THE EFFECTS OF JOB EXPECTATIONS AND COLLEAGUES
ON NEW TEACHERS’ COMMITMENT LEVELS
By
Mark Low

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILSOPHY
Curriculum, Teaching, and Education Policy
2012

ABSTRACT
THE EFFECTS OF JOB EXPECTATIONS AND COLLEAGUES
ON NEW TEACHERS’ COMMITMENT LEVELS
By
Mark Low
Teachers’ levels of commitment to their schools and to the teaching profession,
particularly among early career teachers, is a concern facing educational institutions across the
United States; this concern is apparent in both public and private schools. Research on teacher
working conditions and teacher retention suggests that districts and dioceses have the potential to
alleviate some of the struggles of early career teachers, thereby easing new teachers’ entry into
school communities and the teaching profession. However, schools are complex social
organizations, and much of the research on this topic misses important features of life in schools.
As well, education researchers have generally focused more on the experiences of beginning
teachers after they have been hired in schools and less on the expectations that those beginning
teachers bring to their new positions.
The author drew on theories of realistic job previews and met expectations, as well as
social capital and social networks, to study the effects of teachers’ expectations and social
networks on their levels of commitment to their schools and the profession. The data came from
survey responses from 119 early-career teachers and 248 of their mentors and colleagues across
44 public and Catholic high schools. To study the effects of the predictors on the commitmentto-school and commitment-to-profession outcomes, the author constructed hierarchical linear and
hierarchical logistic regression models, respectively.
Analyses indicated that the extent to which early-career teachers’ expectations for their
work lives match their experiences in schools affects their spring levels of commitment to their

schools. Specifically, increasing the extent of met expectations increases the level of school
commitment even after controlling for the prior level of commitment. Analyses indicated that
having a mentor, the frequency with which mentors and mentees communicated, and whether the
mentor and mentee taught the same content area were all statistically significantly related to
early-career teachers’ levels of commitment to their schools in the spring even after controlling
for their prior levels of commitment. In particular, having a mentor and having a mentor who
teaches the same content area as the mentee each associate with a lower level of commitment to
the school in the spring. At the same time, communicating more frequently with mentors is
associated with higher levels of commitment to their schools in the spring for early-career
teachers.
The null hypotheses that the effects of the met-expectations predictor, as well as the set of
social network predictors, on the commitment-to-profession was equal to zero could not be
rejected based on the analyses in this study. Limitations of the study include not being able to
rule out the possibility that the participants underwent natural development during the year of the
study and it was that natural development that drove the study findings, as opposed to the
predictors of interest. As well, the sample was not selected at random and so the possibility that
the findings were an artifact of an interaction between the selection of the sample and the various
predictors cannot be ruled out.

© 2012, Mark Low

ACKNOWLEDGMENTS
I had the good fortune of meeting Dr. Peter Youngs within the first few days that I began
graduate school. He has been an invaluable source of support and encouragement throughout my
studies at Michigan State. His intensity in listening and his thoughtful review of papers and ideas
are qualities to be emulated. Dr. Ken Frank played an instrumental role in the conceptualization
of this study. In particular, his invitation to a conference on social networks in the fall of 2006
led to the “aha!” moment in which I imagined a dissertation that was both relevant and
interesting. Dr. Mary Kennedy led a course on teacher quality that deeply shaped my worldview
on teachers and their roles in schools. I have very much appreciated her honest and direct review
of my work. Dr. Gary Sykes is a model colleague, intellectual, and course instructor. I have
thoroughly enjoyed his presence and insightfulness in the classroom, the meeting room, and at
the driving range.

v

TABLE OF CONTENTS
LIST OF TABLES

viii

LIST OF FIGURES

xi

CHAPTER ONE
INTRODUCTION AND SUMMARY OF FINDINGS
Introduction
Design
Findings
Discussion

1
1
3
5
9

CHAPTER TWO
THE INFLUENCE OF MET EXPECTATIONS ON BEGINNING TEACHERS’
COMMITMENT LEVELS IN PUBLIC AND CATHOLIC SCHOOLS
Introduction
Literature Review
Organizational Commitment
Research on Teacher Commitment and Teachers’ Job Expectations
Theoretical Framework
Methods
Participants
Procedures
Measures
Results
Descriptive Statistics
Inferential Statistics
Discussion
Contributions of the Study
Limitations of the Study

13
13
14
14
16
23
25
25
26
27
31
31
34
51
51
54

CHAPTER THREE
THE INFLUENCE OF COLLEGIAL NETWORKS ON BEGINNING TEACHERS’
COMMITMENT LEVELS IN PUBLIC AND CATHOLIC SCHOOLS
Introduction
Framework
Social Capital
Social Networks
Hypotheses
Methods
Participants
Procedures
Measures
Results

56
56
58
58
61
63
64
64
65
67
69

vi

Descriptive Statistics
Inferential Statistics
Discussion
Contributions of the Study
Limitations of the Study

69
76
105
105
112

FOOTNOTES

116

APPENDIX A
EARLY-CAREER TEACHER SURVEY #1, FALL 2007

119

APPENDIX B
EARLY-CAREER TEACHER SURVEY #2, SPRING 2008

126

APPENDIX C
MENTOR/COLLEAGUE SURVEY, SPRING 2008

133

APPENDIX D
SUPPLEMENTAL TABLES FOR CHAPTER TWO

138

APPENDIX E
SUPPLEMENTAL TABLES FOR CHAPTER THREE

140

REFERENCES

144

vii

LIST OF TABLES

Table 2.1:

Technical Properties of Commitment-to-School Measure for Fall and Spring

29

Table 2.2:

Technical Properties of Met Expectations Measure

30

Table 2.3:

Results of Fully Unconditional Model for Commitment-to-School Outcome

36

Table 2.4:

Descriptive Statistics for Model Estimating Effects of Met Expectations on
School Commitment

39

Table 2.5

Correlation Matrix for Model Estimating Effects of Met Expectations on
School Commitment

40

Table 2.6:

Results of Model Estimating Effects of Met Expectations on School
Commitment

42

Table 2.7:

Results of Fully Unconditional Model for Commitment-to-Profession
Outcome

47

Table 2.8:

Descriptive Statistics for Model Estimating Effects of Met Expectations on
Professional Commitment

49

Table 2.9:

Correlation Matrix for Model Estimating Effects of Met Expectations on
Professional Commitment

50

Table 2.10:

Results of Model Estimating Effects of Met Expectations on Professional
Commitment

51

Table 3.1:

Technical Properties of Commitment-to-School Measure for Fall and Spring

68

Table 3.2:

Results of Fully Unconditional Model for Commitment-to-School Outcome

78

Table 3.3:

Descriptive Statistics for Model Estimating Effects of Communicating with
Mentors and Colleagues on School Commitment

81

Table 3.4:

Correlation Matrix for Model Estimating Effects of Communicating with
Mentors and Colleagues on School Commitment

82

Table 3.5:

Results of Model Estimating Effects of Communicating with Mentors and
Colleagues on School Commitment

84

Table 3.6:

Descriptive Statistics for Model Estimating Effects of Mentor
Characteristics on School Commitment

87

Table 3.7:

Correlation Matrix for Model Estimating Effects of Mentor Characteristics
on School Commitment

88

viii

Table 3.8a:

Results of Models Estimating Effects of Content-Area and Grade-Level
Matches with Mentors on School Commitment

90

Table 3.8b:

Results of Models Estimating Effects of Mentor Commitment and Influence
on School Commitment

91

Table 3.9:

Descriptive Statistics for Model Estimating Effects of Colleague
Characteristics on School Commitment

93

Table 3.10:

Correlation Matrix for Model Estimating Effects of Colleague
Characteristics on School Commitment

94

Table 3.11a: Results of Models Estimating Effects of Content-Area and Grade-Level
Matches with Colleagues on School Commitment

96

Table 3.11b: Results of Models Estimating Effects of Colleague Commitment and
Influence on School Commitment

97

Table 3.12:

Results of Fully Unconditional Model for Commitment-to-Profession
Outcome

100

Table 3.13:

Descriptive Statistics for Model Estimating Effects of Communicating with
Mentors and Colleagues on Professional Commitment

102

Table 3.14:

Correlation Matrix for Model Estimating Effects of Communicating with
Mentors and Colleagues on Professional Commitment

103

Table 3.15:

Results of Model Estimating Effects of Communicating with Mentors and
Colleagues on Professional Commitment

104

Table D1:

Hierarchical Form of Equation 2.1

138

Table D2:

Hierarchical Form of Equation 2.2

138

Table D3:

Hierarchical Form of Equation 2.3

139

Table D4:

Hierarchical Form of Equation 2.4

139

Table E1:

Hierarchical Form of Equation 3.1

140

Table E2:

Hierarchical Form of Equation 3.2

141

Table E3:

Hierarchical Form of Equation 3.3

142

Table E4:

Hierarchical Form of Equation 3.4

142

Table E5:

Hierarchical Form of Equation 3.5

143

ix

Table E6:

Hierarchical Form of Equation 3.6

x

143

LIST OF FIGURES

Figure 2.1:

Response frequencies of early-career teachers when asked in the fall: “If
you could go back to your college days and start over again, would you
become a teacher or not?”

44

Figure 2.2:

Response frequencies of early-career teachers when asked in the spring: “If
you could go back to your college days and start over again, would you
become a teacher or not?”

44

Figure 3.1:

Proportions of early-career teachers who indicated on the fall and the spring
surveys that they had a mentor or that they interacted with at least one other
colleague at their school on professional matters.

72

Figure 3.2:

Response frequencies of early-career teachers when asked in the fall: “If
you could go back to your college days and start over again, would you
become a teacher or not?”

99

Figure 3.3:

Response frequencies of early-career teachers when asked in the spring: “If
you could go back to your college days and start over again, would you
become a teacher or not?”

99

xi

CHAPTER ONE: INTRODUCTION AND SUMMARY OF FINDINGS
Introduction
Teachers’ levels of commitment to their schools and to the teaching profession,
particularly among early career teachers, is a concern facing educational institutions across the
country; this concern is apparent in both public and private schools. The National Center for
Education Statistics (1997) found that only 18 percent of teachers in public schools and only 37
percent of teachers in Catholic schools expressed high levels of teacher commitment; the report
defined commitment as the “degree of positive, affective bond between the teacher and the
school” (p. 2). Low levels of teacher commitment are problematic for schools and the teaching
profession because of commitment’s connection to teacher goals, effort, persistence, and
effectiveness (Ebmeier, 2003; National Center for Educational Statistics, 1997). Further, teacher
commitment has been linked to teachers’ intentions to remain in their schools and the profession
(Weiss, 1999).
Research on teacher working conditions and teacher retention suggests that districts and
dioceses have the potential to alleviate some of the struggles of early career teachers, thereby
easing new teachers’ entry into school communities and the teaching profession (see, e.g.,
Brewer, 1996; Ebmeier, 2003; Smith & Ingersoll, 2004). That is, through their hiring and
induction programs, districts and dioceses could affect the expectations and experiences of early
career teachers.
However, schools are complex social organizations, and much of the research on this
topic misses important features of life in schools. For instance, teacher responses to district and
diocesan policies are mediated through local processes found in schools, and these local
processes can affect new teachers’ beliefs and actions (Coburn, 2003). Further, because of the

1

multitude of relationships that exist in schools, teachers receive support but also experience
pressure from their colleagues and employers in their daily work lives (Bidwell, 2000). In an
effort to better understand the supports provided to new teachers, as well as the pressures placed
upon new teachers, this dissertation drew on theories of social capital (Coleman, 1988; Lin,
1999; Portes, 1998).
As well, education researchers have generally focused more on the experiences of
beginning teachers after they have been hired in schools and less on the expectations that those
beginning teachers brought to their new positions. Reviews in the field of organizational
psychology (see, e.g., Porter & Steers, 1973; Wanous, Poland, Premack, & Davis, 1992), as well
as relatively recent research in the field of education (Liu & Johnson, 2006), provide an
argument for considering teachers’ pre-employment experiences when trying to understand their
post-employment labor market decisions and commitment levels. To explore the effects of earlycareer teachers’ expectations for their work lives on their commitment levels, this dissertation
drew on the concept of realistic job previews (Breaugh, 1983; Hom, Griffeth, Palich, & Bracker,
1998; Phillips, 1998).
This dissertation contributes to the existing literature concerned with the roles of met
expectations and social networks on early career teachers’ commitment levels. In short, this
study investigated two questions. The first question that I investigated in this study was, “What
are the effects of early career teachers’ expectations for their work lives on their levels of
commitment to their schools and the profession?” I report on how I explored the first question, as
well as what I learned, in the second chapter of this document. The second question that I
investigated in this study was, “What are the effects of mentoring and social networks on early
career teachers’ levels of commitment to their schools and the profession?” I report on how I

2

explored that question, as well as what I learned, in the third chapter of this document. In the
following sections of this chapter, I describe the methods that I used and the findings that I
obtained in this study.
Design
In both of the essays in this dissertation, I report on the responses from 119 early-career
high school teachers who completed two questionnaires during the 2007-2008 academic year. I
use the term early career to indicate a teacher in the first or second year of full-time teaching. In
the first essay on the effects of teachers’ met expectations on commitment, I limit my analyses to
the early-career teachers. In the second essay, I include the responses from 248 of the earlycareer teachers’ mentors and colleagues in the analyses. The mentors and colleagues represent
the early-career teachers’ within-school social networks.
Each of the early-career participants completed two surveys; they submitted one between
November 2007 and January 2008, and they submitted the other between May and June of 2008.
Each of the mentor and colleague participants completed one survey; they submitted it between
February and April of 2008. In total, study participants came from 24 Catholic schools in eight
dioceses across three states and 20 public high schools in nine districts across two states.
Through interactions with district and diocesan administrators, my dissertation director
and I obtained the names, positions, and contact information (school postal and email addresses)
for first- and second-year teachers. I employed a five-contact procedure (Dillman, 2007) in the
recruitment of teachers to participate in the study. For the early-career teachers, I achieved a
response rate of 64 percent for the fall survey administration, and in the spring, I retained 75
percent of the teachers who had completed a fall questionnaire. In total, I obtained responses

3

from 119 teachers on both the fall and spring questionnaires, which corresponds to an overall
study response rate of 48 percent for the early-career teachers.
I generated the social network sample of the study by including two questions in the
early-career teacher survey that asked them to identify their official mentor and up to six withinschool colleagues with whom they interacted on professional matters. I received completed
surveys from 228 mentors and colleagues, which corresponds to a response rate of 70 percent for
this portion of the study.
The first early-career teacher survey contained questions on teachers’ teaching experience
and assignments, the frequency and nature of their interactions with their mentors and
colleagues, the frequency with which they had participated in various professional development
activities, their teaching practices, their perceptions of their schools and departments, their hiring
and work experiences, and their career plans. The second early-career teacher survey contained
similar items to the first survey as well as questions on teachers’ preparations for teaching, their
experiences with teachers’ associations, and their race and gender. The mentor-colleague survey
consisted of questions on teachers’ teaching positions and practices, their perceptions of their
schools and departments, their work experiences, their preparations for teaching, their career
plans, and their race and gender. In total, the first early-career teacher survey contained
approximately 170 items; the second early-career teacher survey included about 240 items; the
mentor-colleague teacher survey contained approximately 150 items.
In the first essay, the predictor of interest is a measure of the degree to which teachers’
expectations of their jobs matched their actual experiences in the job. In the second essay, the
predictors of interest included the frequency with which early-career teachers communicated
with mentors and colleagues and the characteristics of their mentors and colleagues. In both

4

essays, I used hierarchical modeling software (Raudenbush, Bryk, Cheong, Congdon, & du Toit,
2004) to evaluate the effects of the predictors on teachers’ school and professional commitment.
In addition to the predictors of interest, I included several covariates in the models, including a
measure of teachers’ prior commitment levels. I used a linear model to evaluate the commitmentto-school outcome, and I used a logistic model to evaluate the commitment-to-profession
outcome. I chose those model structures because the commitment-to-school measure is on an
interval scale, whereas the commitment-to-profession measure is binary. In the next section, I
provide an overview of the findings from each essay.

Findings
In each essay, I began the process of building models by evaluating a fully unconditional
model for the commitment-to-school outcome in order to determine the extent to which
participants’ responses varied across schools. The sample sizes in the two essays were slightly
different, but the extent of variation among schools was consistent: I found that between 22
percent and 24 percent of the variation in the commitment-to-school outcome existed between
schools. The school effect was statistically significant in each case.
I evaluated the effects of the met expectations predictor on the commitment-to-school
outcome in a hierarchical linear model with several covariates. The teacher-level covariates that I
included were the fall measure of commitment to school and an indicator for whether the earlycareer teacher obtained a bachelor’s degree through a teacher preparation program. The schoollevel covariate in the model was an indicator of whether 40 percent or more of the students at the
school belonged to a racial minority group.

5

I excluded a number of potential covariates because they did not explain at least one
percent of the variation in the outcome. The teacher-level covariates that I excluded were
indicators of whether the early-career teacher held a standard teaching certificate from the state,
was female, was Caucasian, was a first-year teacher, and taught math or science. As well, I
excluded school-level indicators for whether the school was public, had 1200 or more students,
and was located in a central city (as opposed to the suburbs or a rural area).
In the met expectations model, each of the teacher-level predictors is statistically
significantly related to the dependent variable. Specifically, if all of the other predictors are held
constant, a one percent of the scale range increase in the degree of commitment to the school in
the fall is expected to result in a .76 (less than one) percent of the scale range increase in the
degree of commitment to the school in the spring. As well, a teacher who received a bachelor’s
degree through a teacher preparation program is expected to have a level of commitment to the
school in the spring that is 7.10 percent of the scale range greater than a teacher who did not
receive a bachelor’s degree through a teacher preparation program, ceteris paribus. And for the
predictor of interest, a one percent of the scale range increase in the degree of met expectations is
expected to yield a .20 (less than one) percent of the scale range increase in the spring level of
school commitment. In terms of proportion of variance explained, the prior level of commitment,
the indicator for obtaining a bachelor’s degree through a teacher preparation program, and the
degree of met expectations explain 49, 3, and 3 percent of the variance in the outcome,
respectively.
I evaluated the effects of the social network predictors on the commitment-to-school
outcome in several different hierarchical linear models. The social network predictors that I
tested included whether an early-career teacher had a mentor, communicated with colleagues,

6

and taught the same content areas or grade levels as the mentor or colleagues. As well, I explored
the effects of the frequency with which the early-career teachers interacted with mentors and
colleagues, the commitment levels of the mentors and colleagues, and the interaction between
the commitment levels of the mentors and colleagues and the frequency with which the earlycareer teacher communicated with the mentors and colleagues. I controlled for early-career
teachers’ prior levels of commitment to their schools in each of the models. As well, if the model
had enough power, I included additional teacher-level and school-level control variables.
The social network predictors that are statistically significantly related to the spring
measure of school commitment include the indicator for whether an early-career teacher has a
mentor, the frequency with which the early-career teacher communicates with a mentor, and the
indicator for whether the mentor taught the same content area as the early-career teacher. The pvalue of the indicator for whether an early-career teacher communicates with colleagues is .112.
Given the size of the sample and the expected effect size for the network predictors, a qualified
interpretation of the coefficient for the indicator of communicating with colleagues is not
unwarranted.
In terms of expected effects on the commitment-to-school outcome, an early-career
teacher who has a mentor is anticipated to have a spring level of school commitment that is 12.5
percent of the scale range less than an early-career teacher who does not have a mentor, ceteris
paribus. However, a one percent increase in the extent to which an early-career teacher
communicates with a mentor is expected to increase the spring level of commitment to school by
.21 percent of the scale range if all of the other variables remain constant. An early-career
teacher who has a mentor who teaches the same content area that they do is expected, ceteris
paribus, to have a spring level of commitment to the school that is 12.4 percent of the scale range

7

lower than an early-career teacher who has a mentor who does not teach the same content area
that they do. Last, and cautiously, an early-career teacher who communicates with colleagues is
expected to have a higher level of spring commitment to the school than early-career teachers
who do not communicate with colleagues. The proportion of variance in the outcome explained
by each of the network predictors is 7 percent for the indicator of having a mentor, 21 percent for
the frequency of communicating with a mentor, 4 percent for the indicator of a match in content
areas for the mentor and mentee, and 1 percent for communicating with colleagues.
For the covariates, I found that the fall measure of an early-career teacher’s level of
commitment to the school, as well as the indicator for whether an early-career teacher received a
bachelor’s degree through a teacher preparation program, are statistically significantly related to
the outcome. Specifically, a one percent of the scale range increase in the level of commitment to
the school in the fall is expected to yield between a .741 percent of the scale range and .851
percent of the scale range increase in the level of commitment to the school in the spring. And an
early-career teacher who received a bachelor’s degree through a teacher preparation program is
expected to have a level of commitment in the spring that is between 6.5 percent of the scale
range and 8.4 percent of the scale range greater than an early-career teacher who did not receive
a bachelor’s degree through a teacher preparation program, ceteris paribus. Last, across the
models, the fall measure of an early-career teacher’s level of commitment to the school explains
between 26 percent and 56 percent of the variance in the outcome, while the indicator for
whether an early-career teacher received a bachelor’s degree through a teacher preparation
program explains between 2 percent and 5 percent of the variance in the outcome.
In building models to evaluate the effects of the predictors of interest on the commitmentto-profession outcome, I began by evaluating a fully unconditional Bernoulli sampling model.

8

The results of that model indicate that an average of 76 percent of the early-career teachers
reported being committed to the profession in the spring across schools. In the full models that
included control variables, I could not reject the null hypotheses that the effect of any of the
predictors of interest was different from zero. That is, I did not find any statistically significant
relationships between the commitment-to-profession outcome and the predictors of interest.
Discussion
For the concept of met expectations, I conclude from the analyses that the concept is
statistically significantly related to the commitment-to-school outcome but not the commitmentto-profession outcome. For the former, I found that the better the match between teachers’
expectations and experiences, the more committed the teachers reported being to their schools.
The finding that met expectations affect early-career teachers’ levels of commitment to
their schools is consistent with research on the concept in the field of organizational psychology
(Wanous, Poland, Premack, & Davis, 1992). One of the contributions of this study is applying
the concept of met expectations to the field of education. In doing so, I build upon foundational
work by Liu and Johnson (2006). However, this study goes beyond prior work on met
expectations by arguing that teachers’ development of expectations for their work lives begins
before the hiring process.
The retention of the null hypothesis that the effect of met expectations on early-career
teachers’ levels of commitment to the profession is zero is not necessarily a counterpoint to the
application of met expectations to the context of schools. One possible explanation for the null
finding could be a misalignment between the measure of met expectations and the concept of
commitment to the profession. Another possible explanation for the null finding on the

9

relationship between met expectations and professional commitment is the limited power in the
statistical model to detect an effect.
For the social network predictors, I conclude that having a mentor negatively affects the
spring commitment levels of early-career teachers, as does having a mentor who teaches the
same content area as the early-career teachers. However, increasing the frequency with which
early-career teachers communicate with mentors positively affects the early-career teachers’
levels of commitment in the spring. Last, I tentatively conclude that communicating with
colleagues positively affects the spring commitment level of early-career teachers.
The effect of having a mentor is in the opposite direction of what I predicted at the outset
of the study, while the possible effect of communicating with colleagues is in the direction that I
predicted. The overall finding that social capital can be both beneficial and harmful is supported
in the literature (Portes, 1998).
One possible explanation for the opposite direction of the effects of mentors and
colleagues could reside in the difference in the orientations of the relationships between the
early-career teachers and their mentor vis-à-vis the early-career teachers and their colleagues. It
could be that mentors, as official agents of the school or district/diocese, deliver resources to the
mentees that are intended to assist them in their development, but which have the effect of
causing them to question their beliefs about curriculum and students, as well as placing
constraints on the mentees’ autonomy. It could be that the early-career teachers seek out
colleagues to help them make sense of the pressures that they receive from the employing unit,
possibly via the mentor.
One potential inconsistency with my assessment that mentors have the effect of providing
more pressure and less support while colleagues provide more support and less pressure is the

10

finding that increasing the amount of communication between mentors and mentees results in an
expectation of higher levels of spring commitment to the school in the mentees. One resolution
for this seeming inconsistency is the possibility that the more mentors and mentees interact, the
more supportive the relationship becomes. That is, perhaps increasing the frequency of
interaction between the mentor and the mentee moves the relationship beyond the mentor largely
communicating the expectations of the employing unit and into a relationship in which the
mentor helps the mentee make sense of the expectations of the employing unit.
As well, I conclude that early-career teachers who received a bachelor’s degree through a
teacher preparation program are expected to have higher levels of commitment to their schools in
the spring than those early-career teaches who did not receive a bachelor’s degree through a
teacher preparation program. This finding, while not predicted at the outset of the study, is
consistent with recent research on the effects of teacher education (Ingersoll, Merrill, & May,
2011), and consequently, justifies further study on the relationship between teacher preparation
and teachers’ commitment levels.
Last, while this study makes important contributions to the literature on within-school
networks and early-career teachers’ commitment levels, it has important limitations. For one,
because I employed a pretest-posttest design, I cannot rule out the possibility that participants
underwent natural development over the course of the study and that this development could be
driving the changes being detected in the study (Campbell & Stanley, 1966). For another,
because the sample was not selected randomly, I cannot rule out the possibility that the findings
are an artifact of the interaction between the selection of the sample and the various predictors
that I evaluated in this study. While my study is not representative of any state or federal
contexts, my findings do justify the future commitment of additional resources to further

11

evaluate the effects of met expectations, mentors and colleagues, and teacher preparation on
early-career teachers’ commitment levels. Future studies with representative samples would
likely lead to the production of guidance for policymakers.

