that measure PTE was
.69, indicating that this factor had adequate reliability to be useful in
research (Gregory, 1996). Item-total statistics information revealed that
the Cronbachs Alpha would not change significantly by the deletion of any
of the five items.

From the examination of the analysis of the two generally accepted
dimensions of teacher efficacy and the reliability analysis, I noted that one
item did not fit neatly into either of those two factors of TE and PTE.
Follow-up exploratory factor analysis of the TES-S items resulted in the
identification of three factors. One of the factors was identical to that
identified by Hoy to measure TE. The items that were considered to
measure PTE separated into two factors. The first included the items of
the TES-S which were previously considered to measure PTE minus
Question 5, which became a one-item factor. Question 5 asks the
respondent to indicate degree of agreement or disagreement to the
statement, If parents would do more for their children, I could do more.
This third factor appeared to relate to teacher trust in parents and, thus,
for the purposes of this study was labeled as Parent Trust (PT). Table 3
shows the items in each scale with associated factor loading and
Cronbachs Alpha. Based on this information, I decided to use these three
factors to represent teacher efficacy in subsequent analyses.
Data Assumptions
Each of the statistical analyses, correlations, and analyses of
variance used to address the research questions has assumptions
regarding the distributions of the variables involved that need to be met in

order for the results to be reliable. The following is a discussion of how the
testing of the assumptions of these methods was done.
Table 3
Principal Component Analysis Using Varimax Rotation
with Kaiser Normalization
Question TE PTE PT
Q7 .73
Q8 .69
Q9 .62
Q3 .57
Q1 .83
Q2 .77
Q4 .71
Q10 .54
Q5 .78
Cronbachs Alpha .69 .74
Bivariate correlations assume the relationship between the two
variables being correlated is linear. Additionally, the distributions of
variables are assumed to be normal. When both the variables are fairly
continuous and their distributions are not significantly different from
normal, the resulting correlation coefficient is considered to be a good
summary of the relationship.

The assumptions for the One-way Analysis of Variance are as
(a) random,
(b) within subgroups the variables are normally distributed
(normality), and
(c) the subgroups variances are equal (homogeneity of variance).
The normality of the variables used in this study was examined by
obtaining skewness and kurtosis statistics and histograms. Scatterplots
were used to examine linearity. Levenes test was used to assess
homogeneity of variance. Most techniques are considered to be robust of
slight and moderate violations of their associated assumptions.
It should be noted that in the present study, the violation of the
normality assumption for teachers years of teaching experience, teachers
level of education, and teachers years at school was determined to be
significant. Therefore, I decided to examine the relationships involving
these variables using One-way Analysis of Variance.
Results for Question 1: Teacher Efficacy
Question 1 is What are the relationships among selected teacher
characteristics, school characteristics, and efficacy for teacher level data?
The preliminary viewing of the data revealed that the characteristics of the
teachers in the sample were quite skewed. Since the sample had been

selected based on school characteristics rather than teacher
characteristics, this was not a surprising outcome. However, it did
influence my decisions on how to use the data since it did not seem
appropriate to use the data as continuous variables with normal
distribution. As noted previously, information about how long a teacher
has been at a particular school may be influenced by the age of the school
and the growth in the student population for a school. This study did not
attempt to control for the factors of age of the school and growth in student
numbers, other than to exclude schools that were less than five years old.
The following sections present the data analyzed to answer the research
Question 1.a.
What is the relationship between Teacher Efficacy and teachers
years of teaching experience?
Examination of skewness and kurtosis statistics and histograms for
teachers years of teaching experience showed that this variable was
significantly nonnormal. This information for the 3 factors comprising
Teacher Efficacy (TE, PTE, and PT) indicated that TE and PTE were
normal, and that PT was slightly skewed but was not severe. Given the
skewness of the teacher characteristics and the way they distributed, I
decided to collapse some data into fewer categories and to recode this

variable into four categories and continue this analysis using One-way
Analysis of Variance.
I chose to cluster years teaching experience into beginning
teachers (0-5 years), tenured teachers (6-10 years), mid-career teachers
(11-15 years), and veteran teachers (16 years or more). Table 4 presents
the means and standard deviations for teacher efficacy (TE, PTE, PT) by
years of experience.
Table 4
Years of Experience Related to Teacher Efficacy
Variable no. M SD
Teaching Efficacy
1 -5 years 105 4.11 0.81
6-10 years 86 4.01 0.99
11-15 years 69 4.05 1.04
16+ years 126 3.95 1.17
Personal Teaching Efficacy
1 -5 years 106 4.87 0.59
6-10 years 87 4.99 0.49
11-15 years 67 5.19 0.58
16+ years 124 5.05 0.61
Parent Trust
1-5 years 106 2.50 1.21
6-10 years 87 2.75 1.08
11-15 years 69 2.61 1.29
16+ years 125 2.90 1.49
The Levenes test for homogeneity of variance was significant for
TE and PT. However, the One-way ANOVA is considered to still be robust

in light of this violation. The results of the ANOVAs found that there were
no group differences for TE, F (3,382) = 0.49, p =.69 and PT, F (3,383) =
2.03, p = .11. Significant group differences were found for PTE, F (3,380)
= 4.50, p = .004. Scheffes post hoc multiple comparisons found the
beginners (1-5 years) to have significantly lower PTE than the mid-career
(10-15 years) teachers (mean difference = -.32, p = .005). This suggests
that the personal teaching experiences of teachers may influence their
efficacy beliefs.
Question 1.b.
What is the relationship between teacher efficacy and teachers
level of education?
Initial examination revealed that 65% of the teachers had graduate
education. Examination of skewness and kurtosis statistics and
histograms for teachers level of education showed that this variable was
significantly nonnormal. Because of this nonnormality, it was decided to
recode this variable into three categories and continue this analysis using
One-way Analysis of Variance.
The Levenes test for homogeneity of variance was significant
for TE (p = .01). However, the One-way ANOVA is considered still to be
robust in light of this violation. The results of the ANOVAs showed no
group differences for PTE, F (2,380) = 2.77, p = .06.and PT, F (2,383)

=0.35, p = .71 and TE, F (2,382) = 0.24, p =.79. Therefore, the level of
education of teachers in this study did not make a significant contribution
to teacher efficacy. Table 5 presents the means and standard deviations
for teacher efficacy by educational level.
Table 5
Teacher Education Level Related to Teacher Efficacy
Teacher Efficacy Factor no. M SD
Teaching Efficacy
BA 133 3.98 0.93
BA+, MA 101 4.07 0.97
MA+, Ph.D./Ed.D. 151 4.03 1.13
Personal Teaching Efficacy
BA 132 4.92 0.52
BA+, MA 100 5.04 0.61
MA+, Ph.D./Ed.D. 151 5.07 0.60
Parent T rust
BA 134 2.64 1.22
BA+, MA 101 2.68 1.26
MA+, Ph.D./Ed.D. 151 2.77 1.40
Question 1.c.
What is the relationship between collective efficacy and the
teachers years in the school?
Preliminary examination of the data for number of years a teacher
had been at the current school revealed that 41.1 % of the teachers had
been at the school three years or less. Examination of skewness and
kurtosis statistics and histograms for teachers years in the school showed

that this variable was significantly nonnormal. Because of the
nonnormality observed for teachers years in the school, it was decided to
recode this variable into four categories and continue this analysis using
One-way Analysis of Variance. I collapsed the data into 0-3 years (new to
school), 4-7 years (moderately acculturated), 8-15 years (acculturated),
and 16+ years (strongly acculturated). The Levenes test for homogeneity
of variance indicated no violation of the homogeneity assumption (p = .19
is >p = .05). The results of the ANOVA found no differences for CTE, F
(3,385) = .63, p= .60. Table 6 presents the means and standard deviations
for collective efficacy by teachers years in the school.
Table 6
Teachers Years in School Related to Collective Efficacy
Collective Efficacy no. M SD
0-3 years 160 4.83 0.60
4-7 years 103 4.78 0.69
8-15 years 84 4.81 0.62
16+ years 42 4.69 0.56
Question 1.d.
What is the relationship between SES and Teacher Efficacy?
The Pearson correlations between SES and TE (r= -.006, p = .91), PTE (r
= -.10, p = .06), and PT (r= -.10, p = .05) did not differ significantly from 0,
indicating no relationship. Examination of scatterplots confirmed that there