12

CHAPTER TWO: THE INFLUENCE OF MET EXPECTATIONS ON BEGINNING
TEACHERS’ COMMITMENT LEVELS IN PUBLIC AND CATHOLIC SCHOOLS
Introduction
Many beginning high school teachers experience reality shock when they enter the
profession and assume full responsibility for teaching classes (Veenman, 1984). Novice teachers
must plan units and lessons, translate subject matter knowledge into curriculum appropriate for
students, establish and maintain productive learning environments, assess students’ work, and
address individual and cultural differences (Feiman-Nemser, 2001). In response to concerns
about new teacher retention and effectiveness, many states and districts have implemented
mentoring and induction programs (Education Week, 2010; Kaufmann, 2007; Smith & Ingersoll,
2004). But researchers and policy makers have generally focused less on what beginning
teachers expect their jobs to be like or the degree to which their expectations match their
experiences.
Recent research by Liu and Johnson (2006) has begun to address this shortcoming in the
research literature. Based on surveys of close to 500 novice teachers in four states, they reported
that new teachers’ job expectations were rarely well aligned with their actual work experiences
because they were typically hired in information-poor environments. In particular, Liu and
Johnson found that for most newly hired teachers, the hiring process emphasized interviews with
principals or district administrators and reviews of their credentials, with fewer opportunities for
them to interview with teachers or department chairs or to be observed during classroom
instruction. The authors argue that interviewing with other teachers and being observed while
teaching could enable teaching candidates and the schools that hire them to better assess their fit
with their schools.

13

This chapter builds on Liu and Johnson’s study to examine whether the extent to which
beginning teachers’ job expectations are met is associated with their commitment levels. I report
on a study of 119 first- and second-year teachers in public and Catholic high schools in three
states. I find that the degree of met expectations is associated with novices’ commitment to their
schools but not to the profession. This suggests that hiring processes that are information-rich
can lead to teachers who will work to promote their school’s mission and goals and who will
seek to remain members of their school faculties (Mowday, Steers, & Porter, 1979).
In the first section of this paper, I define teacher commitment and discuss its potential
implications for teacher hiring. The second section reviews the research literature on teacher
commitment and job expectations. In the third section, I introduce the theoretical framework,
which shaped the research design and data analysis. Fourth, I describe the methods, including
district and diocesan samples, teacher samples, and research instrument. In the fifth section, I
present a series of regression analyses that I conducted to assess associations between degree of
job expectations and novice teacher commitment. In the sixth section, I interpret the main
findings. Finally, I conclude by addressing how this study builds on other research on teacher
commitment and job expectations, consider the implications of this analysis for future research,
and identify the study’s limitations.
Literature Review
Organizational Commitment
As a concept that is useful to the study of organizations and occupations, scholars have
been attending to the definitions, antecedents, and consequences of commitment since at least the
middle of the twentieth century (see, e.g., Becker & Carper, 1956). Commitment has generally
been recognized as having two facets: observable actions and psychological dispositions

14

(Becker, 1960; Mowday, Steers, & Porter, 1979). In terms of the former, individuals choose
particular actions over others because of their attachment to organizations. With regard to the
latter, individuals hold certain beliefs over others because of their connection to organizations.
Mowday et al. (1979) provided a comprehensive, three-part definition of organizational
commitment: “(1) a strong belief in and acceptance of the organization’s goals and values; (2) a
willingness to exert considerable effort on behalf of the organization; and (3) a strong desire to
maintain membership in the organization” (p. 226). In short, committed individuals work to
maintain and promote the organization’s mission and goals, and they want to continue to be a
member of the organization; this definition has served as the understanding of commitment in
this study.
Assuming that teachers are active agents who make conscious decisions about how to
teach and whether to be employed, it seems reasonable that those teachers who are committed to
their schools would make efforts to promote the educational mission of their schools and seek to
retain employment with their schools. The characterization of commitment offered by Mowday,
Steers and Porter (1979) has intuitive policy implications for educational administrators with
respect to hiring new teachers. First, the characterization suggests that administrators should
consider whether job applicants support the mission of the school/district. Second, the
characterization suggests that administrators should determine whether job applicants are willing
to expend reasonable amounts of effort for the school/district. Third, the characterization
suggests that administrators should assess the career aspirations of job applicants. In the next
section, I review the research literature to investigate whether evidence exists to support the
intuitive implications.

15

Research on Teacher Commitment and Teachers’ Job Expectations
An individual likely has multiple personal and professional commitments (Firestone &
Rosenblum, 1988; Gouldner, 1960; Kushman, 1992). After conducting interviews with samples
of administrators, teachers, and students in ten urban high schools across five cities, Firestone
and Rosenblum (1988) concluded that teachers hold commitments to place, content and students.
They argued: “these three different dimensions of commitment provide a basis for different kinds
of behavior” (p. 288). That is, teachers who are committed to students above content display a
different set of behaviors compared to those who are committed to content above students;
similarly, for place. The authors assert that commitment to place results in desires to continue to
be employed and work in a variety of roles for the school, but commitment to place does not
ensure the development of supportive relationships with students or a press for high academic
achievement from students. In contrast, Firestone and Rosenblum contended that a commitment
to content results in serious efforts toward refining and developing the act of teaching; teachers
who are committed to their content press for student learning. Lastly, the authors concluded that
a commitment to students results in authentic care and concern for the wellbeing of students;
such commitment from a number of teachers in a school could result in the development of a
positive and supportive climate for students. It should be noted that the commitments held by
teachers to different objects are not necessarily mutually exclusive. That is, commitment to place
does not preclude commitments to students or content and vice versa.
In addition to their identification of various objects of teachers’ commitment, Firestone
and Rosenblum (1988) also identified various school-level factors that seem to develop and
maintain teachers’ commitment levels. The school factors include teachers finding a sense of
purpose to their work; having opportunities to interact with colleagues and administrators on

16

professional matters; receiving support from administrators, particularly around issues of student
discipline; working in an environment that promotes high expectations of teaching and learning;
and having autonomy in their own classrooms, especially on matters of instruction. Rosenholtz
and Simpson (1990), in their survey of 1200 elementary school teachers in Tennessee, also found
school-level factors to be significant predictors of teachers’ commitment to their schools. In
particular, their findings indicated that teachers’ autonomy with their teaching practices,
teachers’ beliefs that the school environment allowed them to affect student learning, and
principals’ abilities to shield teachers from outside interference and interruptions were each
important predictors of teachers’ organizational commitment (p. 251).
Apart from identifying school-level factors that affect teachers’ commitment levels,
Rosenholtz and Simpson (1990) made a substantial contribution to the literature by separating
the school-level effects for the various career stages among teachers. In short, they found that
teachers’ career stages significantly affect which school-level factors are most salient in
contributing to their commitment to their schools; notably, career stage did not affect the
proportion of variance explained by school-level factors (58 percent for each career stage).
Novice teachers (defined as those teachers in their first five years of their career) were most
influenced by principals’ abilities to support their work with resources and a minimization of
interruptions. In contrast, the commitment levels of experienced teachers were most influenced
by the degree of autonomy that they possessed in their work and the degree to which they
believed the school environment allowed them to affect student learning. Rosenholtz and
Simpson concluded that while experienced and veteran teachers are most affected by schoollevel factors that affect their abilities to teach content matter to students, novice teachers are most
affected by factors that affect their ability to manage their students and classrooms efficiently.

17

This conclusion has implications for where best to target policies intended to enhance the
commitment levels of teachers.
In addition to students, subject matter, and school, teachers are believed to possess a
more general commitment to the profession of teaching itself. In an effort to quantify
commitment to the profession, numerous studies have made use of the National Center for
Education Statistics’ Schools and Staffing Survey (SASS) items that ask teachers about their
satisfaction with their career choice and their plans for remaining in teaching (Ingersoll &
Alsalam, 1997; Riehl & Sipple, 1996; Weiss, 1999). With respect to career choice, teachers are
asked to respond to the question, “if you could go back to your college days and start over again,
would you become a teacher or not?”; with respect to intent to remain in the profession, teachers
are asked to respond to the question, “how long do you plan to remain in teaching?”; for each
question, teachers can select from one of five categories of agreement/disagreement. The SASS
and its companion, the Teacher Follow-up Survey (TFS), have been administered to a nationally
representative sample of schools and teachers every three to six years since 1987.
Recent studies have attempted to identify the antecedents and consequences of teachers’
organizational and professional commitments. Somech and Bogler (2002) collected 983
questionnaires from teachers with various degrees of experience in 52 Israeli middle and high
schools; the study had a low response rate of 41 percent, and so the findings should not be
considered to be representative of all Israeli teachers. Their surveys asked teachers to rate the
extent of their involvement in making decisions about school- and classroom-level policies, to
assess their commitment to the school and to the profession, and to respond to items that
measured the teachers’ organizational citizenship behaviors (OCB). Somech and Bogler found
positive, statistically significant relationships between having opportunities to participate in

18

school- and classroom-level decision-making and being committed to school and profession. As
well, the authors found significant associations between commitment levels and the display of
OCBs. The authors concluded that those teachers with higher levels of organizational and
professional commitment were more likely to direct extra effort toward students and school than
those teachers with lower levels of commitment.
Chan, Lau, Nie, Lim, and Hogan (2008) considered the mediating effects of teacher
efficacy and identification with school on other school-level and individual-level predictors of
teacher commitment. They obtained data from 2130 teachers in 40 primary schools and 1585
teachers in 39 secondary schools in Singapore; they achieved response rates of 85 percent and 78
percent, respectively. Using structural equation modeling (SEM), they found that both efficacy
and school identification mediated the effects of teaching experience, interactions with
colleagues, and perceived organizational politics on teachers’ commitment to the profession.
Their findings support an argument that teachers’ perceptions of themselves and of their
environment play a key role in mediating their responses to individual, classroom, and school
circumstances. For example, Chan et al. concluded that, “although this study did not examine
reward systems (as a way to enhance commitment), the perception of organizational politics may
reflect how the reward system is perceived. And this may be more important than the actual
reward system” (p. 624). In short, it is necessary to consider teachers’ perceptions of their
environments when interpreting teachers’ reactions to those environments.
In sum, researchers have found that several factors can contribute to beginning teacher
commitment, including principal leadership, school culture, reduced interruptions, and autonomy
(Rosenholtz & Simpson, 1990; Weiss, 1999). In addition, teacher efficacy, school identification,
teaching schedules, and supportive school environments seemed to promote commitment among

19

experienced teachers (Chan, Lau, Nie, Lim, & Hogan, 2008; Riehl & Sipple, 1996). But few
studies in education have examined how commitment is affected by the degree to which new
teachers’ job expectations are aligned with their actual work experiences. The remainder of this
section considers research on job expectations and employee outcomes from the fields of
organizational psychology and education.
In a review of research from organizational psychology on employee turnover, Porter and
Steers (1973) argued that met expectations represented a relevant way to understand the effects
of job satisfaction on employee labor market decisions. They presented an analysis in which,
“job satisfaction (was) viewed as the sum total of an individual’s met expectations on the job” (p.
169). Their article included an assessment of the role of met expectations in explaining the
effects of realistic job previews, pecuniary and non-pecuniary reward systems, peer group
interactions, job task assignments, and job autonomy on employee retention decisions. The
authors concluded their article by proposing that employee turnover could be reduced by
enhancing the rewards of the job, increasing employee autonomy, and providing realistic
expectations for the work of the job and its potential rewards.
In their review, Porter and Steers (1973) argued that, “since different employees can have
quite different expectations with respect to payoffs or rewards in a given organizational or work
situation, it would not be anticipated that a given variable (e.g., high pay, unfriendly work
colleagues, etc.) would have a uniform impact on withdrawal decisions” (p. 152). If their
proposition is valid, I would predict inconsistent findings in evaluations of various education
policies, particularly those aimed at retaining teachers. There is some evidence that this is the
case: for example, there is mixed evidence on the effects of teacher salaries on turnover (Kirby &
Grissmer, 1993; Lankford, Loeb, & Wyckoff, 2002; Murnane, Singer, & Willett, 1989; Rickman

20

& Parker, 1990; Rumberger, 1987; Theobald, 1990). A number of factors could account for the
discrepancies in these studies’ findings; publications in the field of organizational psychology
support an argument that one potential cause of the discrepancies in the attrition literature is an
inattention to the role of met expectations (Chatman, 1989; Kristof, 1996; Kristof-Brown,
Zimmerman, & Johnson, 2005).
Wanous, Poland, Premack, and Davis (1992), in their meta-analysis of 31 studies across
various occupations, found that the alignment between the anticipated job experiences and the
actual job experiences of employees affected employees’ job satisfaction, commitment to the
organization, intention to remain on the job, actual job retention, and performance. More
specifically, the authors found a particular ordering to their correlations, which they argued
supported a theoretical model for the operation of met expectations: “the strongest results are
found for job satisfaction and organizational commitment, followed by intent to remain and, last,
job survival” (p. 292). They concluded that met expectations most directly affect employees’
commitment levels and job satisfaction; those attitudes then have consequences on retention
intentions and behaviors. This means that met expectations have a mediated effect on retention
decisions.
Researchers in the field of organizational psychology have studied the effects of
attempting to deliberately provide employees with realistic expectations for their work lives
(Kristof, 1996; Phillips, 1998). The idea, known as a realistic job preview, is to communicate
both the positive and negative aspects of the position to job applicants. Studies have found
realistic job previews to be effective in raising employee commitment and reducing employee
attrition (Kristof-Brown, Zimmerman, & Johnson, 2005; Phillips, 1998). Several mechanisms
have been proposed to explain how realistic job previews operate: these include having

21

applicants select out of the hiring process upon learning more about the job, lowering applicants’
expectations for various aspects of the job, having applicants view their employers as honest and
concerned bosses, and reducing the role ambiguity that is often associated with new positions
(Meglino, Ravlin, & DeNisi, 2000). While the evidence is strong with respect to the effects of
realistic job previews, there is less consistent evidence regarding their underlying mechanisms
(Hom, Griffeth, Palich, & Bracker, 1999).
In the field of education, many researchers have observed that new teachers confront a
“reality shock” as they enter the profession (Veenman, 1984). However, few researchers have
systematically studied what beginning teachers expect their jobs to be like and how their actual
work experiences differ from those expectations. One notable exception to this gap in the
literature on the relationship between teacher expectations and teacher experiences is the work of
Liu and Johnson (2006). They argued that the job expectations and job experiences of new
teachers do not often align well because they are often hired in information-poor environments.
After surveying 486 first- and second-year teachers across four states, they concluded that,
“many schools are not taking full advantage of decentralized hiring and its potential for
improving the amount and quality of information exchanged between teaching candidates and
those who do the hiring” (p. 351). They concluded that the information-poor hiring contexts have
negative consequences on the satisfaction, effectiveness, and retention of new teachers.
Liu and Johnson (2006) made a contribution to research and practice by applying the
concept of realistic job previews to the field of education. As well, their finding that teachers
were often hired in information-poor environments illuminates for the research and policy
communities ways in which hiring processes might be made more informative for both
employers and potential employees. As they observed: “teacher hiring should properly be viewed

22

as a two-way process” (p. 327). The two-way process, Liu and Johnson argued, allows
employees and employers to each assess whether the job and the organization are good matches
for the potential employee.
Theoretical Framework
In this study, I draw on research from the fields of education and organizational
psychology to theorize and empirically investigate the role of met expectations in predicting new
teachers’ commitment levels (Kristof, 1996; Liu & Johnson, 2006; Porter & Steers, 1973). From
a public policy perspective, I see met expectations as a way to bridge the gap in the existing
literature on teacher commitment: I read the current debate as one that attempts to identify either
a set of individual predictors or a set of organizational predictors that could better serve as a
policy lever to increase teacher commitment and decrease teacher attrition. I propose an
alternative approach that considers individual interpretations of organizational characteristics;
that is, I propose an approach that brings individual characteristics together with organizational
characteristics. Specifically, I believe that the expectations that teachers hold for their work lives
mediate their affective responses to their experiences on the job. The concept of met expectations
provides a framework for the evaluation of the relationship between teachers’ expectations and
their actual experiences.
I use met expectations to represent the degree to which a teacher’s expectations for her
job experiences match her actual experiences on the job. These experiences encompass a wide
variety of situations and include interactions with a substantial number of individuals who are
also a part of the school environment. As teachers enter schools, they encounter students and
parents, colleagues, administrators, and instructional and non-instructional support providers. As
well, teachers obtain teaching assignments and curricular materials, participate in professional

23

development activities, serve on committees, attend meetings, and take on general
responsibilities for maintaining order and promoting the mission of the school.
Before their first day on the job, beginning teachers possess expectations for many of the
interactions and encounters that they will experience at various points throughout their year of
teaching. Liu and Johnson (2006) studied the hiring process as one location for the development
of new teachers’ expectations; I build on their work in this study, but where Liu and Johnson
used a set of items to assess the amount and accuracy of information provided to teachers in the
hiring process (and to make an argument about the degree of realism there was in the job
preview), I propose using a similar set of items to assess the degree of match between teachers’
expectations and experiences (and to make an argument about the extent that expectations were
met).
I think about these items differently for a couple of reasons. First, the items ask teachers
to provide a retrospective assessment of knowledge gained during the hiring process; the
participants in my study responded between November and January, while the teachers in the Liu
and Johnson (2006) study responded between April and June. In both cases, the retrospective
nature of the items makes it difficult to build a convincing case that the participants’ responses
reflect an accurate assessment of what they learned during their hiring process. However, I do
think that a reasonable case can be made for an argument that the items reflect some difference
between expectations and experiences. I cannot attribute that difference to the amount of
information gained during the hiring process, but I can accept that teachers do perceive a
difference between what they expected various aspects of teaching to be like and what those
various aspects were actually like.

24

Second, teachers’ expectations of their job experiences arise well before the hiring
process (Lortie, 1975). Teachers potentially develop expectations for their work lives from at
least two sources: experiences as students in K-12 schools and participation in some form of a
teacher preparation program. Nearly every teacher has accumulated better than 10,000 hours of
seat time as a student in elementary and secondary schools; Bruning, Schraw, and Ronning
(1999) cited research that indicated that 10,000 hours of deliberate practice is about the amount
of time it takes to acquire expertise in some domain. In their seat time, students are not
deliberately practicing to be teachers, but that quantity of experience certainly shapes their
expectations of the proper form and function of classroom life. In addition, teacher preparation
programs do provide teacher candidates with deliberate practice on being a teacher. However,
not all new teachers complete a teacher preparation program and not all teacher preparation
programs provide the same quality and quantity of deliberate practice. Nonetheless, most
teachers have some form of preparation and I predict that this preparation has some consequence
on teachers’ expectations for their work lives.
My hypothesis in this study is that the degree to which a new teacher’s expectations are
met will be associated with the new teacher’s levels of commitment to her school and to the
teaching profession. In the next section, I describe the methods that I adopted to evaluate my
hypothesis.
Methods
Participants
In this paper, I report on the 119 high school teachers who completed two questionnaires
during the 2007 – 2008 academic year. All of these teachers were in their first or second year of
full-time teaching, and they taught a variety of content areas. In total, study participants came

25

from 24 Catholic schools in eight dioceses across three states and 20 public high schools in nine
districts across two states.
I did not draw the sample of teachers randomly; instead, my dissertation director and I
recruited these particular teachers because they taught in schools that were located within
districts and dioceses that matched certain criteria for inclusion in the research study. First,
because I was interested in studying the role of the hiring process in affecting new teacher
commitment, I needed to sample from schools that were actively hiring instructional staff.
Second, because I was interested in studying the role of teacher networks in affecting new
teacher commitment (I report on this aspect of the study in Chapter 3), I needed to sample from
schools that had sizable teaching staffs. Third, because previous studies have found higher levels
of teacher mobility and attrition in schools with student populations that are high on minority and
low on socioeconomic measures, I wanted to build variation into the sample by way of student
characteristics for districts/dioceses. Consequently, I drew a sample of districts and dioceses that
was large enough to have hired new teachers for the 2006 – 2007 and 2007 – 2008 academic
years, that had schools with teaching staffs with more than two teachers per academic
department, and that varied on student measures of race and socioeconomic levels.
Procedures
Through our interactions with district and diocesan administrators, my dissertation
director and I obtained the names, positions, and contact information (school postal and email
addresses) for 309 teachers. I employed a five-contact procedure (Dillman, 2007) in the
recruitment of teachers to participate in the study; this procedure was founded upon ideas of
social exchange, as opposed to economic exchange, as a motive for action (p. 14). For the fall
administration of the survey, which occurred from November 2007 through January 2008, the

26

first contact consisted of a letter introducing the study; the second through fourth contacts
consisted of letters and emails asking the teachers to complete an online survey; the fifth (final)
1

contact asked the teachers to complete a paper survey that I included with the mailing. I inserted
a two-dollar bill in the second contact letter; in the cover letter of the fifth contact, I promised a
25-dollar gift card upon completion of the survey. For the spring administration of the survey in
May and June of 2008, I followed a similar procedure except that I did not include a two-dollar
bill with any of the contacts. I achieved a response rate of 64 percent for the fall survey
administration; for the spring, I retained 75 percent of the teachers who had completed a fall
questionnaire. In total, I obtained responses from 119 teachers on both the fall and spring
2

questionnaires, which corresponds to a response rate of 48 percent.

Measures
The fall questionnaire contained approximately 170 items, while the spring questionnaire
contained approximately 240 items. The questions asked about participants’ teaching positions,
frequency and nature of interactions with colleagues, professional development activities,
teaching practices, perceptions of their schools and departments, work and hiring experiences,
career decisions, preparations for teaching, and questions capturing participants’ gender, race,
and ethnicity (the specific survey questions are located in Appendix A and Appendix B). In
constructing the survey, I relied upon published research and previously established instruments
that dealt with topics such as schools as organizations (see, e.g., Bidwell, Frank, & Quiroz, 1997;
Bidwell & Yasumoto, 1999), the flow of resources through social networks (see, e.g., Frank,
1998; Frank, Zhao, & Borman, 2004), levels of commitment to schools and profession (National

27

Center for Educational Statistics, 2002), and teacher hiring (see, e.g., Liu, 2002; Liu & Johnson,
2006). As well, I attended to the design recommendations of Dillman (2007).
Dependent variables. In both the fall and spring administrations of the questionnaires, I
captured two different types of teacher commitment: commitment to the school and commitment
to the profession. In the analyses, the spring time-point serves as the outcome, while the fall
time-point serves as a control. For commitment to the school, I used Rasch-modeling software to
create a composite variable from three Likert-type survey items (Bond & Fox, 2007); Table 2.1
provides the items and their technical properties. The three items were used by Bryk and
Schneider (2002) to measure teacher commitment to school community (the authors actually
combined four items, but I removed one from the scale because it was not fitting the scale
properly). For commitment to profession, I used a single item that asked teachers whether they
would select teaching again as a career if they had the opportunity to return to college; this item
is found in the SASS and has been used by other researchers who have studied teacher
commitment (e.g., Ingersoll & Alsalam, 1997).

28

Table 2.1
Technical Properties of Commitment-to-School Measure for Fall and Spring
Fall Measure
Item

Spring Measure

Difficulty

Infit ZSTD

Difficulty

Infit ZSTD

I wouldn’t want to work in any other
school

1.91

1.10

1.06

.70

I would recommend this school to parents
seeking a place for their child

-.93

-.50

-1.19

-.30

I usually look forward to each working day
at this school

-.98

.83

.12

-.30

Person Reliability:

0.60

0.67

Note. Response options included strongly disagree, disagree, agree, strongly agree, and not sure.

Independent variables. The predictor of interest in this study is the measure of the degree
to which teachers’ expectations of various aspects of the school environment are met. For this
variable, I used Rasch-modeling software to create a composite variable from nine Likert-type
survey items to which teachers responded as a part of the fall questionnaire; Table 2.2 provides
the items and their technical properties. The fit statistic indicates that the item on curriculum is
less compatible with the model than expected (Bond & Fox, 2007). I decided to keep the item on
curriculum in the model because of its conceptual value to the composite and because removing
the item does not improve the overall person reliability statistic of the scale.

29

Table 2.2
Technical Properties of Met-Expectations Measure
Item

Difficulty

Infit ZSTD

During my hiring process (and the time before I actually started
working at this school), I acquired an accurate picture of…
The teachers at this school

.31

-.70

The students at this school

.04

.60

The principal’s leadership style

.27

.20

The curriculum that I would be responsible for teaching

.12

2.40

-.25

.60

.16

-1.70

-.30

-.50

.35

-2.00

-.75

.20

My teaching assignment
The school support system
The amount of personal control I would have in my classroom
The opportunities that I would have to shape the environment of the
school
The mission and philosophy of the school
Person Reliability:

0.78

Note. Response options included strongly disagree, disagree, agree, strongly agree, and not sure.

In addition to the predictor of interest, I evaluated the suitability of numerous teacherlevel and school-level control variables. Teacher-level controls included years of experience,
gender, race, teacher preparation, certification status, and teaching assignment; school-level
controls included sector, school size, percent of students who were minorities, and school
location. In addition, I controlled for teachers’ prior levels of commitment by including
commitment to the school (as measured in the fall), or commitment to the profession (as
measured in the fall) in the models.
30

Results
Descriptive Statistics
Teacher characteristics. Most of the participants taught language arts, math, science or
social studies (cumulative frequency of 80 percent); participated in a four- or five-year
bachelor’s degree granting teacher preparation program (frequency of 81 percent); majored or
minored in their content area (frequency of 82 percent); and held a standard or probationary
teaching certificate from their state (frequency of 71 percent). Some variation in the descriptive
statistics did exist both within and across sectors (e.g., 19 percent of the teachers from the
Catholic school sample did not have any form of teacher certification, compared with 0 percent
in the public school sample).
In 2007, males made up approximately 44 percent of the population of high school
teachers in the United States and racial minorities made up approximately 16 percent of the same
population (Planty et al., 2007). In the sample, 34 percent of the participants were male and 11
percent reported belonging to a racial minority. This could mean that males and racial minorities
were underrepresented in the study. For the case of male teachers, I examined the gender
distributions associated with the response rate, and there is evidence that males were less likely
to respond than females: 44 percent of the eligible teachers provided by the districts and dioceses
were male. Because of the data available, I was not able to conduct a parallel investigation of
response rates based on race.
School characteristics. The sample of 44 schools had substantial variation between the
schools in the public and Catholic sectors in terms of size and percent of students belonging to a
racial minority. Most of the Catholic schools enrolled between 600 and 1200 students (the mean
number of students was 624 with a standard deviation of 264), while half of the public schools

31

enrolled more than 1800 students (the mean number of students was 1811 with a standard
deviation of 753). In terms of the percentage of students who belonged to a racial minority,
almost all of the Catholic schools had student bodies that were less than 40 percent racial
minority (frequency of 92 percent), while over half of the public schools had student bodies that
were more than 40 percent minority (frequency of 60 percent). The mean percentage of students
belonging to a racial minority in the Catholic schools was 19 percent (with a standard deviation
of 22); the mean percentage of students belonging to a racial minority in the public schools was
48 percent (with a standard deviation of 23).
Predictor of interest: Met expectations. For the entire sample of teachers, most
participants agreed or strongly agreed to having acquired some degree of knowledge on various
characteristics of their school prior to the start of the school year (cumulative frequencies mostly
above 60 percent). Variation existed between the sectors and across school characteristics,
however. For instance, while nearly 90 percent of all participants reported having some
knowledge of the mission and philosophy of their school, only 61 percent of all participants
reported having some knowledge of the opportunities that they would have to shape the
environment of the school. In comparing the sectors, higher percentages of Catholic-school
teachers reported having acquired some knowledge of various characteristics of their school
relative to public school teachers. As an example, nearly 76 percent of Catholic-school teachers
obtained some knowledge about their future colleagues, as opposed to a little more than 56
percent of public school teachers. Overall, the ratio of teachers that acquired an accurate picture
of various aspects of their schools before they began teaching to those that did not acquire an
accurate picture was three to one for Catholic school teachers and two to one for public school
teachers.

32

Covariates: Fall commitment levels. In the fall, most teachers reported that they were
committed to the teaching profession (cumulative frequency of 86 percent) and none of the
teachers were certain that they would not become a teacher if they had to make the decision
again. Similarly, most teachers reported being immediately committed to the district/diocese,
school, and content area in which they were teaching (cumulative frequencies of 74 percent, 79
percent, and 81 percent, respectively). In addition, a majority of teachers would recommend their
schools to parents (cumulative frequency of 77 percent). However, when considering the
possibility of working in any school, only 45 percent of teachers reported that they would not
change schools. As well, commitment percentages dropped when teachers speculated about
continuing to work in their present district/diocese, school, and content area in five years. This is
relevant because a sizable majority of teachers reported that they intended to teach until
retirement or as long as they were able (cumulative frequency of 70 percent).
Outcomes: Spring commitment levels. Compared to the fall, fewer teachers reported that
they were committed to the teaching profession on the spring survey, although a substantial
majority still reported being committed (cumulative frequency of 76 percent). The decrease in
commitment was higher for those teachers in Catholic schools than those in public schools: the
percentage of Catholic school teachers who reported being committed to the profession fell 15
percent from the fall to the spring, while the percentage of public school teachers fell 4 percent
over the same time period. In both sectors, several teachers indicated that they would not choose
teaching as a career if they had to make the decision again (frequency of 4 percent) and several
said that they planned to leave the profession as soon as they could (frequency of 3 percent). In
spite of these percentage changes, a majority of teachers reported being committed to the
profession, as well as to their schools (cumulative frequency of 72 percent), content areas (79

33

percent), and district/dioceses (66 percent) in the short term. In addition, a majority of teachers
would recommend their schools to parents (cumulative frequency of 68 percent), but this
percentage is lower in the spring than it was in the fall. A majority of teachers felt loyal to their
school and looked forward to working in their schools each day (cumulative frequencies of 76
percent and 82 percent, respectively), but slightly less than half of the teachers reported that they
would not change schools if they could; interestingly, this last measure is four percent higher in
the spring than it was in the fall.
Inferential Statistics
Correlations. In this study, I attempted to learn whether the match between what teachers
expected their schools to be like and what they found their schools to be like affected their levels
of commitment to their schools and the teaching profession. Ordinal correlation analyses (e.g.,
the calculation of Kendall’s tau-a) allow for the direct assessment of this query.
On the question of met expectations, I found that those teachers who had a better fit in
terms of what they expected their schools to be like and what they found their schools to be like
reported higher levels of commitment to their schools and profession than did their colleagues
who had a worse fit. For the relationship between met expectations and level of commitment to
the school, Kendall’s tau-a equaled 0.29 (99.9 percent C.I. of 0.08, 0.50) for the fall measure and
0.31 (99.9 percent C.I. of 0.11, 0.51) for the spring measure. For the relationship between met
expectations and level of commitment to the profession, Kendall’s tau-a equaled 0.36 (99.9
percent C.I. of 0.18, 0.54) for the fall measure and 0.30 (99.9 percent C.I. of 0.14, 0.46) for the
spring measure.
Kendall’s tau-a is a proportion of congruent relations, which means it is more analogous
to R-squared than it is to Pearson’s correlation coefficient (rho) (Cliff, 1996, p. 40). A Kendall’s

34

tau-a measure of 0.30 is analogous to a Pearson correlation coefficient of 0.55; a Kendall’s tau-a
measure of 0.36 is analogous to a Pearson correlation coefficient of 0.60. In sum, I found a
statistically significant relationship between the extent to which teachers’ expectations of their
schools were met and their levels of commitment to their schools and the profession; in terms of
social science research, these relationships are moderately strong.
Multilevel linear regression models. The correlation analyses allowed me to directly
measure the relationship between the predictor variable of interest and the outcome variables. As
I indicated above, I found significant relationships between the predictor and outcomes. In this
section, I examine whether those relationships remain significant when various teacher-level and
school-level controls are considered. Because many of the participants taught in the same schools
as other participants, some of the teachers may have had their responses influenced in similar
ways by shared school-level characteristics. Single-level models assume independence of
observations and, consequently, are inappropriate and inefficient for nested data (Raudenbush &
Bryk, 2002). Hierarchical models account for the potential dependence of responses; as such, I
evaluated a set of multilevel models to assess the relationships among predictors and outcome.
To begin, I estimated a fully unconditional model for the school commitment outcome in
order to evaluate whether there was significant variation among the schools on this variable.
Equation 2.1 displays the combined form of the model (the hierarchical form of the model is
located in Table D1 in Appendix D):

(SCH_COMMIT_SPR)ij =

γ00 + u0j + rij.

(2.1)

35

The dependent variable, SCH_COMMIT_SPR, is the level of commitment to the school in the
spring for early-career teacher i in school j; γ00 is the average of the school means for the
outcome measure; u0j is the unique effect of each school; and rij is the unique effect of each
person. The results of the model that estimates Equation 2.1 are located in Table 2.3. Based on
2

the results, I conclude that there is a significant school effect (χ (43, n = 117) = 78.778, p <
0.001). The intraclass correlation coefficient for this model equals .236, which means that
approximately 24 percent of the variance in the model is due to differences between schools.

Table 2.3
Results of Fully Unconditional Model for Commitment-to-School Outcome
Fixed Effects

Coefficient

SE

t-ratio

Constant, γ00

61.956

3.223

19.224*** < 0.001

Random Effects

Variance
(SD)

df

χ

School effect, u0j

191.907
(13.853)

43

78.778*** < 0.001

Teacher effect, rij

620.947
(24.919)

2

p-value

p-value

Note. Level-1 n = 117; level-2 n = 44.
*** p < .001.

Next, I constructed a model that evaluated the effect of early-career teachers’ met
expectations on their school commitment level in the spring after controlling for their school
commitment level in the fall and other teacher-level and school-level covariates. In the model,

36

which I display in combined form in Equation 2.2 (below; the hierarchical form of the model is
located in Table D2 in Appendix D), SCH_COMMIT_FALL is the measure of school
commitment for early-career teachers in the fall, and MET_EXPECTATIONS is the measure of
the degree to which teachers’ expectations for their work lives matched their experiences in the
fall. That is, MET_EXPECTATIONS addresses the question, to what extent were teachers’
expectations for working in their schools met?
In addition to the prior measure of commitment and the predictor of interest, I also
evaluated and included a number of control variables. The teacher-level control variable that I
tested and retained in Equation 2.2 is the indicator for whether an early-career teacher received a
bachelor’s degree through a teacher preparation program, TEDEG_YN. I held the level-one
predictors fixed across schools (in the model development, I allowed MET_EXPECTATIONS
and SCH_COMMIT_FALL to vary randomly, but their variance components approached zero,
and so to simplify the model, I decided to keep them fixed).
I tested and decided to exclude a number of teacher-level variables because they did not
explain more than one percent of the variance in the outcome and because their inclusion in the
model resulted in the rejection of the null hypothesis that the level-1 error variance was
homogenous. The teacher-level variables that I excluded were indicators for whether the earlycareer teacher had obtained a regular state teaching certificate, taught math or science, was in
their first-year of full-time teaching, was female, and was Caucasian.
The school-level control variable that I evaluated and kept in the model is an indicator for
whether more than 40 percent of the students in the school belong to a racial minority group,
OVER40SCH_YN. I examined but did not retain school-level indicators of whether the school
was public, located in a central city, and had more than 1200 students enrolled in it. I excluded

37

those school-level controls because they did not explain more than one percent of the variance in
the outcome.

(SCH_COMMIT_SPR)ij =

γ00 + γ01(OVER40SCH_YN)j

(2.2)

+ γ10(SCH_COMMIT_FALL)ij
+ γ20(MET_EXPECTATIONS)ij
+ γ30(TEDEG_YN)ij + u0j + rij.

The dependent variable is the same as in Equation 2.1. The intercept, γ00, is the adjusted school
mean for schools with less than 40 percent of their students belonging to a racial minority and
controlling for the level-1 predictors. The parameter γ01 is the effect of schools with more than
40 percent of their students belonging to a racial minority; the parameters γ10 through γ30 are the
effects of their respective variables on the outcome. The components u0j and rij represent the
residual error among schools conditional on OVER40SCH_YN and the residual error among
teachers conditional on the teacher-level predictors, respectively. The descriptive statistics for
the variables in Equation 2.2 are in Table 2.4; the zero-order correlations for the variables in
Equation 2.2 are in Table 2.5.

38

Table 2.4
Descriptive Statistics for Model Estimating Effects of Met Expectations on School Commitment
Name

Mean

SD

Min.

Max.

Description

SCH_COMMIT_SPR

61.68

28.51

Scale measure of level of
.00 100.00 commitment to the school in the
spring

SCH_COMMIT_FALL

63.22

27.42

Scale measure of level of
.00 100.00 commitment to the school in the
fall

MET_EXPECTATIONS

56.29

18.17

Scale measure of the degree to
.00 100.00 which teachers’ job expectations
matched their job experiences

TEDEG_YN

.64

.48

.00

1 if obtained a bachelor’s degree
1.00 through a teacher preparation
program; 0 otherwise

OVER40SCH_YN

.30

.46

.00

1 if 40% or more of the students
1.00 at the school belong to a racial
minority; 0 otherwise

Note. Level-1 n = 117; level-2 n = 44.

39

Table 2.5
Correlation Matrix for Model Estimating Effects of Met Expectations on School Commitment
Variables

1

2

3

4

5

1. SCH_COMMIT_SPR
2. SCH_COMMIT_FALL

.798***

3. MET_EXPECTATIONS

.431***

.402***

4. TEDEG_YN

.203*

.076

-.006

5. OVER40SCH_YN

-.088

-.075

-.065

.047

Note. Level-1 n = 117; level-2 n = 44. Values for comparisons between two variables in which
both variables are measured on interval scales are Pearson correlation coefficients; values for
comparisons between variables in which at least one of the variables is measured on a binary
scale are Spearman correlation coefficients.
* p <.05; *** p < .001.
The results of the model that estimates Equation 2.2 are located in Table 2.6. Each of the
teacher-level fixed effects explains a significant amount of the variation in the outcome. A
teacher’s fall commitment level (γ10 = .762, t(70) = 12.206, p < .001), the degree to which a
teacher’s expectations are met (γ20 = .205, t(70) = 2.188, p = .032) and whether a teacher
received a bachelor’s degree through a teacher preparation program (γ30 = 7.097, t(70) = 2.186,
p = .032) are each significantly associated with the teacher’s spring commitment level when
evaluated at a p-value of .05. The results indicate that the school effect is no longer significant
2

once the school-level and teacher-level predictors are included in the model (χ (42, n = 117) =
46.769, p = .283).
Because the outcome measure and the continuous predictors are on interval scales that
stretch from 0 to 100, the coefficients can be interpreted in terms of percents of the scale range.
40

Specifically, the unstandardized coefficient for SCH_COMMIT_FALL indicates that, ceteris
paribus, a one percent of the scale range increase in the fall level of school commitment is
expected to result in a .762 (less than one) percent of the scale range increase in the spring level
of school commitment; a one percent of the scale range increase in the degree of met
expectations is expected to result in a .204 (less than one) percent of the scale range increase in
the outcome when all of the other variables are held constant. As well, ceteris paribus, an earlycareer teacher who received a bachelor’s degree as part of a teacher preparation program is
expected to have a school commitment level in the spring that is 7.097 percent of the scale range
greater than an early-career teacher who did not receive a bachelor’s degree through a teacher
preparation program.
The expected effects in terms of standard deviations, as well as the index of the
proportion of variance explained, provide ways to compare the relative importance of each of the
predictors. In terms of standard deviations, a one standard deviation increase in the level of
commitment in the fall is expected to yield an increase in the level of commitment to the school
in the spring of .733 standard deviations when all other variables are held constant. A one
standard deviation increase in the degree of met expectations is expected to result in an increase
in the level of commitment to the school in the spring of .131 standard deviations, ceteris
paribus. Last, an early-career teacher who received a bachelor’s degree through a teacher
preparation program is expected to have a level of commitment to the school in the spring that is
.249 standard deviations greater than an early-career teacher who did not receive a bachelor’s
degree through a teacher preparation program when all other variables remain constant. For
proportion of variance explained, the results indicate that SCH_COMMIT_FALL explains

41

approximately 49 percent of the variance in the outcome, MET_EXPECTATIONS explains 3
percent, and TEDEG_YN explains 3 percent.

Table 2.6
Results of Model Estimating Effects of Met Expectations on School Commitment
Variance
Explained

Fixed Effect

Coefficient

SE

t-ratio

p-value

Constant, γ00

-1.946

5.751

-.338

.737

OVER40SCH_YN, γ01

-2.150

3.440

-.625

.535

SCH_COMMIT_FALL, γ10

.762

.062

12.206*** < .001

.486

MET_EXPECTATIONS, γ20

.205

.094

2.188*

.032

.031

TEDEG_YN, γ30

7.097

3.246

2.186*

.032

.031

Random Effect

Variance
(SD)

df

χ

p-value

School effect, u0j

.559
(.748)

42

46.769

.283

Teacher effect, rij

280.441
(16.746)

2

.054

Note. Level-1 n = 117; level-2 n = 44.
*p < .05; ***p < .001.

Multilevel logistic regression models. In the previous section, I assessed the suitability of
various predictors for the commitment to school outcome by using linear regression. For the
commitment to profession outcome, I cannot use standard linear regression because the outcome
measure is not continuous. One option would be to use ordinal logistic regression to model the

42

effects of the various predictors across the five different ordered categories of the outcome
response variable. However, when I explored the response patterns in detail, I discovered a
number of violations of the assumptions for ordinal logistic regression models. Specifically, I
found inadequate cell counts and an inadequate dispersion of responses across the five categories
of the variable. Further, the sample size was inadequately small to reasonably attempt to fit an
ordinal regression model to the data (Long, 1997).
As a consequence of the inability to fit an ordinal regression model to the data, I decided
to collapse the five response categories of the outcome variable into two categories and model
those responses using binary logistic regression. The survey question that I used for the
professional commitment measure asked teachers, “If you could go back to your college days
and start over again, would you become a teacher or not?”. Teachers could respond in one of five
ways: “certainly would become a teacher”; “probably would become a teacher”; “chances about
even for and against”; “probably would not become a teacher”; or “certainly would not become a
teacher”. I collapsed the two affirmative response options together and dubbed those responses
being committed to the profession; I combined the indecisive response option with the two
negative response options and dubbed those responses being uncommitted to the profession. In
Figure 2.1, I display the frequencies for the five responses to the professional commitment
measure when asked of teachers in the fall; in Figure 2.2, I display the frequencies for the five
responses to the professional commitment measure when asked of teachers in the spring.

43

60

41

10

Certainly Probably
would
would

7

0
Chances Probably Certainly
even
would not would not

Figure 2.1. Response frequencies of early-career teachers when asked in the fall: “If you could
go back to your college days and start over again, would you become a teacher or not?”
59

31
19

4
Certainly Probably
would
would

5

Chances Probably Certainly
even
would not would not

Figure 2.2. Response frequencies of early-career teachers when asked in the spring: “If you
could go back to your college days and start over again, would you become a teacher or not?”

44

Both binary logistic and ordinal regression models use the same link function in the
evaluation of the data: the models calculate the log of the odds of one of the responses occurring
in the dependent variable given the values of the set of predictors included in the model. The
difference between binary logistic and ordinal regression models lies in the event whose
probability is being calculated; for binary logistic, the event is one response as opposed to the
other; for ordinal, the event is one response or one set of responses as opposed to all of the other
responses that are of a higher order than the single response or the highest ordered response of
the set.
As with the commitment to the school outcome, I chose to use a multi-level model to
evaluate the commitment to the profession outcome because the respondents were nested within
schools. However, unlike the continuous structure of the commitment-to-school outcome, binary
measures are not suited for standard hierarchical linear models for a number of reasons: the
predicted values of the outcome could be greater than one or less than zero, the level-one error
term cannot have a random distribution, and the variance of the level-one error term cannot be
homogenous (Raudenbush & Bryk, 2002). Consequently, I evaluated a set of hierarchical
generalized linear models (HGLMs) in order to estimate the effects of the predictor variables on
the commitment to the profession outcome variable. As well, I chose to use unit-specific models
(as opposed to population-average models) when I investigated the effects of hiring experiences
on the commitment to profession outcome because I wanted to model group to group differences
and examine the amount of variation within and between groups (Roux, 2004).

45

To begin, I estimated a fully unconditional Bernoulli sampling model with a logit link
function. Equation 2.3 displays the combined form of the model (the hierarchical form of the
model is located in Table D3 in Appendix D):

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) %
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij

!

= γ00 + u0j.

(2.3)

The term on the left side of the equation is the log of the odds of a teacher reporting being
committed to the teaching profession in the spring of 2008. The outcome measure
PRO_COMMIT_SPR_YN is the dichotomous variable that takes the value of one if the earlycareer teacher indicated that if they went back to college they “certainly” or “probably” would
become a teacher; the variable takes the value of zero if the early-career teacher responded that
the chances were “about even for or against” or “probably” or “certainly” would not become a
teacher. The intercept, γ00, is the average log-odds of commitment across schools; the error term,
u0j, represents the variance between schools in the average log-odds of commitment.
The results of the model estimating Equation 2.3 are located in Table 2.7. The
unstandardized coefficient indicates that the expected log-odds of professional commitment in
the spring at an average school is 1.182, which corresponds to an odds ratio of 3.261, which
3

corresponds to a probability of .765. This means that in the average school, approximately 77
percent of the early-career teachers reported being committed to the profession in the spring;
calculating a confidence interval around the estimate yields the finding that 95 percent of the
schools have between 47 and 92 percent of their early-career teachers being commitment to the
profession in the spring.

46

Table 2.7
Results of Fully Unconditional Model for Commitment-to-Profession Outcome
Fixed Effects

Coefficient (SE)

Odds Ratio

t-ratio

p-value

Constant, γ00

1.182 (.230)

3.261

5.131***

< .001

Random Effects

Variance Component (SD)

df

χ

p-value

School effect, u0j

.450 (.203)

43

41.525

> .500

2

Note. Level-1 n = 118; level-2 n = 44.
*** p < .001
Next, I built a model that evaluated the effects of met expectations on early-career
teachers’ commitment to the profession in the spring. The predictor of interest in this model is
MET_EXPECT_B, which is a dichotomous variable that indicates whether the score for person i
on the Rasch-calculated composite variable MET_EXPECTATIONS was above the median. I
used a dummy coding in this model for the met expectations measure, as opposed to the Raschcalculated composite variable in the previous model, because a logit step test indicated that the
continuous predictor, MET_EXPECTATIONS, was not linearly related to the log of the odds of
the dependent variable. A linear relation between the predictor and the log of the odds of the
outcome is one of the requirements for logistic regression (Garson, 2011).
In addition to the predictor of interest, I included an indicator of whether an early-career
teacher was committed to the profession in the fall (PRO_COMMIT_FALL_YN). As well, I
evaluated a number of control variables and kept them in the model if their p-values were less
than 0.25 when they were entered into the model with just the fall indicator of professional
commitment (see, e.g., Hosmer & Lemeshow, 2000; cited by Norušis, 2008). At the teacher-

47

level, I retained the indicator for whether an early-career teacher received a bachelor’s degree
through a teacher preparation program, TEDEG_YN, as well as an indicator of whether the
early-career teacher was in the first or second year of full-time teaching, FIRSTYEAR_YN. At
the school level, I retained an indicator of whether the school had more than 1200 students in it,
OVER1200SCH_YN. I did not keep indicators for an early-career teacher’s certification status,
gender, race, school sector, school location, or proportion of students in the school who belonged
to a racial minority. I present the combined form of the model in Equation 2.4 (the hierarchical
form of the model is located in Table D4 in Appendix D):

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) %
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij

=

γ00 + γ01(OVER1200SCH_YN)ij

(2.4)

+ γ10(PRO_COMMIT_FALL_YN)ij
+ γ20(MET_EXPECT_B)ij
+ γ30(TEDEG_YN)ij

!

+ γ40(FIRSTYEAR_YN)ij + u0j.
The outcome variable, PRO_COMMIT_SPR_YN, retains its definition from Equation
2.3. The intercept, γ00, equals the average log-odds of professional commitment in the spring
across schools that had fewer than 1200 students in them and for early-career teachers who were
not committed to the profession in the fall, did not earn their bachelor’s degree through a teacher
preparation program, and were in their second year of full-time teaching. The parameters γ10
through γ40 are the effects of their respective variables on the log-odds of the outcome variable;
the residual, u0j, is the error among schools conditional on OVER1200SCH_YN. The descriptive
statistics for Equation 2.4 are displayed in Table 2.8; the zero-order correlations for Equation 2.4
are displayed in Table 2.9.

48

Table 2.8
Descriptive Statistics for Model Estimating Effects of Met Expectations on Professional
Commitment
Name
PRO_COMMIT_SPR_YN

PRO_COMMIT_FALL_YN

MET_EXPECT_B

Mean
.76

.86

.77

SD

Min.

Max.

Description

.00

.42

1 if certainly or probably would
1.00 become a teacher again – fall
measure; 0 otherwise

.00

.35

1 if certainly or probably would
1.00 become a teacher again –
spring measure; 0 otherwise

.00

.43

1 if MET_EXPECTATIONS
1.00 score was above the median; 0
otherwise

TEDEG_YN

.64

.48

.00

1 if obtained a bachelor’s
degree through a teacher
1.00
preparation program; 0
otherwise

FIRSTYEAR_YN

.50

.50

.00

1.00

1 if in first year of full-time
teaching; 0 otherwise

OVER1200SCH_YN

.34

.48

.00

1.00

1 if school enrolls 1200 or more
students; 0 otherwise

Note. Level-1 n = 118; level-2 n = 44.

49

Table 2.9
Correlation Matrix for Model Estimating Effects of Met Expectations on Professional
Commitment
Variable

1

2

3

4

5

6

1. PRO_COMMIT_SPR_YN
2. PRO_COMMIT_FALL_YN

.565***

3. MET_EXPECT_B

.313**

.466***

4. TEDEG_YN

.157^

.040

-.077

5. FIRSTYEAR_YN

-.080

.024

.020

-.053

6. OVER1200SCH_YN

.133

-.040

-.174

.098

-.229*

Note. Level-1 n = 118; level-2 n = 44. Values for comparisons between two variables in which
both variables are measured on interval scales are Pearson correlation coefficients; values for
comparisons between variables in which at least one of the variables is measured on a binary
scale are Spearman correlation coefficients.
^ p < .10; * p < .05; ** p < .01; *** p < .001.
The results of the model that estimates Equation 2.4 are displayed in Table 2.10. The fall
indicator of whether an early-career teacher is committed to the teaching profession is
statistically significantly related to the outcome (γ10 = 3.921, t(70) = 4.035, p < .001). The
coefficient indicates that we would expect an early-career teacher who was committed to the
teaching profession in the fall to have a log-odds of being committed to the teaching profession
in the spring that was 3.921 units greater than that of an early-career teacher who was not
committed to the profession in the fall, ceteris paribus. In terms of predicted probabilities, we
would expect 91 percent of the early-career teachers’ who were committed to the profession in
the fall to be committed to the profession in the spring, ceteris paribus. As well, we would
expect 17 percent of the early-career teachers who were not committed to the profession in the

50

fall to be committed to the profession in the spring, ceteris paribus. Last, I cannot reject the null
hypothesis that the coefficient on the indicator of met expectations is equal to zero.

Table 2.10
Results of Model Estimating Effects of Met Expectations on Professional Commitment

Fixed Effects

Coefficient (SE)

Odds
Ratio

Constant, γ00

-3.137 (1.143)

.043

-2.744**

.009

OVER1200SCH_YN, γ01

1.310 (.912)

3.705

1.436

.158

PRO_COMMIT_FALL_YN, γ10

3.921 (.972)

50.437

4.035***

< .001

MET_EXPECT_B, γ20

1.079 (.834)

2.941

1.293

.200

TEDEG_YN, γ30

.862 (.619)

2.369

1.393

.168

FIRSTYEAR_YN, γ40

-.770 (.670)

.463

-1.150

.254

Random Effects

Variance Component (SD) df

χ

p-value

School effect, u0j

1.544 (1.242)

47.300

.265

42

t-ratio

p-value

2

Note. Level-1 n = 116; level-2 n = 43.
** p < .01; *** p < .001.
Discussion
Contributions of the Study
This study evaluated the effects of met expectations on early-career teachers’ levels of
commitment to their schools and the profession. I found that the concept of met expectations was
statistically significantly related to the concept of school commitment but not to the concept of

51

commitment to the profession. Specifically, I found that an increase in the degree of the fit
between early-career teachers’ expectations and experiences resulted in an expected increase in
the level of commitment to the school. As well, I found that teachers who obtained their
bachelor’s degree through a teacher preparation program had a statistically significantly higher
level of commitment to their schools in the spring than did teachers who did not receive their
bachelor’s degree through a teacher preparation program. In this section, I discuss these findings
in terms of the literature that I reviewed at the outset of the paper.
The finding that met expectations affect early-career teachers’ levels of commitment to
their schools is consistent with research on the concept in the field of organizational psychology
(Wanous, Poland, Premack, & Davis, 1992). One of the contributions of this study is applying
the concept of met expectations to the field of education. In doing so, I build upon foundational
work by Liu and Johnson (2006). However, this study goes beyond prior work on met
expectations by arguing that teachers’ development of expectations for their work lives begins
before the hiring process.
This study’s finding that teachers who earned a bachelor’s degree through a teacher
preparation program had higher levels of commitment to their schools supports the argument that
teachers’ expectations for their work lives develop before the hiring process begins. It seems
reasonable that students in teacher preparation programs spend more time discussing many of the
aspects of life in schools relative to individuals who do not participate in a teacher preparation
program. As well, it seems reasonable that students in teacher preparation programs would have
more opportunities to spend more time in classrooms as observers than individuals who do not
participate in teacher preparations programs. It is my hypothesis that the additional time which
students in teacher preparation programs spend discussing and participating in life in schools

52

leads to the development of more realistic expectations for their jobs relative to individuals who
do not participate in teacher preparation programs. This hypothesis could lay the foundation for a
study on the effects of teacher preparation.
The retention of the null hypothesis that the effect of met expectations on early-career
teachers’ levels of commitment to the profession is zero is not necessarily a counterpoint to the
application of met expectations to the context of schools. One possible explanation for the null
finding could be a misalignment between the measure of met expectations and the concept of
commitment to the profession. Each of the items that I used to develop the met expectations
scale referred specifically to some within-school characteristic. None of the items in the scale
asked about the degree of match between teachers’ expectations of the profession and their actual
experiences in the profession. Future research could be done to develop a set of items that
measured teachers’ met expectations of the profession.
Another possible explanation for the null finding on the relationship between met
expectations and professional commitment is the limited power in the statistical model to detect
an effect. Long (1997) suggested a minimum of 100 cases as a rule of thumb plus 10 cases for
each predictor that is included in the model. This study meets that minimum threshold. However,
if the size of the effect is small, then larger samples are required to detect the effect. As such, I
recommend that future efforts to study the effects of met expectations on professional
commitment adopt a design with a larger sample than the one this study used.
One of the central contributions of this study is controlling for the prior measure of the
outcome variable in the analyses. By controlling for the prior level of the outcome, I can assign
directionality to the relationship between the predictors and the outcome. That is, because I
controlled for teachers’ prior levels of commitment, and because I measured teachers’ levels of

53

met expectations in the fall, I can assert that met expectations are not only related to spring levels
of school commitment, but also that met expectations affect spring levels of school commitment.
Similarly, I can assert that obtaining a bachelor’s degree through a teacher preparation program
affects spring levels of school commitment.
The ordering of the relationship would have implications for policymakers if the findings
were generalizable. If I found the relationship to be merely associational, I would not
recommend efforts to improve the fit between new teachers’ expectations and experiences.
However, because the relationship is directional, I would recommend efforts to make teachers’
expectations for their jobs more realistic if the sample was representative (which it is not). I
anticipate that locations in which these efforts are made will find that their teachers possess
higher levels of commitment to their schools (relative to where these efforts are not made). But
this needs to be studied with a representative sample before policy applications are warranted. In
the next section, I specifically address the limitations of this study.
Limitations of the Study
Even though this study makes important contributions to research on met expectations
and early-career teachers’ commitment levels, it has important limitations. First, the reliability of
the outcome measure is low in terms of standards for social science research. Next, the design
does not control for the natural development of the participants (Campbell and Stanley, 1966). It
could be that this natural development is what is explaining the variance in the early-career
teachers’ levels of commitment to their schools, rather than the degree of met expectations. A
better design would account for the potential confounding variable of natural development by
randomly selecting the sample or capturing more than two time-points of data (and ideally
capturing data over multiple years if the resources of the study allowed).

54

Finally, I am also unable rule out the possibility that the findings in this paper are due to
the interaction between the predictors that I evaluated in this study and the non-random sample
that I drew for the study. In other words, I cannot rule out the possibility that the findings in this
study apply only to the sample of teachers that participated in this study. If I wanted to apply the
findings in this study to teachers that were not a part of the sample, I would have needed to select
the sample randomly from the population of interest.
Despite these limitations, this study presents enough evidence to warrant further
investigation of the concept of met expectations and the role of teacher preparation programs in
affecting early-career teachers’ levels of commitment to their schools. I am confident that future
studies on these topics that employed representative samples in their designs would lead to the
production of guidance for policymakers.

55

CHAPTER THREE: THE INFLUENCE OF COLLEGIAL NETWORKS ON BEGINNING
TEACHERS’ COMMITMENT LEVELS IN PUBLIC AND CATHOLIC SCHOOLS
Introduction
Catholic and public schools experience high levels of teacher turnover each year. Firstyear teachers leave at the rates of 16 percent per year in Catholic schools and 11 percent per year
in public schools; as well, first-year teachers change schools at the rates of 12 percent per year
and 16 percent per year, respectively (Smith & Ingersoll, 2004). In reviewing the literature on the
potential impact of teacher turnover on schools as organizations, Ingersoll (2001) concluded that
high turnover could affect the commitment and cohesion of the members of the organization,
thereby affecting school performance. Reducing school performance is problematic itself, but
particularly so when it is realized which type of schools and students are hurt most often.
Hanushek, Kain, and Rivkin (2004a), in their study on teacher retention in Texas, found that
disadvantaged and low-achieving students were more likely to be in schools with high levels of
teacher turnover than were their more advantaged and high-achieving peers. In short, teacher
attrition is problematic for the school as an organization and is particularly harmful to
disadvantaged, low-achieving students.
Many research studies point toward teacher working conditions as a likely contributor to
teacher attrition. Hanushek, Kain, and Rivkin (2004a) considered working conditions in terms of
student characteristics and found evidence that those characteristics affected teacher retention
decisions. In addition, Hanushek, Kain, and Rivkin (2004b) found teacher salary to be associated
with retention decisions; other research studies support this association (Brewer, 1996; Kirby,
Berends, & Naftel, 1999). Johnson and Birkeland (2003) examined working conditions in terms
of the professional cultures of schools and found that collegial relationships affected teacher

56

retention decisions. Similarly, Smith and Ingersoll (2004) concluded that in-field mentors,
teacher networks, and structured time for professional collaboration were significant factors in
reducing teacher attrition. As well, Ebmeier (2003) suggested that improving teacher retention in
schools could be achieved by facilitating “functional and supportive relationships with other
teachers” (p. 138).
These findings suggest that districts and dioceses have the potential to alleviate some of
the struggles of early career teachers, thereby easing new teachers’ entry into school
communities and the teaching profession. However, schools are complex social organizations,
and much of the research on this topic misses important features of life in schools. For instance,
teacher responses to district and diocesan policies are mediated through local processes found in
schools, and these local processes can affect new teachers’ beliefs and actions (Bryk &
Schneider, 2002; Coburn, 2003). Further, because of the multitude of relationships that exist in
schools, teachers both receive supports and experience pressures from their colleagues and
employers in their daily work lives (Bidwell, 2000; Frank, Zhao, & Borman, 2004; Kennedy,
2005).
This chapter contributes to the existing literature concerned with the role of induction
experiences on early-career teachers’ commitment levels. In short, this study investigates two
questions. First, “What are the effects of mentoring on early-career teachers’ levels of
commitment to their schools and the profession?” Second, “What are the effects of within-school
colleagues on early-career teachers’ levels of commitment to their schools and the profession?”
In the next section, I describe how social networks could affect teachers’ commitment levels.

57

Framework
The framework for this paper builds chiefly around theoretical ideas of social capital and
social networks. By way of social capital, the frame finds influence in the writings and works of
Bourdieu (1986), Coleman (1988), Lin (1999), and Portes (1998), among others. However, it is
prudent to observe that the idea of capital originates in the writings of Karl Marx. Lin noted that
Marx described capital as being composed of two elements, surplus and investment. According
to Lin, Marx used the elements in reference to class differences: the dominant class, or
capitalists, acquired surplus value from the sale of commodities produced by the laborers; this
represents the surplus element. In addition, the capitalists invested in the further production of
commodities, expecting a future return on their investment; this represents the investment
element. Lin refers to Marx’s theory as the “classical theory of capital” and characterizes
theories and discussions of capital (such as human, cultural, and social) that move away from the
context of class differences as “neo-capitalist theories” (p. 29).
Social Capital
Social capital, as I make use of it in this study, lies in the company of the neo-capitalist
theorists. In the context of schools, Marx’s distinction between capitalists and laborers does not
find relevant standing. That is not to say that employees and employers in school settings operate
on equal footing, but the nature of their positional differences are not well described by Marx’s
classical theory of capital: employers in school settings do not receive surplus from the sale of
their employees’ labors.
However, in spite of the contextual differences between school settings and the settings
of nineteenth-century workers, Marx’s basic premise that capital consists of the elements of
surplus and investment still holds in my usage of social capital in this study. That is, I use social

58

capital to mean the resources available to an individual because of his or her position in a
network of relationships with others. Marx’s notion of surplus resides in the ability of an
individual to access resources, while Marx’s notion of investment lies in an individual’s
development and maintenance of the social network.
An extrapolation of Marx’s investment component of capital suggests that social capital,
by definition, relies on the existence of relationships among individuals. That is, social capital is
a property of social networks; without networks, there could not be social capital. The obvious
question that arises is whether social capital is a necessary property of social networks. In other
words, if social networks exist, must also social capital exist? A turn to Lin (1999) helps to
resolve this quandary. Lin conceptualizes a definition of social capital as “resources embedded in
a social structure which are accessed and/or mobilized in purposive actions” (p. 35). This
definition moves the theoretical foundation of social capital to consider both a structural
component (the embedded resources) and a relational component (the accessibility and the
activation of relationships). Such a move is fundamentally important as it addresses the
definitional quandary of where social capital lies (or with whom it lies). Lin’s definition argues
that social capital does not simply lie with an individual, but rather Lin’s definition asserts that
social capital resides both in individuals and in their relationships. That is, social capital consists
both of the resources that individuals possess (resources embedded in the network) and of the
potential for activating relationships in order to access those embedded resources (network
characteristics affect the potential of activation).
In operationalizing the elements of social capital, Lin (1999) attends to three processes
for study: (a) an individual’s investment in social capital, (b) the extent of an individual’s access
to and activation of social capital, and (c) the returns to an individual from his or her activation

59

of resources in the network. For Lin, returns take shape in two ways: acquiring new resources
and maintaining existing relationships; note that these returns echo Marx’s notions of surplus and
investment, respectively.
Portes (1998) notes that sociologists have written on the consequences of social capital in
generally positive ways, but he observes that social capital has the potential to produce negative
consequences as well. Portes writes that internalized norms for behavior allow for the flow of
resources in social situations, but they also work to narrow the spectrum of what the network
deems acceptable behavior. Such a narrowing of possible behaviors could lead to group
conformity and the restriction of individual freedoms, which could diminish the capacity for
growth within the network.
In the context of new teachers, the potential for both positive and negative consequences
of social relations seems quite plausible. For instance, new teachers could potentially benefit by
learning classroom management strategies from their more senior colleagues. However, new
teachers could be discouraged if their colleagues disapprove of their instructional tactics.
Consequently, I adopt a conceptualization of social capital that is concerned with both the
supports and pressures that teachers receive from their social networks. In short, supports are
those resources gained by new teachers that benefit them in their work; these include curricular
materials, information regarding school and curriculum issues, as well as advice and support
regarding classroom management and interacting with parents, among other examples. Pressures
are those messages received by new teachers that place expectations on their behavior and work;
these include the implementation of curricular materials in particular ways, managing students in
particular ways, as well as being assigned to committees and extracurricular appointments,
among other examples.

60

Social Networks
Social networks consist of relationships, or ties, among members of a social system; for
instance, the relationships between early-career teachers and their professional colleagues in the
setting of a school. However, all ties are not created equal. That is, different relationships have
different qualities about them, and it is necessary to attend to these qualitative differences. For
instance, an early-career teacher might receive consistent, sustained support (or pressure) from
one relationship, but another relationship might only provide infrequent support (or pressure) of
a specific type. As an example, an early-career teacher might receive frequent support regarding
curricular materials from her department chair, but she might receive infrequent support
regarding the school’s electronic grading program from the technology coordinator; the former is
an example of a strong tie, while the latter is an example of a weak tie. It is necessary to make
distinctions regarding the qualitative nature of relationships because the strength of ties has
consequences for the flow of resources between pairs of individuals, as well as the functioning of
the larger social structure (Granovetter, 1973, 1983).
In his development of a theoretical basis for using social network analyses to connect
micro and macro sociological events, Granovetter (1973) defined relational strength, or tie
strength, as “a combination of the amount of time, the emotional intensity, the intimacy (mutual
confiding), and the reciprocal services which characterize the tie” (p. 1361). In our previous
example, it is apparent that the early-career teacher’s relationship with her department chair
entails more time and is of greater intensity than is the early-career teacher’s relationship with
the technology coordinator. Nonetheless, both kinds of ties provide tangible resources or
expertise to the early-career teacher, and so both weak and strong ties are important to the earlycareer teacher. Granovetter (1983) describes the importance of weak and strong ties in this way:

61

“weak ties provide people with access to information and resources beyond those available in
their own social circle; but strong ties have greater motivation to be of assistance and are
typically more easily available” (p. 209). Considering the example of the early-career teacher,
the technology coordinator likely does not lie within the early-career teacher’s social circle and
so they interact infrequently. However, it seems highly plausible that the early-career teacher
would interact quite frequently with her department chair, and so they would likely assume
positions in the same social circle.
If we consider the expanse of a social environment (e.g., a school faculty), then existing
within that expansive social structure are numerous substructures, also referred to in the research
literature as “cliques” (Coleman, 1961), “cohesive subgroups” (Frank, 1995, 1998), “social
circles” (Granovetter, 1983), and more simply, “groups” (Burt, 2001). Each of these various
designations of environmental substructures contain slightly different technical (or non-technical,
in many cases) characterizations of substructures, from Coleman’s aesthetic sociogram to
Frank’s empirical procedure using odds-ratios and multidimensional scaling techniques.
Nonetheless, each of these substructure designations works toward the same basic idea: namely,
substructures are composed of individuals who generally interact with each other more than with
other individuals outside of the substructure. In this study, I attempt to measure the influence of
the subgroup to which an early-career teacher belongs by asking the early-career teacher to name
the members of their subgroup; I then administered surveys to the members of the subgroup.
The structural features of schools formally organize teachers into subunits. For instance,
in elementary schools, teachers often interact with others who teach the same grade level; in
middle and high schools, teachers often interact with others who teach the same content area;
school structures foster these interactions through daily common planning time or periodic

62

professional development activities. As well, school geographies often locate similar grade levels
and content areas in the same region of the school building, thereby encouraging the interactions
of certain individuals through physical proximity.
However, these formal structural subunits are not, by definition, equivalent to subgroups.
Rather, subgroups form in a more organic nature with ties representing the establishment of
personal relationships, as opposed to structural designations. For instance, Frank (1996) found
that subgroups in one school formed around racial lines and gender, as well as by content area.
That subgroups developed according to racial and gender characteristics is less the result of
school structures and more the result of personal preferences. Nonetheless, it makes intuitive
sense that those school structures that promote the interactions among certain individuals (e.g.,
common planning time or space for members of particular content-area departments) would
affect the development of subgroups by increasing the frequency with which those individuals
interact. In short, a combination of structural features and personal preferences lead to the
generation of subgroups in schools.
Hypotheses
I expect that the benefits to early-career teachers of interacting with mentors and
colleagues will outweigh the pressures that are placed upon the early-career teachers through
their interactions. And I expect those benefits to be realized in the school commitment levels of
the early-career teachers. Specifically, if everything else is equal, I expect that the early-career
teachers who have mentors will have higher levels of commitment to their schools than will the
early-career teachers who do not have mentors. I expect to obtain similar effects for colleagues.
As well, if everything else is equal, I expect that communicating more frequently with mentors
and colleagues will increase early-career teachers’ levels of commitment to their schools relative

63

to communicating less frequently. Because I have limited my definition of networks to the
within-school colleagues, I do not anticipate finding an effect for any of the network variables on
the commitment to the profession outcome. In the next section, I describe the methods that I
employed in this study to evaluate these hypotheses.
Methods
Participants
In this paper, I report on the responses from 119 early-career high school teachers and
248 of their mentors and colleagues. I use the term early career to indicate a teacher in the first
or second year of full-time teaching. Each of the early-career participants completed two
surveys; they submitted one between November 2007, and January 2008, and they submitted the
other between May and June of 2008. Each of the mentor and colleague participants completed
one survey; they submitted it between February and April of 2008.
Prior to recruiting individuals to participate in the study, my dissertation advisor and I
identified and recruited a number of public school districts and Catholic dioceses for the study.
We limited the search for public school districts to two states in the Midwest. This limitation was
pragmatic: I needed to minimize the travel costs that would be incurred when conducting the
face-to-face interview portion of the study (the interviews are not within the scope of this paper
and so I do not report on them here). Within the two states, I examined demographic information
on districts in medium to large cities and their suburbs.
We contacted the superintendents in a dozen districts that had teaching staffs with more
than two teachers per academic department and that had variation in student measures of race
and socioeconomic status. The teaching staffs needed to have more than two teachers per
department in order to have a possibility of finding new teachers who could nominate teaching

64

colleagues with whom they interacted on professional matters. We sought variation in the
characteristics of students because prior research indicates that student characteristics matter to
teachers’ career decisions (Hanushek, Kain, & Rivkin, 2004a).
After we received the permission of district administrators to conduct the study in their
public schools, I contacted the superintendents of the Catholic dioceses that represented either
the geographic areas in which the public school districts were located or demographically similar
areas across three states; I recruited some dioceses in demographically similar areas and added
one state to the sample in order to increase the number of potential participants from Catholic
high schools. In total, we received surveys from participants in 24 Catholic high schools in eight
dioceses across three states and 20 public high schools in nine districts across two states.
Procedures
We requested the names and contact information for the early-career teachers in the
participating districts and dioceses; administrators provided us with information for 309
individuals. I used a multiple-contact procedure (Dillman, 2007) to recruit participants to the
study. For each of the three survey administrations, the multiple contacts spanned approximately
eight weeks and consisted of a series of letters, emails, and postcards. The first contact described
the study; the second through fourth contacts provided more information about the study and
directed participants to the web address for the electronic survey; the fifth and final contact
4

consisted of a letter and a paper survey.

Dillman (2007) recommends that researchers attempt to frame interactions with potential
study participants in terms of social exchange, as opposed to economic exchange. According to
Dillman, a small bill sent at the outset of a study was less an economic exchange and more an
effort to establish a trusting relationship with potential participants. To that end, I included a
65

two-dollar bill in the second letter that I sent to the early-career teachers in November of 2007; I
also included a two-dollar bill in the second letter that I sent to the mentors and colleagues in
February of 2008.
In an effort to promote the retention of early-career teachers between the fall and spring
administrations of the two surveys, I sent each early-career teacher a 25-dollar gift card after they
submitted their surveys. I did not send gift cards to the mentors and colleagues; there is a
substantial amount of research on survey response rates that indicates the promise of a payment
upon completion of a survey is not as effective at boosting response rates relative to sending a
small incentive with the initial request to complete the survey (Dillman, 2007).
Forty-eight percent of the eligible early-career teachers from the initial sample completed
5

two surveys during the 2007-2008 academic year. Specifically, 159 teachers completed the first
survey that I administered, which corresponded to a response rate of 64 percent (I deemed 61 of
the 309 teachers on the initial list that we received from district and diocesan administrators to be
ineligible to participate in the study because they either were not in their first or second year of
teaching or they did not teach in a high school). I sent the second survey to 159 teachers who had
completed the first survey; I received completed second surveys from 119 teachers, which
corresponded to a response rate of 75 percent.
I generated the social network sample of the study by including two questions in the
early-career teacher survey that asked teachers to identify their official mentor and up to six
within-school colleagues with whom they interacted on professional matters. The 119 earlycareer teachers identified 408 unique mentors and colleagues; 28 of the nominees were fellow
early-career teachers whose surveys I had already received. I received 220 completed surveys
from the 380 nominated mentors and colleagues who were not already in the early-career teacher

66

dataset; this corresponds to a response rate of 70 percent for this portion of the study (I deemed
66 of the 380 nominated individuals to be ineligible to participate in the study either because
they were administrators or they did not work at the same school as the early-career teachers who
nominated them).
Measures
The first early-career teacher survey contained questions on teachers’ teaching experience
and assignments, the frequency and nature of their interactions with their mentors and
colleagues, the frequency with which they had participated in various professional development
activities, their teaching practices, their perceptions of their schools and departments, their hiring
and work experiences, and their career plans. The second early-career teacher survey contained
similar items to the first survey as well as questions on teachers’ preparations for teaching, their
experiences with teachers’ associations, and their race and gender. The mentor-colleague survey
consisted of questions on teachers’ teaching positions and practices, their perceptions of their
schools and departments, their work experiences, their preparations for teaching, their career
plans, and their race and gender. In total, the first early-career teacher survey contained
approximately 170 items; the second early-career teacher survey included about 240 items; the
mentor-colleague teacher survey contained approximately 150 items. The questions for the first
and second early-career teacher survey, as well as the mentor-colleague teacher survey, are
presented in Appendices A, B, and C, respectively.
Dependent variables. I measured two types of teacher commitment in each of the
surveys; the measures from the second early-career teacher survey serve as the outcome
variables in the analyses that I present in this chapter. In one set of questions, adapted from items
used by Bryk and Schneider (2002) in their study of the role of trust in schools, I measured the

67

extent to which teachers were committed to the schools in which they taught. I combined the
responses to the individual ordinal items into a single interval scale using Rasch modeling (Bond
& Fox, 2007); Table 3.1 provides the items and their technical properties.

Table 3.1
Technical Properties of Commitment-to-School Measure for Fall and Spring
Fall Measure
Item

Spring Measure

Difficulty

Infit ZSTD

Difficulty

Infit ZSTD

I wouldn’t want to work in any other
school

1.91

1.10

1.06

.70

I would recommend this school to parents
seeking a place for their child

-.93

-.50

-1.19

-.30

I usually look forward to each working day
at this school

-.98

.83

.12

-.30

Person Reliability:

0.60

0.67

Note. Response options included strongly disagree, disagree, agree, strongly agree, and not sure.

In addition to commitment to school, I also measured the extent to which teachers were
committed to the profession of teaching. To measure commitment to the profession, I used an
item that has been used in the administrations of the Schools and Staffing Survey by the National
Center for Education Statistics (National Center for Educational Statistics, n.d.). The item asks
respondents, “if you could go back to your college days and start over again, would you become
a teacher or not?”, and offers five response options: certainly would become a teacher; probably
would become a teacher; chances about even for and against; probably would not become a
teacher; and certainly would not become a teacher.

68

Independent variables. For the analyses in this paper, the predictors of interest were
measures of early-career teachers’ interactions with their mentors and school-based colleagues.
In particular, the analytic models included variables that indicated whether early-career teachers
had mentors and colleagues and the frequency with which early-career teachers interacted with
their mentors and colleagues. As well, the models included variables that assessed the influence
of mentors and colleagues after controlling for characteristics of the mentors and colleagues.
In addition to the predictors of interest, the analytic models also contained several
variables that served as statistical controls. At the teacher level, controls included early-career
teacher’ years of experience, gender, race, formal preparation, certification status, and teaching
assignment; at the school level, controls included sector, school size, percent of students who
were racial minorities, and school location. Last, the models controlled for early-career teachers’
prior levels of commitment by including the measures of commitment captured in the first earlycareer teacher survey.
Results
Descriptive Statistics
Characteristics of the early-career teachers. The 119 early-career teachers in this study
taught an assortment of grade levels and content areas. For grade levels taught, 73 percent of the
teachers taught freshmen and 76 percent taught sophomores; 67 percent of the teachers taught
juniors and 61 percent taught seniors; 4 percent of the teachers taught at least one grade at the
elementary level. In terms of content areas, 24 percent of the teachers taught mathematics, while
19 percent taught language arts. In total, 76 percent of the teachers taught one or more of the core
content areas; others taught world languages, humanities, business or technology education, art,

69

physical education, family and consumer science, or agriscience. None of the teachers taught
special, technical, or vocational education.
As well, the early-career teachers possessed a variety of certifications and paths of entry
into the teaching profession. For certification, 71 percent of the teachers possessed either a
regular or a probationary state teaching certificate, while 19 percent had either a provisional,
temporary, or emergency certificate. Ten percent of the teachers held no state certification. In
terms of paths to entry into the profession, a majority of the teachers in the sample (64 percent)
entered teaching by completing a bachelor’s degree granting program (including programs that
spanned five years). Others entered the profession by completing a master’s degree granting
program (17 percent) or by completing a program designed to expedite the transition of nonteachers into the teaching profession (13 percent). Seven percent of the teachers entered the
profession without completing any formal program (although some of these teachers had
completed at least some coursework related to teaching).
The early-career teachers represented relatively similar numbers of first and second year
teachers, and a majority of the teachers were Caucasian and female. Specifically, of the 119
early-career teachers, 59 of them were in their first year of teaching, while 60 were in their
second. Sixty-six percent of the teachers were female, and 97 percent of the teachers were
Caucasian.
6

Characteristics of the mentors and colleagues. Like the early-career teachers, the
mentors and colleagues taught a variety of grade levels and subject matters. For grade levels, 65
percent of the colleagues taught freshmen and 67 percent taught sophomores; 74 percent of them
taught juniors and 72 percent taught seniors; 3 percent of the colleagues taught at least one grade
at the elementary level. In terms of content areas, 25 percent of the colleagues taught language

70

arts, 19 percent taught math, 24 percent taught science, and 12 percent taught social studies; 25
percent of the colleagues taught at least one course outside of the core content areas.
A majority of the mentors and colleagues held standard state certification and entered the
profession through a bachelor’s degree program. For certification status, 86 percent of the
mentors and colleagues possessed either a regular or probationary state teaching certificate,
while three percent had either a provisional, temporary, or emergency certificate. Eleven percent
of the mentors and colleagues held no state certification; five percent possessed a certificate from
the National Board for Professional Teaching Standards. In terms of paths to entry into the
profession, 63 percent of the mentors and colleagues entered teaching by completing a bachelor’s
degree granting program (including programs that spanned five years). Others entered the
profession by completing a master’s degree granting program (19 percent) or by completing a
program designed to expedite the transition of non-teachers into the teaching profession (four
percent). Thirteen percent of the mentors and colleagues entered the profession without
completing any formal program.
Characteristics of the schools. The 44 schools in this study varied by school sector and in
terms of size and percent of students belonging to a racial minority. Most of the Catholic schools
enrolled between 600 and 1200 students (the mean number of students was 624 with a standard
deviation of 264), while half of the public schools enrolled more than 1800 students (the mean
number of students was 1811 with a standard deviation of 753). In terms of the percentage of
students who belonged to a racial minority, almost all of the Catholic schools had student bodies
that were less than 40 percent racial minority (frequency of 92 percent), while over half of the
public schools had student bodies that were more than 40 percent minority (frequency of 60
percent). The mean percentage of students who belonged to a racial minority in the Catholic

71

schools was 19 percent (with a standard deviation of 22); the mean percentage of students
belonging to a racial minority in the public schools was 48 percent (with a standard deviation of
23).
Characteristics of the social networks. In this study, I captured information on the
frequency and nature of the interactions that the early-career teachers had with their mentors and
colleagues. In Figure 3.1, I display the proportions of early-career teachers who had mentors and
colleagues in the fall and the spring.

94
71

98

71

Fall Spring
Mentors

Fall Spring
Colleagues

Figure 3.1. Proportions of early-career teachers who indicated on the fall and the spring surveys
that they had a mentor or that they interacted with at least one other colleague at their school on
professional matters.

Seventy-one percent (n = 85) of the early-career teachers had a mentor in the 2007-2008
school year; 69 percent (n = 59) of the mentors completed a survey. For all of the nominated
mentors, between 71 and 76 percent taught the same content area as their mentee, between 41
and 69 percent had a bachelor’s degree in the content area that the mentee taught, and between

72

7

56 and 86 percent possessed certification in the content area that the mentee taught. Between 44
and 87 percent of the early-career teachers and the colleagues with whom they interacted taught
the same content areas. As well, between 31 and 78 percent of the colleagues possessed a
bachelor’s degree in the content area that their nominating early-career teacher taught, and
between 41 and 89 percent of the colleagues were certified in the content area taught by the
early-career teacher who identified them as a colleague.
Mentees interacted with their mentors more frequently in the months of September and
October than in the months of April and May. In the fall, 13 percent of the mentees engaged in
professional interactions with their mentors every day, 54 percent interacted with their mentors at
least once each week, and two percent never interacted with their mentors. In the spring, nine
percent of the mentees interacted with their mentors on professional matters every day, 46
percent interacted with their mentors at least once each week, and five percent never interacted
with their mentors. In the fall, mentees most frequently discussed teaching strategies, curriculum,
and managing student behavior with their mentors; the means (with standard deviations in
parentheses) for number of days per month that mentees and mentors interacted on those topics
were 3.4 (4.8), 3.2 (4.8), and 3.0 (4.2), respectively. In the spring, mentees most frequently
discussed curriculum and teaching strategies (but not managing student behavior) with their
mentors; the means (with standard deviations) for the number of days per month that menteementor pairs interacted on curriculum and teaching strategies were 2.4 (4.1) and 2.2 (3.3),
respectively.
Early-career teachers reported interacting with approximately the same number of
colleagues in the months of September and October as in the months of April and May. Ninetyfour percent of the early-career teachers indicated that they engaged in professional interactions

73

in the months of September and October with one or more colleagues who worked at their
school; in the survey instrument, I defined professional interactions as interactions about
curriculum, instruction, students, school policies, parents, et cetera. A majority (65 percent) of
the early-career teachers indicated that they interacted with at least three colleagues in their
schools in the months of September and October, and 23 percent reported engaging in
professional interactions with six colleagues; six was the maximum number of colleagues that
teachers could identify on the survey instrument. In April and May, 98 percent of the earlycareer teachers indicated that they engaged in professional interactions with one or more
colleagues who worked at their school, 69 percent reported engaging in professional interactions
with at least three colleagues, and 20 percent indicated interacting with six colleagues.
Early-career teachers reported interacting with their colleagues on professional topics
with approximately the same frequency in September and October and in April and May. In the
fall, 23 percent of the early-career teachers engaged in professional interactions with one or more
of their school-based colleagues every day, 76 percent interacted with one or more of their
colleagues at least once a week, and 12 percent interacted with one or more of their colleagues
less than once a month or not at all. In the spring, 23 percent of the early-career teachers
interacted with one or more of their school-based colleagues on professional matters every day,
81 percent interacted with one or more of their colleagues at least once a week, and four percent
interacted with one or more of their colleagues less than once a month or not at all.
The professional topics on which the early-career teachers and their colleagues interacted
most frequently were the same in the fall and the spring. In the months of September and
October, the early-career teachers most frequently discussed curriculum, student behavior, and
teaching strategies with one or more of their school-based colleagues; the means (with standard

74

deviations in parentheses) for the number of days per month that early-career teachers interacted
with their colleagues on those topics were 5.8 (6.5), 5.6 (6.4), and 4.9 (5.3), respectively. In the
months of April and May, those same topics were again the most frequently discussed, albeit less
frequently in each case: the means (with standard deviations) for the number of days per month
that the early-career teachers and their colleagues interacted about curriculum, student behavior,
and teaching strategies were 5.0 (5.3), 4.8 (5.7), and 4.8 (5.4), respectively.
In both the fall and the spring surveys, the early-career teachers indicated the extent to
which they believed that they could access, when necessary, their mentors and colleagues for
information and advice on a number of professional topics. In the fall, 80 percent or more of the
early-career teachers agreed that they had access to their mentors and colleagues on each of the
eight professional topics. The topics that received the most agreement were student behavior and
teaching strategies. In the spring, 89 percent or more of the early-career teachers agreed that they
had access to their mentors and colleagues on each of the topics. Similar to the fall, the topics
that received the most agreement were student behavior and teaching strategies.
Early-career teachers rated the support that they received on professional topics from
their colleagues higher than the support they received from their mentors; this trend was found in
both the fall and the spring surveys. On a scale of not, somewhat, very, and extremely important,
75 percent of the early-career teachers who had mentors indicated that the professional support
they received from their mentors was very or extremely important in the months of September
and October. For the same months, 87 percent of the early-career teachers who interacted with
colleagues indicated that the professional support that they received was very or extremely
important. For the months of April and May, 59 percent of the early-career teachers who had
mentors rated the professional support that they received from their mentors as very or extremely

75

important. The proportion of early-career teachers who interacted with colleagues and who rated
the professional support they received as very or extremely important was the same in the spring
as in the fall (i.e., 87 percent).
Inferential Statistics
Many of the participants in this study work in the same schools as other participants. If
we assume that the school context affects the way in which the participants responded to
questions on the survey, then the responses of the different participants are not independent of
each other. And it seems reasonable, given the types of questions addressed in this study (e.g.,
school commitment), that the school context would affect participant responses. Independence of
responses is one of the assumptions of single-level linear regression models. Because the
assumption of independence is likely violated in this study, using hierarchical models, which
takes account of the dependence among responses, is prudent for this study (Raudenbush &
Bryk, 2002).
Multilevel linear regression models. To explore the relationships between the earlycareer teachers’ levels of commitment to their schools in the spring and their interactions with
their mentors and colleagues, I developed a series of maximum likelihood (ML), multi-level,
multiple linear regression models. In the models, the outcome was commitment to the school as
measured in the spring. In one of models, the predictor variables of interest were network
variables as measured on the early-career teachers’ surveys; in additional models, I modeled
network variables as measured on the surveys of the mentors and colleagues.
To begin, I estimated a fully unconditional model that used the measure of an earlycareer teacher’s level of commitment to their school as the outcome variable. Equation 3.1

76

displays the combined form of the model (the hierarchical form of the model is located in Table
E1 in Appendix E):

(SCH_COMMIT_SPR)ij =

γ00 + u0j + rij.

(3.1)

The dependent variable is the level of commitment to the school for early-career teacher i in
school j in the spring; γ00 is the average of the school means for the dependent variable; u0j is the
unique effect of each school; rij is the unique effect of each teacher. I present the results of
Equation 3.1 in Table 3.2. Based on the results, I conclude that a significant school effect exists
2

for the data (χ (42, n=114)=74.174, p=0.002); the significant school effect indicates that using a
multilevel framework to evaluate the effects of various predictors on the outcome variable is
necessary and appropriate. The intraclass correlation coefficient for Equation 3.1 is .222, which
means that 22 percent of the variation in the dependent variable lies between schools.

77

Table 3.2
Results of Fully Unconditional Model for Commitment to School Outcome
Fixed Effects

Coefficient

SE

t-ratio

Constant, γ00

61.576

3.222

19.106*** <0.001

Random Effects

Variance
Component

df

χ

p-value

School effect, u0j

180.187

42

74.174**

0.002

Teacher effect, rij

630.260

2

p-value

Note. Level-1 n = 118; level-2 n = 43.
**p < .01; *** p < .001.
Next, I built a model that related early-career teachers’ levels of commitment to their
schools to the frequency with which they communicated with their mentors and colleagues. The
variables of most conceptual interest in this model were whether an early-career teacher had a
mentor, MENT_YN; how frequently mentees and mentors communicated, MENT_FREQ;
whether an early-career teacher communicated with within-school colleagues on professional
topics, COLL_YN; and the frequency with which early-career teachers communicated with their
colleagues, COLL_FREQ. In addition to those predictors, I tested and included a number of
control variables. The controls included the early-career teachers’ prior levels of commitment to
their schools, SCH_COMMIT_FALL; whether an early-career teacher received a bachelor’s
degree through a teacher preparation program, TEDEG_YN; and whether the teacher was
female, FEMALE_YN. I also included a school-level indicator of whether more than 40 percent
of the students at the school belonged to a racial minority, OVER40SCH_YN.

78

I designated six of the level-1 predictors as fixed and one as random. Specifically, I
assumed that the relationship between the outcome and an early-career teacher’s level of
commitment in the fall, whether the teacher received a bachelor’s degree through a teacher
preparation program, the frequency with which the teacher interacted with a mentor and
colleagues, whether the teacher communicated with colleagues, and the teacher’s gender was
consistent across schools. I assumed that the effect of having a mentor on the spring measure of
school commitment could be influenced by unspecified characteristics of the school. As such, I
treated MENT_YN as a random effect. This decision was supported by a test of the homogeneity
of level-1 variance. When MENT_YN was treated as a fixed effect, I rejected the null hypothesis
2

that the level-1 variance in the model was homogenous across schools (χ (32, n=43)=69.011,
p<0.001). However, when I allowed MENT_YN to vary randomly across schools, I retained the
2

null hypothesis that the level-1 variance in the model was homogenous across schools (χ (6,
n=10)=5.584, p>.500).
I excluded measures if, after they were added to the model, they failed to improve the
total amount of variance that was explained by the model by less than two percent. For the level1 predictors, I excluded indicators of whether an early-career teacher had regular state
certification, taught math or science, was in the first year of full-time teaching, and was
Caucasian. For the level-2 predictors, I excluded indicators of whether the school was located in
a central city, was public, and had more 1200 students enrolled.
Equation 3.2 displays the combined form of the model (the hierarchical form of the
model is located in Table E2 in Appendix E):

79

(SCH_COMMIT_SPR)ij =

γ00 + γ01(OVER40SCH_YN)j

(3.2)

+ γ10(SCH_COMMIT_FALL)ij + γ20(MENT_YN)ij
+ γ30(MENT_FREQ)ij + γ40(COLL_YN)ij
+ γ50(COLL_FREQ)ij + γ60(TEDEG_YN)ij
+ u0j + u2j(MENT_YN)ij + rij.

The dependent variable is the same as in Equation 3.1. The intercept, γ00, is the adjusted school
mean for schools with less than 40 percent of their students belonging to a racial minority and
controlling for the level-1 predictors. The parameter γ01 is the effect of schools with more than
40 percent of their students belonging to a racial minority; the parameters γ10 through γ60 are the
effects of their respective variables on the outcome. The components u0j, u2j, rij represent the
residual error among schools conditional on OVER40SCH_YN, the residual error among schools
on MENT_YN, and the residual error among teachers conditional on the teacher-level predictors,
respectively. The descriptive statistics for the variables in Equation 3.2 are in Table 3.3; the zeroorder correlations are in Table 3.4.

80

Table 3.3
Descriptive Statistics for Model Estimating Effects of Communicating with Mentors and
Colleagues on School Commitment
Name
SCH_COMMIT_SPR

SCH_COMMIT_FALL
MENT_YN
MENT_FREQ
COLL_YN

COLL_FREQ

TEDEG_YN

OVER40SCH_YN

Mean
61.54

SD
28.42

Min.

Max.

Description

.00

Scale measure of level of
100.00 commitment to the school in the
spring

63.34

27.21

.00

Scale measure of level of
100.00 commitment to the school in the
fall

.73

.45

.00

1.00 1 if have mentor; 0 otherwise

32.93

25.99

.00

100.00

.94

.24

.00

1.00

48.03

18.75

.00

Scale measure of the frequency
100.00 of communicating with one or
more colleagues

.63

.48

.00

1 if obtained bachelor’s through
1.00 teacher preparation program; 0
otherwise

.00

1 if 40% or more of the students
1.00 at the school belong to a racial
minority; 0 otherwise

.30

.46

Note. Level-1 n = 114; level-2 n = 43.

81

Scale measure of the frequency
of communicating with a mentor
1 if communicate with
colleagues; 0 otherwise

Table 3.4
Correlation Matrix for Model Estimating Effects of Communicating with Mentors and
Colleagues on School Commitment
Variable

1

2

3

4

5

6

7

8

1. SCH_COMMIT_SPR
2. SCH_COMMIT_FALL

.793***

3. MENT_YN

.014

.024

4. MENT_FREQ

.143

.142

5. COLL_YN

.226*

.226* .172^

.192*

6. COLL_FREQ

.163

.150

.194*

.361*** .417***

7. TEDEG_YN

.203*

.080

.187*

.022

.108

.137

8. OVER40SCH_YN

-.093

-.089

.042

.065

.163^

.209* .046

.728***

Note. Level-1 n = 114; level-2 n = 43. Values for comparisons between two variables in which
both variables are measured on interval scales are Pearson correlation coefficients; values for
comparisons between variables in which at least one of the variables is measured on a binary
scale are Spearman correlation coefficients.
^p < .10; *p < .05; ***p < .001.

The results of the model that estimates Equation 3.2 are displayed in Table 3.5. Among
the fixed effects, measures of an early-career teacher’s prior level of commitment to the school,
whether an early-career teacher had a mentor, the frequency with which a mentee-mentor pair
interacted, and whether the early-career teacher received a bachelor’s degree through a teacher
preparation program are all statistically significantly related to the measure of an early-career
teacher’s level of commitment to the school in the spring. I cannot reject the null hypotheses that
the effects on the outcome for the indicator for whether an early-career teacher communicates

82

with colleagues, nor the frequency with which an early-career teacher interacts with colleagues
are different from zero.
Because the outcome measure and the continuous predictors are on interval scales that
stretch from 0 to 100, the coefficients can be interpreted in terms of percents of the scale range.
Specifically, the unstandardized coefficient for SCH_COMMIT_FALL indicates that, ceteris
paribus, a one percent of the scale range increase in the fall level of school commitment is
expected to result in a .787 (less than one) percent of the scale range increase in the spring level
of school commitment; a one percent of the scale range increase in the frequency with which
mentee-mentors pairs interact is expected to result in a .213 (less than one) percent of the scale
range increase in the outcome when all of the other variables are held constant. As well, ceteris
paribus, an early-career teacher who has a mentor is expected to have a spring level of
commitment to the school that is 12.530 percent of the scale range less than a teacher who does
not have a mentor. In addition, all else equal, an early-career teacher who received a bachelor’s
degree as part of a teacher preparation program is expected to have a school commitment level in
the spring that is 8.380 percent of the scale range greater than an early-career teacher who did not
receive a bachelor’s degree through a teacher preparation program.
The random effects of Equation 3.2 are statistically significant. In particular, the variance
component for schools indicates that significant variation in the outcome exists across schools
2

even after the teacher-level controls are included in the model (χ (8, n=10)=18.485, p=.018). As
well, the variance component for MENT_YN indicates that the effect of having a mentor on
2

commitment to the school in the spring varies significantly across schools (χ (9, n=10)=18.778,
p=.027).

83

Table 3.5
Results of Model Estimating Effects of Communicating with Mentors and Colleagues on School
Commitment
Variance
Explained

Fixed Effect

Coefficient

SE

t-ratio

p-value

Constant, γ00

.242

7.487

.032

.974

OVER40SCH_YN, γ01

-1.444

3.866

-.374

.711

.027

SCH_COMMIT_FALL, γ10

.787

.062

12.777***

< .001

.561

MENT_YN, γ20

-12.530

6.905

-1.815^

.077

.066

MENT_FREQ, γ30

.213

.111

1.926^

.067

.205

COLL_YN, γ40

14.839

8.986

1.651

.112

.005

COLL_FREQ, γ50

-.112

.123

-.908

.373

.039

TEDEG_YN, γ60

8.380

3.410

2.458*

.022

.049

Random Effect

Variance
(SD)

df

χ

p-value

School effect, u0j

129.989
(11.401)

8

18.485*

.018

MENT_YN, u2j

324.742
(18.020)

9

18.778*

.027

Teacher effect, rij

237.605
(15.414)

Note. Level-1 n = 114; level-2 n = 43.
^p < .10; *p < .05; ***p < .001.

84

2

In Equation 3.2, I modeled the effects of having mentors and colleagues, as well as
communicating with mentors and colleagues, on the spring levels of school commitment for
early-career teachers. In Equation 3.3 and Equation 3.4 (below), I model those effects after
controlling for certain characteristics of the mentors and colleagues, respectively. In particular, I
used Equations 3.3 and 3.4 to investigate the relationships between the early-career teachers’
levels of commitment to their schools in the spring and indictors for whether early-career
teachers taught the same content area and grade levels as their mentors and colleagues, the
commitment levels of their mentors and colleagues, and the interaction between the frequency
with which the early-career teachers communicated with their mentors and colleagues and the
commitment levels of those mentors and colleagues. The latter is defined as an influence term in
social network analyses.
In the sample, there are 59 and 94 early-career teachers for whom I obtained a survey
from their mentor and one or more colleagues, respectively. Green (1991) suggested that a
sample with 63 units would support two predictors in a multiple linear regression model with
medium effect sizes, while a sample of 97 would support six predictors. Given this, I decided to
separate the evaluations of the mentors and colleagues into different models in order to maximize
the number of covariates from Equation 3.2 that I could assess. As well, I decided to create
separate models for each of the network predictors. In an effort to balance under- and overspecifying the models, I include two predictors in the mentor models and a maximum of six
predictors in the colleague models.
Equation 3.3 displays the combined form of a model that estimates the effects of mentor
characteristics on mentee levels of commitment to their schools (the hierarchical form of the
model is located in Table E3 in Appendix E):

85

(SCH_COMMIT_SPR)ij =

γ00 + γ10(SCH_COMMIT_FALL)ij

(3.3)

+ γ20(MENT_CHAR)ij + u0j + rij.
The dependent variable is the same as in Equation 3.1. The intercept, γ00, is the adjusted school
mean controlling for the level-1 predictors. The parameters γ10 and γ20 are the effects of their
respective variables on the outcome. The components u0j and rij represent the residual error
among schools and among teachers, respectively; the residual error among teachers is
conditional on the teacher-level predictors.
The independent variable SCH_COMMIT_FALL retains its definition from Equation
3.2. The variable MENT_CHAR is a placeholder in the equation for one of four variables that
represent some characteristic of the mentor. Because of the sample size, I evaluated Equation 3.3
in four different ways: each time I inserted a different variable in the MENT_CHAR place. The
variables that I evaluated were MENT_CA_YN, which indicates whether the mentor and mentee
taught the same content area (e.g., mathematics); MENT_GL_YN, which indicates whether the
mentor and mentee taught the same grade level; MENT_COMMIT, which is a Rasch-calculated
measure for the commitment level of the mentor; and MENT_INFLUENCE, which is an
interaction term between the frequency with which the mentee communicated with the mentor
and the mentor’s commitment level. The descriptive statistics for Equation 3.3 are displayed in
Table 3.6; the zero-order correlations between the variables in Equation 3.3 are displayed in
Table 3.7.

86

Table 3.6
Descriptive Statistics for Model Estimating Effects of Mentor Characteristics on School
Commitment
Name
SCH_COMMIT_SPR

SCH_COMMIT_FALL

Mean
66.92

SD
25.89

Min.

Max.

Description

1.84

Scale measure of level of
100.00 commitment to the school in the
spring
Scale measure of level of
100.00 commitment to the school in the
fall

67.58

24.25

.00

MENT_CA_YN

.76

.43

.00

1.00

1 if mentor and mentee teach
same content area; 0 otherwise

MENT_GL_YN

.90

.30

.00

1.00

1 if mentor and mentee teach
same grade level; 0 otherwise

MENT_COMMIT

MENT_INFLUENCE

73.01

45.71

32.00

26.44

.00

.00

Note. Level-1 n = 59; level-2 n = 29.

87

Scale measure for mentor’s
100.00 level of commitment in the
winter
Scale measure of the product
between mentor’s commitment
95.33 level and frequency of
interaction between mentor and
mentee

Table 3.7
Correlation Matrix for Model Estimating Effects of Mentor Characteristics on School
Commitment
Variable

1

2

3

4

5

6

1. SCH_COMMIT_SPR
2. SCH_COMMIT_FALL

.725***

3. MENT_CA_YN

.089

.331*

4. MENT_GL_YN

-.015

-.109

-.056

5. MENT_COMMIT

.295*

.245^

.151

-.198

6. MENT_INFLUENCE

.259*

.237^

.398**

.003

.774***

Note. Level-1 n = 59; level-2 n = 29. Values for comparisons between two variables in which
both variables are measured on interval scales are Pearson correlation coefficients; values for
comparisons between variables in which at least one of the variables is measured on a binary
scale are Spearman correlation coefficients.
^ p < .10; * p < .05; ** p < .01; *** p < .001.

The results of the models that estimate Equation 3.3 are displayed in Tables 3.8a and
3.8b. In each of the models, the prior measure of school commitment is statistically significantly
related to the spring measure of school commitment. The unstandardized coefficients indicate
that a one percent of the scale range increase in the fall measure of school commitment is
expected to result in an increase in school commitment in the spring of between .741 and .851
percent of the scale range, ceteris paribus. In terms of standard deviations, a one standard
deviation increase in the fall measure of school commitment is expected to result in an increase
in school commitment in the spring of between .694 and .797 standard deviations, all else equal.
In terms of variance explained, the fall measure of school commitment accounts for between 26
and 30 percent of the variation in the outcome.

88

In addition to the prior level of commitment, the indicator for whether the mentor and the
mentee teach the same content area is statistically significantly related to the outcome. The
unstandardized coefficient indicates that, everything else equal, early-career teachers who have a
mentor who teaches the same content area that they do are expected to have a spring level of
commitment that is 12.400 percent of the scale range less than early-career teachers who have a
mentor who does not teach the same content area that they do. In terms of standard deviations,
early-career teachers who have a mentor who teaches the same content area as they do are
expected to have a spring level of commitment to their schools that is .479 standard deviations
less than the spring level of commitment of teachers whose mentors do not teach the same
content area that they do, ceteris paribus. The proportion of variance explained by the indicator
for whether mentees and mentors teach the same content area is approximately four percent.
For the measures of whether mentors and mentees teach the same grade level, the levels
of commitment of the mentors, and the interaction between the levels of commitment of the
mentor and the frequency with which mentors and mentees communicate, I cannot reject the null
hypothesis that their coefficients are different from zero when evaluated for a p-value of 0.10.
The small sample size used in the analysis, which results in a low-power analysis, means that the
statistically insignificant findings for the measures of the commitment levels of the mentors and
the interaction between the commitment of mentors and the frequency of communication
between mentors and mentees should be interpreted with care. The ratio between the coefficient
and the standard error of the grade level match variable indicates that there is less risk in
committing a Type II error by interpreting that variable as being unrelated to the outcome
measure, ceteris paribus. Last, in each of the models, I cannot reject the null hypothesis that the

89

variation in the outcome across schools is different from zero after entering the teacher-level
variables into the models.

Table 3.8a
Results of Models Estimating Effects of Content-Area and Grade-Level Matches with Mentors on
School Commitment
Content Area Match Model

Grade Level Match Model

Fixed Effect

Coefficient
(SE)

Coefficient
(SE)

Constant, γ00

19.182**
(7.322)

SCH_COMMIT_FALL, γ10

.851***
(.104)

.287

MENT_CA_YN, γ20

-12.400**
(5.815)

.042

Variance
Explained

Variance
Explained

12.106
(10.691)
0.777***
(.103)

2.846
(7.798)

MENT_GL_YN, γ20

Random Effect

Variance
(SD)

School effect, u0j

18.370
(4.286)

Teacher effect, rij

286.790
(16.935)

2

[df] χ

[28] 33.763

.260

.000

Variance
(SD)
26.175
(5.116)
303.869
(17.432)

Note. Level-1 n = 59; level-2 n = 29.
^p < .10; *p < .05; **p < .01; ***p < .001.

90

2

[df] χ

[28] 32.304

Table 3.8b
Results of Models Estimating Effects of Mentor Commitment and Influence on School
Commitment
Mentor Commitment Model
Fixed Effect

Coefficient
(SE)

Variance
Explained

Constant, γ00

9.670
(8.234)

SCH_COMMIT_FALL, γ10

.741***
(.103)

.283

MENT_COMMIT, γ20

.101
(.076)

Mentor Influence Model
Coefficient
(SE)

Variance
Explained

.003

12.032
(7.870)
0.748***
(.104)

.099
(.092)

MENT_INFLUENCE, γ20

Random Effect

Variance
(SD)

School effect, u0j

21.551
(4.642)

Teacher effect, rij

298.601
(17.280)

2

[df] χ

[28] 30.382

.295

.015

Variance
(SD)
29.929
(5.471)

2

[df] χ

[28] 31.834

295.057
(17.177)

Note. Level-1 n = 59; level-2 n = 29.
^p < .10; *p < .05; ***p < .001.

Equation 3.4 displays the combined form of a model estimating the effects the
characteristics of the colleagues of the early-career teachers on the spring levels of commitment
to school for the early-career teachers (the hierarchical form of the model is located in Table E4
in Appendix E). In building the model, I retained the predictors of interest whatever their
significance levels and proportion of variance explained. For the covariates, I included predictors
91

that explained two percent or more of the level-1 variation in the outcome when I regressed the
outcome on the predictor and the prior level of commitment; I excluded the level-2 predictors
because none of them accounted for the between-school variance in a statistically significantly
way.

(SCH_COMMIT_SPR)ij =

γ00 + γ10(SCH_COMMIT_FALL)ij

(3.4)

+ γ20(COLL_CHAR)ij + γ30(TEDEG_YN)ij + u0j + rij.

The outcome, SCH_COMMIT_SPR, retains its definition from Equation 3.1. The intercept, γ00,
is the adjusted school mean controlling for the level-1 predictors; the parameters γ10 through γ30
are the effects of their respective variables on the outcome. The components u0j and rij represent
the residual error among schools and the residual error among teachers conditional on the
teacher-level predictors, respectively.
The independent variable SCH_COMMIT_FALL is the same as in Equation 3.2. The
variable COLL_CHAR holds the place of one of four variables that represent some characteristic
of the mentor. As with Equation 3.3, I evaluated Equation 3.4 in four different ways. The
variables that I evaluated were COLL_CA_YN, which indicates whether one or more of the
colleagues of the early-career teacher taught the same content area as the early-career teacher
(e.g., mathematics); COLL_GL_YN, which indicates whether one or more of the colleagues of
the early-career teacher taught the same grade level as the early-career teacher;
COLL_COMMIT, which is a Rasch-calculated measure of the commitment level of the
colleagues, averaged across colleagues; and COLL_INFLUENCE, which is an interaction term
between the frequency with which the early-career teacher communicated, on average, with the
92

colleagues and the average commitment level of the colleagues. The descriptive statistics for
Equation 3.4 are displayed in Table 3.9; the zero-order correlations between the variables in
Equation 3.4 are displayed in Table 3.10.

Table 3.9
Descriptive Statistics for Model Estimating Effects of Colleague Characteristics on School
Commitment
Name

Mean

SD

Min.

Max.

Description

Scale measure of level of
1.84 100.00 commitment to the school in the
spring

SCH_COMMIT_SPR

64.33

26.27

SCH_COMMIT_FALL

65.64

25.30

COLL_CA_YN

.90

.30

.00

1 if one or more colleagues teach
1.00 the same content area as the earlycareer teacher; 0 otherwise

COLL_GL_YN

.96

.21

.00

1 if one or more colleagues teach
1.00 the same grade level as the earlycareer teacher; 0 otherwise

COLL_COMMIT

COLL_INFLUENCE

TEDEG_YN

73.35

64.69

10.89 100.00

Scale measure of level of
commitment to the school in the fall

20.73

Scale measure for average of the
.00 100.00 colleagues’ levels of commitment
in the winter

23.66

Scale measure of the product
between the average commitment
level of the colleagues and the
.00 121.48
average frequency of interaction
between the early-career teacher
and the colleagues

.64

.48

.00

Note. Level-1 n = 92; level-2 n = 41.

93

1 if obtained bachelor’s through
1.00 teacher preparation program; 0
otherwise

Table 3.10
Correlation Matrix for Model Estimating Effects of Colleague Characteristics on School
Commitment
Variable

1

2

3

4

5

6

7

1. SCH_COMMIT_SPR
2. SCH_COMMIT_FALL

.738***

3. COLL_CA_YN

.093

.083

4. COLL_GL_YN

.024

.076

.109

5. COLL_COMMIT

.160

.086

.086

-.152

6. COLL_INFLUENCE

.104

.084

.059

.034

.779***

7. TEDEG_YN

.202^

.056

-.246*

.063

.062

.150

Note. Level-1 n = 92; level-2 n = 41. Values for comparisons between two variables in which
both variables are measured on interval scales are Pearson correlation coefficients; values for
comparisons between variables in which at least one of the variables is measured on a binary
scale are Spearman correlation coefficients.
^ p < .10; * p < .05; *** p < .001.

I present the results of the models that estimate Equation 3.4 in Tables 3.11a and 3.11b.
The measure of an early-career teacher’s commitment to the school in the fall is statistically
significantly related to the spring measure of the same construct in each of the models.
Specifically, we would expect that a one percent of the scale range increase in the fall level of
commitment would yield between a .750 and .762 (i.e., less than one) percent of the scale range
increase in an early-career teacher’s level of commitment to the school in the spring if all of the
other variables remained constant. In terms of proportion of variance explained, these models
estimate that the fall measure of school commitment explains between 42 and 44 percent of the
variance in the outcome.

94

In three of the four models, the indicator for whether an early career teacher received a
bachelor’s degree through a teacher preparation program is statistically significantly related to
the spring measure of an early-career teacher’s level of commitment to the school when judged
against a p-value of less than 0.10. In the Mentor Commitment Model, the t-ratio is 1.660, which
corresponds to a p-value of 0.103. Given the size of the t-ratio for the indicator, as well as the
consistent findings of statistical significance in the other models, I believe sufficient evidence
exists to interpret the coefficient for the indicator in the Mentor Commitment Model. So based
on the four models, we would expect that a teacher who receives a bachelor’s degree through a
teacher preparation program to have a level of commitment in the spring that was between 6.369
and 7.906 percent of the scale range higher than the level of commitment of an early-career
teacher who did not receive a bachelor’s degree through a teacher preparation program, ceteris
paribus. The amount of variance in the outcome that is explained by TEDEG_YN ranges from
two to three percent.
The colleague network variable in each of the four models is not statistically significantly
related to the outcome measure. That is, I cannot reject each null hypothesis that the effect of the
colleague network variable on the spring level of commitment is different from zero. As well, the
proportion of variance explained in the outcome by the colleague network variables is less than
one percent in each case. In sum, for these models, the colleague network variables are not
explaining spring levels of school commitment for the early-career teachers.

95

Table 3.11a
Results of Models Estimating Effects of Content-Area and Grade-Level Matches with Colleagues
on School Commitment
Content Area Match Model

Grade Level Match Model

Fixed Effect

Coefficient
(SE)

Coefficient
(SE)

Constant, γ00

3.343
(8.131)

SCH_COMMIT_FALL, γ10

.750***
(.073)

.427

COLL_CA_YN, γ20

7.424
(6.396)

.004

Variance
Explained

Variance
Explained

16.603
(9.914)
0.762***
(.074)

-7.038
(9.064)

TEDEG_YN, γ30

7.906^
(3.958)

Random Effect

Variance
(SD)

School effect, u0j

.408
(.638)

Teacher effect, rij

308.683
(17.569)

.032

2

[df] χ

[40] 40.288

-.004

6.915^
(3.852)

COLL_GL_YN, γ20

.431

.023

Variance
(SD)
.683
(.827)
311.04
(17.636)

Note. Level-1 n = 92; level-2 n = 41.
^p < .10; ***p < .001.

96

2

[df] χ

[40] 41.484

Table 3.11b
Results of Models Estimating Effects of Colleague Commitment and Influence on School
Commitment
Mentor Commitment Model
Fixed Effect

Coefficient
(SE)

Variance
Explained

Constant, γ00

2.709
(8.158)

SCH_COMMIT_FALL, γ10

.751***
(.073)

.436

COLL_COMMIT, γ20

.113
(.089)

Mentor Influence Model
Coefficient
(SE)

Variance
Explained

.007

8.597
(7.129)
.756***
(.074)

.029
(.079)

TEDEG_YN, γ30

6.369
(3.836)

Random Effect

Variance
(SD)

School effect, u0j

.638
(.799)

Teacher effect, rij

307.669
(17.541)

.018

2

[df] χ

[40] 40.302

-.009

6.548^
(3.891)

COLL_INFLUENCE, γ20

.418

.019

Variance
(SD)
.632
(.795)

2

[df] χ

[40] 41.196

312.734
(17.684)

Note. Level-1 n = 92; level-2 n = 41.
^p < .10; *p < .05; ***p < .001.

Multilevel logistic regression models. I investigated the effects of the early-career
teachers’ interactions with their mentors and colleagues on the early-career teachers’ spring
levels of commitment to the profession with a series of binomial sampling models. In the

97

models, the outcome variable was commitment to the profession as measured in the spring. In
one of the models, I included network variables as measured on the early-career teachers’
surveys as the predictors. In other models, I included network variables as measured on the
mentor and colleague surveys.
I began by estimating a fully unconditional Bernoulli sampling model with a logit link
function; Equation 3.5 displays the combined form of the model (the hierarchical form of the
model is located in Table E5 in Appendix E):

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) %
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij

!

= γ00 + u0j.

(3.5)

The term on the left side of the equation is the log of the odds of an early-career teacher
reporting being committed to the teaching profession in the spring. The intercept, γ00, is the
average log-odds of commitment across schools and the error term, u0j, represents the variance
between schools in the average log-odds of commitment.
The outcome measure PRO_COMMIT_SPR_YN is a dichotomous variable that takes a
value of one if the early-career teacher indicated that if they went back to college they
“certainly” or “probably” would become a teacher; the variable takes a value of zero if the earlycareer teacher responded that the chances were “about even for or against” or “probably” or
“certainly” would not become a teacher. In Figure 3.2, I display the frequencies for the five
responses to the professional commitment measure when asked of teachers in the fall; in Figure
3.3, I display the frequencies for the five responses to the professional commitment measure
when asked of teachers in the spring.

98

58

41

10

Certainly Probably
would
would

7

0
Chances Probably Certainly
even
would not would not

Figure 3.2. Response frequencies of early-career teachers when asked in the fall: “If you could
go back to your college days and start over again, would you become a teacher or not?”

57

31
19

4
Certainly Probably
would
would

5

Chances Probably Certainly
even
would not would not

Figure 3.3. Response frequencies of early-career teachers when asked in the spring: “If you
could go back to your college days and start over again, would you become a teacher or not?”

99

I present the results of the model that estimates Equation 3.5 in Table 3.12. The
unstandardized coefficient indicates that the expected log-odds of professional commitment in
the spring at an average school is 1.159, which corresponds to an odds ratio of 3.186, which
corresponds to a predicted probability of .761. This means that in the average school, 76 percent
of the early-career teachers reported being committed to the profession in the spring; calculating
a confidence interval around the estimate yields the finding that 95 percent of the schools have
between 56 and 89 percent of their early-career teachers being commitment to the profession in
the spring.

Table 3.12
Results of Fully Unconditional Model for Commitment-to-Profession Outcome
Fixed Effects

Coefficient (SE)

Odds Ratio

t-ratio

p-value

Constant, γ00

1.159 (.232)

3.186

4.990***

< .001

Random Effects

Variance Component (SD)

df

χ

p-value

School effect, u0j

.219 (.467)

42

41.367

> .500

2

Note. Level-1 n = 116; level-2 n = 43.
*** p < .001

Next, I constructed a pair of models that included an indicator of whether an early-career
teacher was committed to the profession in the fall, PRO_COMMIT_FALL_YN, and a set of
network predictors. In one of the models, the network predictors were the indicator for whether
an early-career teacher had a mentor, MENT_YN, and a measure of the frequency with which
the mentor and mentee communicated on various professional topics, MENT_FREQ. In the
100

second model, I also included two network predictors (COLL_YN and COLL_FREQ), but they
related to colleagues rather than mentors. I display the combined form of these models in
Equation 3.6 (the hierarchical form is located in Table E6 in Appendix E), where
NETWORK_YN is a placeholder for the MENT_YN or COLL_YN variables and
NETWORK_FREQ is a placeholder for the MENT_FREQ or COLL_FREQ variables:

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) %
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij

=

γ00 + γ10(PRO_COMMIT_FALL_YN)ij

(3.6)

+ γ20(NETWORK_YN)ij
+ γ30(NETWORK_FREQ)ij + u0j.

!

The outcome variable, PRO_COMMIT_SPR_YN, retains its definition from Equation
3.5. For the mentor model, the intercept, γ00, equals the average log-odds of professional
commitment in the spring across schools for early-career teachers who were not committed to the
profession in the fall, did not have a mentor, and communicated with a mentor a study-wide
average amount. For the colleague model, the intercept has a similar definition except it relates
to colleagues rather than mentors. The parameters γ10 through γ30 are the effects of their
respective variables on the log-odds of the outcome variable; the residual, u0j, is the error among
schools. The descriptive statistics for Equation 3.6 are displayed in Table 3.13; the zero-order
correlations between the variables in Equation 3.6 are displayed in Table 3.14.

101

Table 3.13
Descriptive Statistics for Model Estimating Effects of Communicating with Mentors and
Colleagues on Professional Commitment
Name

Mean

SD

Min.

Max.

Description

PRO_COMMIT_SPR_YN

.76

.43

.00

1.00 1 if committed; 0 otherwise

PRO_COMMIT_FALL_YN

.85

.36

.00

1.00 1 if committed; 0 otherwise

MENT_YN

.72

.45

.00

1.00 1 if have mentor; 0 otherwise

MENT_FREQ

COLL_YN

COLL_FREQ

32.46

26.12

.94

Scale measure of the frequency
.00 100.00 of communicating with a
mentor

.24

47.49

.00

1.00

1 if communicate with
colleagues; 0 otherwise

Scale measure of the frequency
.00 100.00 of communicating with one or
more colleagues

19.15

Note. Level-1 n = 116; level-2 n = 43.

102

Table 3.14
Correlation Matrix for Model Estimating Effects of Communicating with Mentors and
Colleagues on Professional Commitment
Variable

1

2

3

4

5

6

1. PRO_COMMIT_SPR
2. PRO_COMMIT_FALL

.564***

3. MENT_YN

.180^

.063

4. MENT_FREQ

.173^

.066

.740***

5. COLL_YN

.111

.100

.161^

.185*

6. COLL_FREQ

.147

.078

.202*

.350***

.410***

Note. Level-1 n = 116; level-2 n = 43. Values for comparisons between two variables in which
both variables are measured on interval scales are Pearson correlation coefficients; values for
comparisons between variables in which at least one of the variables is measured on a binary
scale are Spearman correlation coefficients.
^p < .10; *p < .05; ***p < .001.

I present the results of the models that estimate Equation 3.6 in Table 3.15. In each of the
models, the fall indicator of whether an early-career teacher is committed to the teaching
profession is statistically significantly related to the outcome. The coefficients indicate that we
would expect an early-career teacher who was committed to the teaching profession in the fall to
have a log-odds of being committed to the teaching profession in the spring that was
approximately 3.9 units greater than the log-odds of being committed to the profession of an
early-career teacher who was not committed in the fall, ceteris paribus. In terms of predicted
probabilities, we would expect approximately 93 percent of the early-career teachers’ who were
committed to the profession in the fall to be committed to the profession in the spring, ceteris
paribus. As well, we would expect approximately 20 percent of the early-career teachers who

103

were not committed to the profession in the fall to be committed to the profession in the spring,
ceteris paribus. Last, I cannot reject the null hypotheses that the coefficients on the network
variables are equal to zero in each of the models.

Table 3.15
Results of Model Estimating Effects of Communicating with Mentors and Colleagues on
Professional Commitment
Mentor Model

Colleague Model

Fixed Effects

Coefficient
(SE)

Odds Ratio

Coefficient
(SE)

Constant, γ00

-2.551
(.884)

.078

-2.717
(1.346)

.066*

PRO_COMMIT_FALL_YN, γ10

3.889
(.809)

48.872***

3.923
(.806)

50.597***

MENT_YN, γ20

.523
(1.107)

1.686

MENT_FREQ, γ30

.010
(.021)

1.010

COLL_YN, γ20

-.125
(1.519)

.882

COLL_FREQ, γ30

.019
(.020)

1.020

Odds Ratio

Random Effects

Variance
(SD)

[df] χ

Variance
(SD)

[df] χ

School effect, u0j

1.183
(1.088)

[42] 45.130

1.324
(1.150)

[42] 47.559

2

Note. Level-1 n = 116; level-2 n = 43.
*** p < .001

104

2

I explored models similar to the one described in Equation 3.6 but by exchanging the
network variables as measured on the early-career teacher surveys for network variables as
measured on the mentor and colleague surveys. Specifically, I investigated the effects of having
a mentor who taught the same content area as the mentee, as well as the interaction effect of the
frequency of communication between mentors and mentees times the commitment level of the
mentors, on the probability of an early-career teacher being committed to the profession in the
spring after controlling for whether they were committed in the fall. I investigated similar effects
for colleagues instead of mentors. In each of the models, I could not reject the null hypotheses
that the network effects were different from zero, and in each of the models, the indicator of
professional commitment in the fall was a statistically significant predictor of professional
commitment in the spring (the tables for these models are available upon request).
Discussion
Contributions of the Study
This study evaluated the effects of early-career teachers’ within-school social networks
on their levels of commitment to their schools and the profession. I found that having a mentor
negatively affected the spring commitment levels of early-career teachers. As well, the frequency
with which the early-career teachers communicated with their mentors affected their levels of
school commitment. I found that having colleagues had positive effects on early-career teachers’
spring levels of commitment to their schools when evaluated for a p-value of less than 0.15. I did
not find a statistically significant relationship between the frequency with which early-career
teachers communicated with their colleagues and their spring school commitment levels. Nor did
I find statistically significant effects for the within-school networks on early-career teachers’

105

levels of commitment to the profession. Last, I found consistent evidence across the models that
teachers who received a bachelor’s degree through a teacher preparation program had higher
levels of commitment to their schools in the spring than did early-career teachers who did not
receive a bachelor’s degree through a teacher preparation program. In this section, I relate these
findings to the framework that guided the design of the study.
Mentors, colleagues, and school commitment. The effect of having a mentor is in the
opposite direction of what I predicted at the outset of the study. In the framework for this paper, I
noted that Portes (1998) wrote about the potential of networks to exert negative consequences on
its members. However, I expected that, in total, the positive flow of resources and their benefits
to early-career teachers would outweigh the constraints placed on them from their network
members. That is, I expected early-career teachers who had and communicated with mentors and
colleagues to have higher levels of school commitment than their peers who did not have
mentors or communicate with colleagues. This study’s finding that early-career teachers who
have a mentor are expected to have a spring level of school commitment that is approximately 13
percent of the scale range less than early-career teachers who do not have a mentor provides
evidence that the potential of the negative consequences of networks is real for early-career
teachers.
At the same time, this study provides a hint of evidence that the potential of the positive
benefits of networks is also real for early-career teachers. The finding (when evaluated at a
significance level of p = 0.15) that early-career teachers who communicate with colleagues are
expected to have a spring level of school commitment that is approximately 15 percent of the
scale range greater than early-career teachers who do not communicate with colleagues points to
the possibility of the supportive nature that networks can provide.

106

Given the different directions of the potential findings for mentors and colleagues, an
obvious question is why are the negative consequences significant for the mentors while the
benefits are significant for the colleagues? Portes (1998) argued that one of the ways for negative
outcomes to arise from networks was through networks’ demands for conformity from their
group members. Portes thought that networks could diminish the autonomy of the individuals
that were in the network by demanding conformance to locally established norms. It is not
difficult to imagine early-career teachers experiencing a tension between the ideas and
perspectives that they bring to their new career and the established norms and processes that
exist in their new employment setting. Early-career teachers could feel restrictions on their
personal freedoms as they become aware of the norms and processes of the school through their
interactions with their networks. But the existence of this tension still does not explain why
mentors had a negative effect and colleagues a positive one.
One possible explanation could reside in the difference in the orientations of the
relationships between the early-career teachers and their mentor vis-à-vis the early-career
teachers and their colleagues. In most cases, the mentor was officially designated by the school
or district/diocese and, consequently, is an agent of the employing unit. Assuming that school or
district/diocesan leaders selected the mentors based on their years of service, experience, or
conformity to the mission of the employing unit, I speculate that at least some of the messages
that flow from the mentors to the early-career teachers are the formal preferences or
requirements of the school or district/diocese. That is, I imagine that at least some of the
resources that mentors provide to mentees are not done so because of requests that were initiated
by the mentees for supportive materials, but rather because the employing unit required the
implementation of the resources. On the other hand, the employing unit does not assign

107

colleagues to the early-career teachers. Instead, colleagues are individuals with whom the earlycareer teachers communicate as necessary. It could be that the early-career teachers seek out
colleagues to help them make sense of the pressures that they receive from the employing unit,
possibly via the mentor. In terms of Lin’s (1999) definitions of the elements of social capital,
early-career teachers activate resources in the network by communicating with colleagues; the
return to those early-career teachers who activate resources is a higher level of commitment to
their schools.
One potential inconsistency with my assessment that mentors have the effect of providing
more pressure and less support while colleagues provide more support and less pressure is the
finding that increasing the amount of communication between mentors and mentees results in an
expectation that mentees will have higher levels of spring commitment to the school. The results
of the model estimating Equation 3.2 indicated that a one percent increase in the frequency in
communication between mentees and mentors was expected to yield a 0.2 percent of the scale
range increase in the spring level of school commitment in the mentees. One resolution for this
seeming inconsistency is the possibility that the more mentors and mentees interact, the more
supportive the relationship becomes. That is, perhaps increasing the frequency of interaction
between the mentor and the mentee moves the relationship beyond the mentor largely
communicating the expectations of the employing unit and into a relationship in which the
mentor helps the mentee make sense of the expectations of the employing unit.
Another striking finding from this study is the indication that the early-career teacher
who teaches the same content area as the mentor is expected to have a spring level of
commitment to the school that is lower than the early-career teacher who does not teach the same
content area as the mentor, ceteris paribus. I expected that the positive flow of resources from

108

mentors to mentees would outweigh the negative flow of constraints. Instead, an early-career
teacher who teaches the same content area as the mentor is expected to have a spring level of
school commitment that is 12 percent of the scale range less than an early-career teacher who
does not teach the same content area as the mentor.
Portes’ (1988) observation that networks have the potential to diminish the autonomy of
individuals offers one possible explanation for the negative effect of content area matches
between mentors and mentees on mentees’ commitment levels. It is not difficult to imagine that
mentors’ years of experience in classrooms have given them beliefs that are different from their
mentees for what is possible with curriculum and students in their content area. Possibly, then,
this difference in beliefs between the mentor and mentee for how the curriculum can and should
be enacted and for what students can be expected to do results in early-career teachers feeling a
diminished sense of autonomy. And possibly beliefs about curriculum do not get discussed in the
same way or to the same degree for mentors and mentees who do not teach the same content
area.
Based on the early-career teacher surveys, mentors and mentees who teach the same
content area discussed curriculum an average of 4.34 days per month while mentors and mentees
who did not teach the same content area discussed curriculum an average of 0.96 days per
month. It seems reasonable that talking about curriculum less frequently would have less of an
effect on the early-career teachers’ perspectives on curriculum than would talking more
frequently. And if those conversations surfaced differences in perspectives, then it would seem
plausible that fewer conversations would be fewer opportunities for early-career teachers to
receive pressure to conform. Further, it seems likely that mentor-mentee pairs who do not share
subject areas would not discuss the enactment of curriculum with the same sophistication as

109

would mentee-mentor pairs who do teach the same subject area. Perhaps less sophisticated
discussions result in less pressure to conform.
Last, the null findings with respect to the effects of the interaction terms between the
frequency with which early-career teachers communicate with their mentors and colleagues and
the commitment levels of the mentors and colleagues point to a need for further theoretical work
on how characteristics of the networks might influence the levels of commitment to the schools
of the early-career teachers. It could be that early-career teachers do not perceive the levels of
commitment of their mentors and colleagues. It could also be that the commitment levels of the
mentors and colleagues do not impact the nature and frequency with which they interact with
early-career teachers. The finding in this study that the content area that the mentor teaches does
impact the relationships lends support to the need to conceptualize other characteristics besides
commitment that could affect network interactions.
Mentors, colleagues, and professional commitment. The null findings for the effects of
the network variables on the spring levels of professional commitment of the early-career
teachers are what I expected at the outset of the study. Mentors and colleagues are schoolspecific resources for the early-career teachers. I speculate that the experiences of the earlycareer teachers with these resources would not generalize to teachers’ perceived experiences in
all schools or districts/dioceses.
Teacher preparation and commitment. Across models in this study, I found the indicator
for whether an early-career teacher received a bachelor’s degree as part of a teacher preparation
program to be a statistically significant predictor of the spring level of school commitment of the
early-career teachers. I did not evaluate the effects of the indicator in the models for professional
commitment due to the limited amount of power to detect effects in those models. I hypothesize

110

teachers who participate in a teacher preparation program as part of their bachelor’s degree
program obtain a more realistic expectation for what their experiences in schools will be like
relative to teachers who do not participate in a teacher preparation program as part of their
bachelor’s degree. I suspect that teacher candidates in college and university teacher preparation
programs log more hours in classrooms and as practicing being a teacher than do entrants to the
profession who do not participate in a preparation program as part of their bachelor’s degree. I
imagine that these additional hours help prepare individuals for the realities of classroom
experiences.
Longitudinal analyses. One of the central contributions that this study makes is testing
the effects of within-school networks using longitudinal data. Each model in this paper, with the
exception of the fully unconditional models, controlled for the prior measures of the outcome
variables. That is, in the models that evaluated the effects of various predictors on early-career
teachers’ levels of commitment to their schools in the spring, I included a measure of the earlycareer teachers’ levels of commitment to their schools in the fall in the model as well. The same
can be said for the measure of professional commitment.
Controlling for the prior allows directionality to be assigned to relationships among
variables because those relationships have a temporal order. For instance, in between obtaining
the measures of commitment in the fall (baseline measure) and the spring (outcome measure),
the early-career teachers interacted with their networks for approximately six months. In the
models, I controlled for the measure of the outcome in the fall, which then allowed me to assess
the measure of the outcome in the spring, knowing that the predictors preceded the outcome in
the spring. Consequently, I can say that the predictors led to the variance in the outcome that is
being explained by each model.

111

Further evidence of the directionality of the effects is provided by the lack of any
statistically significant relationships between predictors and outcome (with the exception of the
covariate that measures the level of commitment) when the models are run in the reverse
direction. For example, when I exchange the measures of commitment in Equation 3.2, thereby
regressing the fall measure of commitment on the network predictors, the spring level of
commitment, and the indicator of receiving a bachelor’s degree as part of a teacher preparation
program, the p-value of each of the predictors (except the spring level of commitment) is greater
than 0.500. Again, the insignificant results of the reverse models provide evidence of the
directionality of the relationships.
Limitations of the Study
While this study makes important contributions to the literature on within-school
networks and early-career teachers’ commitment levels, it has important limitations (beyond the
low reliability of the commitment-to-school outcome measure). Campbell and Stanley (1966)
provided a thorough review of the possible threats to internal and external validities in research
studies. According to their review, my study would be classified as a “pre-experimental design,”
and specifically, as a “one-group pretest-posttest design” (p. 8). The potential rival hypotheses
that the authors identified as being threats to the internal validity of a one-group pretest-posttest
design include history, maturation, testing, instrumentation, regression, and the interaction of
selection and maturation. For threats to the external validity of the design, the authors identified
the interaction of testing and the predictors, the interaction of selection and the predictors, and
reactive arrangements. In this section, I define and address their validity concerns in the context
of my study.

112

Campbell and Stanley (1966) define history as a threat in terms of some extraneous
situation that occurred between the pretest and posttest that could have accounted for the effects
rather than the predictors included in the model. Because the teachers in my study worked in 44
different schools across three different states, any extraneous event would have had to occur at a
fairly high level of awareness within society. I do not recall such an event in the winter of 20072008. That is not to say with certainty that there was not such an event, but it is to say that I am
fairly comfortable assuming that some extraneous situation did not systematically affect the
results of this study. (Note that I would likely not have disregarded the possibility of an
extraneous situation had my study been conducted in the winter of 2010-2011 due to the
widespread attacks on collective bargaining rights of public employees that occurred in many of
the states in the Midwest.)
Maturation is the process of participants undergoing natural development over the course
of the study and the threat is that this development could be what is causing the changes being
detected in the study (Campbell & Stanley, 1966). It seems plausible that the commitment levels
of early-career teachers could vary across different points in the school year. It could be the case
that teachers are more optimistic about their work, and consequently more committed, in the
beginning of the year than at the end. And it could be that this change in commitment is more a
consequence of spending time in classrooms and less a consequence of interacting with others in
their school communities. I did not control for this potential confounding effect; in future studies,
it would likely be wise to collect measures of commitment at more than two time points within
the same school year and also across multiple school years.
It is unlikely that the effects of testing, reactive arrangements, instrumentation, and
regression affected the findings of my study. The measure of school commitment consisted of

113

three items that likely could comfortably be asked in casual conversations between any two
individuals. The same is likely true of the one item that was used as the measure of professional
commitment. It seems improbable that the items themselves would cause changes in the
perceptions of the early-career teachers, nor does it seem probable that any of the responses to
the items appear to be any more or less socially acceptable. Last, I did not select the sample
based on their extreme scores, and so regression to the mean is not likely a concern in this study.
To this point, I have addressed the main effects of extraneous variables that could
threaten the internal validity of my study. Campbell and Stanley (1966) also identify the
possibility of threats from the interaction of extraneous variables themselves, which would affect
the internal validity of the study, as well as the interaction of the extraneous variables and the
predictors of interest; the latter interactions would affect the external validity of the study. In
terms of the interactions among extraneous variables, I am skeptical of the possibility that
interactions of extraneous variables are confounding the findings of my study. My skepticism
arises because I have essentially ruled out the effects of confounding extraneous variables with
the exception of maturation. For example, if history takes the value of zero because it is a nonfactor in my study, then the non-zero confounding effect of maturation times the zero effect of
history yields a zero effect for the interaction between the two extraneous variables.
While I do not think that interactions between pairs of extraneous variables represent
credible threats to the internal validity of my study, I do believe that the interactions between
certain extraneous variables and the predictors of interest threaten the external validity of my
study. In particular, I cannot rule out the possibility that the findings that I presented in this paper
are an artifact of the interaction between the selection of the sample and the various predictors
that I evaluated in this study. The consequence of this is that this study has value for advancing

114

theoretical ideas on the relationships among commitment levels, networks, and the preparation of
teachers, but it does not offer guidance to policymakers. In order to offer guidance to
policymakers, a study must have external validity; that is, it must be representative of the
population to which the policy would apply. My study is not representative of any state or
federal contexts, but my findings do justify the future investment of additional resources to
further evaluate the effects of mentors and colleagues on early-career teachers’ commitment
levels. Future studies with representative samples would likely lead to the production of
guidance for policymakers.

115

FOOTNOTES

116

FOOTNOTES
1

Changing the survey response mode introduces the possibility of an instrument effect on

participants’ responses; that is, by changing modes, measurement error becomes a concern (de
Leeuw, 2005). As well, these instrument effects are particularly problematic on scalar items
(Dillman, 2007). For instance, mode effects are not likely to be found when participants respond
to a question asking for their gender; however, in questions asking them to provide their level of
agreement or disagreement on culturally-sensitive positions, a mode effect is a real possibility.
These mode effects on scalar items have been found to be a concern when the mode changes
result in participants receiving different sensory stimulations (i.e., aural versus visual). There is
less evidence that participants respond differently between modes with similar sensory
stimulations, particularly when a uni-mode design has been adopted for the different survey
formats; an example would be one visual mode versus another visual mode (Dillman, 2007). In a
uni-mode design, one attempts to present the same stimulus to participants completing the
questionnaire in different modes. For example, one would construct a paper survey and an online
survey with similar visual and graphical stimuli. I adopted this approach to the survey design.
2

The overall response rate of 48 percent for early-career teachers who completed both

surveys is considered modest by standards of social science research. However, for a study with
a sample that is not representative of a larger population, the response rate is more of a concern
to ensure sufficient power in analyses and less of a concern to ensure that the sample is
representative of the population from which it was drawn. This is, in fact, because the latter
cannot be ensured, or known, in the absence of random sampling. Even in studies in which the
sample is selected at random from the population, the response rate is more of a subjective
indicator of the effectiveness of the data collection methods and less of an objective measure of

117

the extent to which the estimated findings from the study can be thought to represent the true
circumstances of the population. If researchers or others want to know the extent to which the
study participants represent the population of interest, then confidence intervals can be calculated
for the estimates. Confidence intervals are only meaningful in cases of random sampling.
3

The coefficient in the output of a logistic regression analysis is defined as the predicted

log of the odds (i.e., predicted log-odds) that the outcome is equal to one conditional on the other
predictors in the model. The predicted log of the odds can be converted to predicted odds
through the exponentiation of the coefficient. As well, a predicted probability can be calculated
from the predicted log of the odds by taking the inverse of one plus the exponentiation of the
opposite of the coefficient (Raudenbush & Bryk, 2002).
4
5
6

See footnote 1.
See footnote 2.
In this section, I report descriptive characteristics for the 232 mentors and colleagues

who completed one survey during the 2007-2008 school year but not the 16 early-career teachers
who completed two surveys during that school year (the characteristics of the 16 teachers who
completed two surveys get represented in the reporting on the early-career teacher sample). In
the regression analyses, I include the 16 early-career teachers in the calculations relying on
information from mentors and colleagues.
7

The lower bound of each range is based on the responses of the individuals who

completed a survey. The upper bound of each range is the maximum value that would be
possible if all of the individuals who did not return a survey satisfied the condition being
measured.

118

APPENDICES

119

APPENDIX A

EARLY-CAREER TEACHER SURVEY #1, FALL 2007
Q1:

In 2007-2008, are you teaching in grades 9 through 12 as the regular classroom teacher?
[yes; no]

Q2:

Please indicate whether you are in your first or second year of full-time teaching in 20072008 (do not count long-term or short-term sub positions as years of teaching, but do
count full-time experience in other districts, dioceses, or schools). [first year of teaching;
second year of teaching; not in first or second year of teaching]

Q3:

What grade level(s) and subject(s) do you teach? [open-ended response]

Q4:

Do you have a mentor who was assigned to you by your school or district/diocese? [yes;
no]

Q5:

How was your mentor assigned to you? [my mentor was assigned to me by my
district/diocese; my mentor was assigned to me by my school; I am not sure how my
mentor was assigned to me; other]

Q6:

Please provide the following information about your mentor: name; position/title. [openended response]

Q7:

When was your mentor assigned to you? [open-ended response]

Q8:

Is your mentor currently a… full-time teacher in your school?; part-time teacher in your
school?; full-time mentor who has been released from teaching?; school-based
administrator?; district/diocesan office administrator?; someone from a university-based
teacher education program? [yes; no]

Q9:

What grade level(s) and subject(s) does you mentor currently teach? [open-ended
response]

Q10: How often do you engage in professional interactions (e.g., interactions about curriculum,
instruction, students, school policies, parents, etc.) with your mentor? [never; less than
once a month; one to three times per month; one to two times per week; three to four
times per week; every day]
Q11: Do those interactions usually take place… before school?; after school?; during lunch?;
during planning period?; other? [yes; no]

120

Q12: For September and October, how often did you address each of the following with your
mentor?: curriculum (main topics and texts to be taught, including scope and sequence)?;
teaching strategies?; student discipline / managing student behavior?; classroom
assessments?; paperwork?; standardized testing?; psychological support (assistance in
dealing with work-related stresses and pressures)?; communicating with parents?; other?
[never; less than once a month; one to three times per month; one to two times per week;
three to four times per week; every day]
Q14: In September and October, did you engage in professional interactions (e.g., interactions
about curriculum, instruction, students, school policies, parents, etc.) with one or more
colleagues who work at your school? [yes; no]
Q15- How often do you engage in professional interactions (e.g., interactions about curriculum,
Q20: instruction, students, school policies, parents, etc.) with one or more colleagues who work
at your school? [open-ended response to name colleagues plus frequency response for
each named colleague: less than once a month; one to three times a month; one to two
times per week; three to four times per week; every day]
Q21: For September and October, how often did you address each of the following with one or
more of the school-based colleagues whom you listed in Questions 15-20?: curriculum
(main topics and texts to be taught, including scope and sequence)?; teaching strategies?;
student discipline / managing student behavior?; classroom assessments?; paperwork?;
standardized testing?; psychological support (assistance in dealing with work-related
stresses and pressures)?; communicating with parents?; other? [never; less than once a
month; one to three times a month; one to two times per week; three to four times per
week; every day]
Q22: If you would like to make any additional comments in response to the questions about
your mentor and colleagues, please use this space. [open-ended response]
Q23: Please indicate your level of agreement or disagreement with each of the statements:
“when necessary, I have access to my mentor and/or colleagues for info and advice
about…” curriculum (main topics and texts to be taught, including scope and sequence)?;
teaching strategies?; student discipline / managing student behavior?; classroom
assessments?; paperwork?; standardized testing?; psychological support (assistance in
dealing with work-related stresses and pressures)?; communicating with parents?; other?
[strongly disagree; disagree; agree; strongly agree; not applicable]
Q24: How important is the professional support that you receive from your mentor and schoolbased colleagues? Mentor?; school-based colleagues? [not at all important; somewhat
important; very important; extremely important; not applicable]

121

Q25: For September and October, how many hours did you spend addressing each of the
following in professional development activities?: curriculum (main topics and texts to be
taught, including scope and sequence)?; teaching strategies?; student discipline /
managing student behavior?; classroom assessments?; paperwork?; standardized testing?;
psychological support (assistance in dealing with work-related stresses and pressures)?;
communicating with parents?; other? [none; one to two hours; three to four hours; six to
eight hours; two days; three or more days]
Q26: In September and October, did you discuss school-related matters (e.g., curriculum,
instruction, students, school policies, parents, etc.) with individuals who do not work at
your school (e.g., friends, relatives, etc.)? [yes; no]
Q27: Please list up to four colleagues, friends, relatives, or others outside of your school with
whom you engage in professional interactions (e.g., interactions about curriculum,
instruction, students, school policies, parents, etc.). [open-ended response]
Q28: How important is the professional support that you receive from the people whom you
listed in Question 27? [not at all important; somewhat important; very important;
extremely important; not applicable]
Q29: Please indicate the extent to which each statement describes you as a teacher: I set a welldefined task for each student; I write detailed lesson plans; I know what each student is
doing; I set a good example for my students; I try to instill a common set of values in my
students; my lessons are based on an explicit set of values; students see me as someone
they can look up to; order and discipline come first in my classroom; I require a quite
classroom. [not at all like me; a little like me; somewhat like me; a lot like me; not
applicable]
Q30: Please indicate the extent to which each statement describes you as a teacher: my students
must do more than learn basic facts; I encourage students to express opinions different
from my own; I engage students in question and answer; I give my students a chance to
discuss issues among themselves; I teach students how to learn; the class material I
choose stimulates students to reflect on their values; my assignments require students to
gather information on their own; I give students the opportunity to explore subject matter
on their own. [not at all like me; a little like me; somewhat like me; a lot like me; not
applicable]
Q31: Please indicate the extent to which each statement describes you as a teacher and your
relations with students: students talk to me about their friendships; students talk to me
about what they do outside school; students see me as a friend; I know a great deal about
students’ families; I make exceptions for special cases of misbehavior; I refuse to
negotiate with students about homework assignments; students rarely see me break a
school rule; I refuse to negotiate with students about grades. [not at all like me; a little
like me; somewhat like me; a lot like me; not applicable]

122

Q32: Please indicate your level of agreement or disagreement with each of the following
statements: I know how to teach students from diverse racial/ethnic backgrounds; I know
how to teach students from diverse linguistic backgrounds (i.e., English language
learners); I know how to teach students from diverse socio-economic backgrounds; I
know how to teach students with special needs (i.e., special needs students or special
education students); I know how to teach students with diverse abilities (i.e., gifted
students, low achieving students, etc.). [strongly disagree; disagree; agree; strongly agree]
Q33: If you would like to make any additional comments in response to the questions about
your teaching practices or interactions with students, please use this space. [open-ended
response]
Q34: Please indicate your level of agreement or disagreement with each of the following
statements: I wouldn’t want to work in any other school; I would recommend this school
to parents seeking a place for their child; I usually look forward to each working day at
this school; I feel loyal to this school. [strongly disagree; disagree; agree; strongly agree;
not sure]
Q35: Please indicate your level of agreement or disagreement with each of the following
statements: it’s okay in this school to discuss feelings, frustrations, and worries with other
teachers; teachers in this school trust each other; teachers in this school respect other
teachers who take the lead in school improvement efforts; teachers in this school respect
those colleagues who are expert in their craft. [strongly disagree; disagree; agree; strongly
agree; not sure]
Q36: Please indicate your level of agreement or disagreement with each of the following
statements: my approach to teaching fits with this school; my professional interests are
the same as those of other teachers in this school; I identify with other teachers in this
school; my professional goals are the same as those of other teachers in this school; I
matter to other teachers in this school; other teachers in this school matter to me.
[strongly disagree; disagree; agree; strongly agree; not sure]
Q37: Please indicate the proportion of teachers in this school who do the following: help
maintain discipline in the entire school, not just their classrooms; take responsibility for
helping one another do well; take responsibility for improving the overall quality of
teaching in the school; feel responsible for helping students develop self-control; set high
expectations for academic work; feel responsible for ensuring that all students learn.
[none; less than half; about half; most; all]
Q39: Please indicate your level of agreement or disagreement with each of the following
statements: it’s okay in this department to discuss feelings, frustrations, and worries with
other teachers; teachers in this department trust each other; teachers in this department
respect other teachers who take the lead in department improvement efforts; teachers in
this department respect those colleagues who are expert in their craft. [strongly disagree;
disagree; agree; strongly agree; not sure]

123

Q40: Please indicate your level of agreement or disagreement with each of the following
statements: my approach to teaching fits with this department; my professional interests
are the same as those of other teachers in this department; I identify with other teachers in
this department; my professional goals are the same as those of other teachers in this
department; I matter to other teachers in this department; other teachers in this
department matter to me. [strongly disagree; disagree; agree; strongly agree; not sure]
Q41: If you would like to make any additional comments in response to the questions about
your perceptions of your school or department, please use this space. [open-ended
response]
Q42: For the months of September and October, please indicate your level of agreement or
disagreement with each of the following statements: I am teaching with adequate
resources and materials to do my job properly; Administrative duties/paperwork do not
interfere with my teaching; my workload is manageable; I feel energetic and enthusiastic
about teaching; I have time and energy for friends and family. [strongly disagree;
disagree; agree; strongly agree; not sure]
Q43: Please indicate your level of agreement or disagreement with each of the following
statements: “during my hiring process (and the time before I actually started working at
this school), I acquired an accurate picture of…” the teachers at this school; the students
at this school; the principal’s leadership style; the curriculum that I would be responsible
for teaching; my teaching assignment; the school support system; the amount of personal
control I would have in my classroom; the opportunities that I would have to shape the
environment of the school; the mission and philosophy of the school. [strongly disagree;
disagree; agree; strongly agree; not sure]
Q44: If you would like to make any additional comments in response to the question about
your hiring experience, please use this space. [open-ended response]
Q45: If you go back to your college days and start over again, would you become a teacher or
not? [certainly would become a teacher; probably would become a teacher; chances about
even for and against; probably would not become a teacher; certainly would not become a
teacher]
Q46: How long do you plan to remain in teaching? [as long as I am able; until I am eligible for
retirement; will probably continue until something better comes along; definitely plan to
leave teaching as soon as I can; undecided at this time]
Q47: Please indicate your level of agreement or disagreement with each of the following
statements: I would prefer to continue teaching in this school next year; I could see
myself teaching in this school in five years; I would prefer to continue teaching the
courses I am presently teaching next year; I could see myself teaching the courses I am
presently teaching in five years; I would prefer to continue teaching in this
district/diocese next year; I could see myself teaching in this district/diocese in five years.
[strongly disagree; disagree; agree; strongly agree; not sure]

124

Q48: Which of the following best describes your career plans? [I plan to pursue another
education-related career at some point; I plan to pursue another career outside of the field
of education at some point; I plan to have children and stop teaching at some point; I am
going to see if I like teaching before I make plans; other]
Q49: So that we can record your participation in this research study, please provide your full
name in the box below. Please recall that your name, the name of your school, and the
name of your district/diocese will not be used in any reports or publications. [open-ended
response]
Q50: If you would like to make any additional comments in response to any of the questions in
this survey, please use this space. [open-ended response]
Q51: Please fill-in today’s date. [open-ended response]

125

APPENDIX B

EARLY-CAREER TEACHER SURVEY #2, SPRING 2008
Q1:

What grade level(s) and subject(s) did you teach in April and May? [open-ended
response]

Q2:

In April and May, did you have a mentor who was assigned to you by your school or
district/diocese? [yes; no]

Q3:

Was the assigned mentor that you had in April/May the only assigned mentor that you
had during the 2007-2008 school year? [yes; no]

Q4:

Please provide the following information about your mentors: name; position/title;
months during which (s)he was your mentor. [open-ended response]

Q5:

Please briefly describe the reason(s) for the change in mentors this year. [open-ended
response]

Q6:

In April and May did you assigned mentor work as a… full-time teacher in your school?;
part-time teacher in your school?; full-time mentor who has been released from
teaching?; school-based administrator?; district/diocesan office administrator?; someone
from a university-based teacher education program? [yes; no]

Q7:

In April and May, what grade level(s) and subject(s) did your assigned mentor teach?
[open-ended response]

Q8:

In April and May, how often did you engage in professional interactions (e.g.,
interactions about curriculum, instruction, students, school policies, parents, etc.) with
your mentor? [never; less than once a month; one to three times per month; one to two
times per week; three to four times per week; every day]

Q9:

Did those interactions usually take place… before school?; after school?; during lunch?;
during planning period?; other? [yes; no]

Q10: For April and May, how often did you address each of the following with your mentor?:
curriculum (main topics and texts to be taught, including scope and sequence)?; teaching
strategies?; student discipline / managing student behavior?; classroom assessments?;
paperwork?; standardized testing?; psychological support (assistance in dealing with
work-related stresses and pressures)?; communicating with parents?; other? [never; less
than once a month; one to three times per month; one to two times per week; three to four
times per week; every day]

126

Q11: If you would like to make any additional comments in response to the questions about
your mentor(s), please use this space. [open-ended response]
Q12: In April and May, did you engage in professional interactions (e.g., interactions about
curriculum, instruction, students, school policies, parents, etc.) with one or more
colleagues who work at your school? [yes; no]
Q13- How often do you engage in professional interactions (e.g., interactions about curriculum,
Q18: instruction, students, school policies, parents, etc.) with one or more colleagues who work
at your school? [open-ended response to name colleagues plus frequency response for
each named colleague: less than once a month; one to three times a month; one to two
times per week; three to four times per week; every day]
Q19: For April and May, how often did you address each of the following with one or more of
the school-based colleagues whom you listed in Questions 15-20?: curriculum (main
topics and texts to be taught, including scope and sequence)?; teaching strategies?;
student discipline / managing student behavior?; classroom assessments?; paperwork?;
standardized testing?; psychological support (assistance in dealing with work-related
stresses and pressures)?; communicating with parents?; other? [never; less than once a
month; one to three times a month; one to two times per week; three to four times per
week; every day]
Q20: If you would like to make any additional comments in response to the questions about
your colleagues, please use this space. [open-ended response]
Q21: Please indicate your level of agreement or disagreement with each of the statements:
“when necessary, I have access to my mentor and/or colleagues for info and advice
about…” curriculum (main topics and texts to be taught, including scope and sequence)?;
teaching strategies?; student discipline / managing student behavior?; classroom
assessments?; paperwork?; standardized testing?; psychological support (assistance in
dealing with work-related stresses and pressures)?; communicating with parents?; other?
[strongly disagree; disagree; agree; strongly agree; not applicable]
Q22: How important is the professional support that you receive from your mentor and schoolbased colleagues? Mentor?; school-based colleagues? [not at all important; somewhat
important; very important; extremely important; not applicable]
Q23: In April and May, how many hours did you spend addressing each of the following in
professional development activities?: curriculum (main topics and texts to be taught,
including scope and sequence)?; teaching strategies?; student discipline / managing
student behavior?; classroom assessments?; paperwork?; standardized testing?;
psychological support (assistance in dealing with work-related stresses and pressures)?;
communicating with parents?; other? [none; one to two hours; three to four hours; six to
eight hours; two days; three or more days]

127

Q24: In April and May, did you discuss school-related matters (e.g., curriculum, instruction,
students, school policies, parents, etc.) with individuals who do not work at your school
(e.g., friends, relatives, etc.)? [yes; no]
Q25: Please list up to four colleagues, friends, relatives, or others outside of your school with
whom you engage in professional interactions (e.g., interactions about curriculum,
instruction, students, school policies, parents, etc.). [open-ended response]
Q26: How important is the professional support that you receive from the people whom you
listed in Question 25? [not at all important; somewhat important; very important;
extremely important; not applicable]
Q27: Please indicate the extent to which each statement describes you as a teacher: I set a welldefined task for each student; I write detailed lesson plans; I know what each student is
doing; I set a good example for my students; I try to instill a common set of values in my
students; my lessons are based on an explicit set of values; students see me as someone
they can look up to; order and discipline come first in my classroom; I require a quite
classroom. [not at all like me; a little like me; somewhat like me; a lot like me; not
applicable]
Q28: Please indicate the extent to which each statement describes you as a teacher: my students
must do more than learn basic facts; I encourage students to express opinions different
from my own; I engage students in question and answer; I give my students a chance to
discuss issues among themselves; I teach students how to learn; the class material I
choose stimulates students to reflect on their values; my assignments require students to
gather information on their own; I give students the opportunity to explore subject matter
on their own. [not at all like me; a little like me; somewhat like me; a lot like me; not
applicable]
Q29: Please indicate the extent to which each statement describes you as a teacher and your
relations with students: students talk to me about their friendships; students talk to me
about what they do outside school; students see me as a friend; I know a great deal about
students’ families; I make exceptions for special cases of misbehavior; I refuse to
negotiate with students about homework assignments; students rarely see me break a
school rule; I refuse to negotiate with students about grades. [not at all like me; a little
like me; somewhat like me; a lot like me; not applicable]
Q30: Please indicate your level of agreement or disagreement with each of the following
statements: I know how to teach students from diverse racial/ethnic backgrounds; I know
how to teach students from diverse linguistic backgrounds (i.e., English language
learners); I know how to teach students from diverse socio-economic backgrounds; I
know how to teach students with special needs (i.e., special needs students or special
education students); I know how to teach students with diverse abilities (i.e., gifted
students, low achieving students, etc.). [strongly disagree; disagree; agree; strongly agree]

128

Q31: If you would like to make any additional comments in response to the questions about
your teaching practices or interactions with students, please use this space. [open-ended
response]
Q32: Please indicate your level of agreement or disagreement with each of the following
statements: I wouldn’t want to work in any other school; I would recommend this school
to parents seeking a place for their child; I usually look forward to each working day at
this school; I feel loyal to this school. [strongly disagree; disagree; agree; strongly agree;
not sure]
Q33: Please indicate your level of agreement or disagreement with each of the following
statements: it’s okay in this school to discuss feelings, frustrations, and worries with other
teachers; teachers in this school trust each other; teachers in this school respect other
teachers who take the lead in school improvement efforts; teachers in this school respect
those colleagues who are expert in their craft. [strongly disagree; disagree; agree; strongly
agree; not sure]
Q34: Please indicate your level of agreement or disagreement with each of the following
statements: my approach to teaching fits with this school; my professional interests are
the same as those of other teachers in this school; I identify with other teachers in this
school; my professional goals are the same as those of other teachers in this school; I
matter to other teachers in this school; other teachers in this school matter to me.
[strongly disagree; disagree; agree; strongly agree; not sure]
Q35: Please indicate the proportion of teachers in this school who do the following: help
maintain discipline in the entire school, not just their classrooms; take responsibility for
helping one another do well; take responsibility for improving the overall quality of
teaching in the school; feel responsible for helping students develop self-control; set high
expectations for academic work; feel responsible for ensuring that all students learn.
[none; less than half; about half; most; all]
Q36: If you would like to make any additional comments in response to the questions about
your perceptions of your school, please use this space. [open-ended response]
Q37: Please indicate your level of agreement or disagreement with each of the following
statements: my approach to teaching fits with this department; my professional interests
are the same as those of other teachers in this department; I identify with other teachers in
this department; my professional goals are the same as those of other teachers in this
department; I matter to other teachers in this department; other teachers in this
department matter to me. [strongly disagree; disagree; agree; strongly agree; not sure]

129

Q38: Please indicate your level of agreement or disagreement with each of the following
statements: it’s okay in this department to discuss feelings, frustrations, and worries with
other teachers; teachers in this department trust each other; teachers in this department
respect other teachers who take the lead in department improvement efforts; teachers in
this department respect those colleagues who are expert in their craft. [strongly disagree;
disagree; agree; strongly agree; not sure]
Q39: If you would like to make any additional comments in response to the questions about
your perceptions of your department, please use this space. [open-ended response]
Q40: For the months of April and May, please indicate your level of agreement or
disagreement with each of the following statements: I am teaching with adequate
resources and materials to do my job properly; Administrative duties/paperwork do not
interfere with my teaching; my workload is manageable; I feel energetic and enthusiastic
about teaching; I have time and energy for friends and family. [strongly disagree;
disagree; agree; strongly agree; not sure]
Q41: If you go back to your college days and start over again, would you become a teacher or
not? [certainly would become a teacher; probably would become a teacher; chances about
even for and against; probably would not become a teacher; certainly would not become a
teacher]
Q42: How long do you plan to remain in teaching? [as long as I am able; until I am eligible for
retirement; will probably continue until something better comes along; definitely plan to
leave teaching as soon as I can; undecided at this time]
Q43: Please indicate your level of agreement or disagreement with each of the following
statements: I would prefer to continue teaching in this school next year; I could see
myself teaching in this school in five years; I would prefer to continue teaching the
courses I am presently teaching next year; I could see myself teaching the courses I am
presently teaching in five years; I would prefer to continue teaching in this
district/diocese next year; I could see myself teaching in this district/diocese in five years.
[strongly disagree; disagree; agree; strongly agree; not sure]
Q44: Which of the following best describes your career plans? [I plan to pursue another
education-related career at some point; I plan to pursue another career outside of the field
of education at some point; I plan to have children and stop teaching at some point; I am
going to see if I like teaching before I make plans; other]
Q45: What is your gender? [female; male]
Q46: Are you of Hispanic or Latino origin? [yes; no]
Q47: What is your race? [American Indian or Alaska Native; Asian; Black or AfricanAmerican; Native Hawaiian or Pacific Islander; White; other]

130

Q48: Please provide the following information about the high school from which you
graduated: name of high school; location (city and state). [open-ended response]
Q49: Which of the following best describes your formal preparation for teaching? [an
“alternative” program designed to expedite the transition of non-teachers to a teaching
career (e.g., a state, district/diocesan or university alternative program); a bachelor’s
degree granting program (B.A. or B.S.); a fifth-year program (not leading to a master’s
degree); a master’s degree granting program (M.A., M.S., M.Ed., M.A.T.); individual
courses (not part of a program leading to a degree); I did not enter teaching through any
of the routes described above; other]
Q50: Did your preparation for teaching include the following?: coursework in how to select
and adapt instructional material; coursework in learning theory or psychology appropriate
to the age of students you teach; your observation of other classroom teaching; formal
feedback on your teaching. [yes; no]
Q51: Which of the following describes the teaching certificate(s) you currently hold?:
certification from the National Board for Professional Teaching Standards (NBPTS);
regular or standard state certificate; probationary certificate (issued after satisfying all
requirements except the completion of a probationary period); provisional or other type
of certificate given to persons who are still participating in what the state calls an
“alternative certification program”; temporary certificate (requires some additional
college coursework, student teaching, and/or passage of a test before regular certification
can be obtained); waiver or emergency certificate (issued to persons with insufficient
teacher preparation who must complete a regular certification program in order to
continue teaching); regular or full certification by an accrediting or certifying body other
than the state; I do not have any of the above certifications in this stats; other]
Q52: Please identify the areas in which you are certified (e.g., secondary math). [open-ended
response]
Q53: If you would like to make any additional comments in response to the questions about
your background or preparation for teaching, please use this space. [open-ended
response]
Q54: Did you apply and/or entertain teaching position offers from more than one school? [yes,
I applied to other schools/districts but only received an offer from my present school; yes,
I received job offers from (an)other school(s); no, I researched other schools/districts but
decided to formally apply just to my present school; no, I only considered my present
school]
Q55: Relative to the first day of school (defined as the first full day of classes for students),
when were you hired at your present school? [more than 30 days before the start of
school; between 30 and 10 days before the start of school; within 10 days of the start of
school; on or after the first day of school]

131

Q56: Relative to the first day of school, when did you receive your teaching assignment
(defined as the courses and grade levels for which you would be responsible)? [more than
30 days before the start of school; between 30 and 10 days before the start of school;
within 10 days of the start of school; on or after the first day of school]
Q57: Relative to the first day of school, when did you receive curricular materials (defined as
curriculum guides, textbooks, or other teachers’ lesson plans, etc.) for your teaching
assignment? [more than 30 days before the start of school; between 30 and 10 days before
the start of school; within 10 days of the start of school; on or after the first day of school;
never received; not applicable (there are not defined curricula for my courses]
Q58: If you would like to make any additional comments in response to the questions about
your hiring experience, please use this space. [open-ended response]
Q59: Are you a member of a teachers’ union or an employee association similar to a union?
[yes, my only option was to belong; yes, I chose to belong; no, I chose not to belong; no I
had no opportunity to belong (i.e., there is no union/association at my school]
Q60: How frequently do you attend union meetings? [I attend every meeting; I attend more
than half of the meetings; I attend fewer than half of the meetings; I never attend
meetings; not applicable (i.e., there is no union/association at my school]
Q61: How would you rate relations between teachers and administrators at your school?
[excellent; good; fair; poor; not sure]
Q62: If you would like to make any additional comments in response to the questions about
union/association activities, please use this space. [open-ended response]
Q63: Please describe your post-secondary degrees in the chart below: associate’s, bachelor’s
#1, bachelor’s #2, master’s #1, master’s #2, ed specialist, doctorate, other BY year,
major, minor, college/university. [open-ended response]
Q64: So that we can record your participation in this research study, please provide your full
name in the box below. Please recall that your name, the name of your school, and the
name of your district/diocese will not be used in any reports or publications. [open-ended
response]
Q65: We will send you a gift card as a token of our appreciation for completing this survey.
We will likely process gift cards in June; if you would like us to send your gift card to an
address other than your school, please provide that address below. [open-ended response]
Q66: If you would like to make any additional comments in response to any of the questions in
this survey, please use this space. [open-ended response]
Q67: Please fill-in today’s date. [open-ended response]

132

APPENDIX C

APPENDIX C: MENTOR/COLLEAGUE SURVEY, SPRING 2008
Q1:

For the 2007-2008 school year, are you currently a… full-time teacher in your school?;
part-time teacher in your school?; full-time mentor who has been released from teaching?;
school-based administrator?; district/diocesan office administrator?; someone from a
university-based teacher education program?; other? [yes; no]

Q2:

Please indicate how many times you have served as a formal mentor teacher (to a fulltime new teacher) or as a formal cooperating teacher (to a student teacher or intern
teacher) prior to the 2007-2008. [open-ended response]

Q3:

What grade level(s) and subject(s) do you currently teach? [open-ended response]

Q4:

What grade level(s) and subject(s) have you taught in the past? Please describe your
previous teaching experiences: these might include work at other schools, work as a
teacher’s aide or paraprofessional, or work as a substitute teacher. [open-ended response]

Q5:

Please indicate the extent to which each statement describes you as a teacher: I set a welldefined task for each student; I write detailed lesson plans; I know what each student is
doing; I set a good example for my students; I try to instill a common set of values in my
students; my lessons are based on an explicit set of values; students see me as someone
they can look up to; order and discipline come first in my classroom; I require a quite
classroom. [not at all like me; a little like me; somewhat like me; a lot like me; not
applicable]

Q6:

Please indicate the extent to which each statement describes you as a teacher: my students
must do more than learn basic facts; I encourage students to express opinions different
from my own; I engage students in question and answer; I give my students a chance to
discuss issues among themselves; I teach students how to learn; the class material I
choose stimulates students to reflect on their values; my assignments require students to
gather information on their own; I give students the opportunity to explore subject matter
on their own. [not at all like me; a little like me; somewhat like me; a lot like me; not
applicable]

Q7:

Please indicate the extent to which each statement describes you as a teacher and your
relations with students: students talk to me about their friendships; students talk to me
about what they do outside school; students see me as a friend; I know a great deal about
students’ families; I make exceptions for special cases of misbehavior; I refuse to
negotiate with students about homework assignments; students rarely see me break a
school rule; I refuse to negotiate with students about grades. [not at all like me; a little
like me; somewhat like me; a lot like me; not applicable]

133

Q8:

Please indicate your level of agreement or disagreement with each of the following
statements: I know how to teach students from diverse racial/ethnic backgrounds; I know
how to teach students from diverse linguistic backgrounds (i.e., English language
learners); I know how to teach students from diverse socio-economic backgrounds; I
know how to teach students with special needs (i.e., special needs students or special
education students); I know how to teach students with diverse abilities (i.e., gifted
students, low achieving students, etc.). [strongly disagree; disagree; agree; strongly agree]

Q9:

If you would like to make any additional comments in response to the questions about
your teaching practices or interactions with students, please use this space. [open-ended
response]

Q10: Please indicate your level of agreement or disagreement with each of the following
statements: I wouldn’t want to work in any other school; I would recommend this school
to parents seeking a place for their child; I usually look forward to each working day at
this school; I feel loyal to this school. [strongly disagree; disagree; agree; strongly agree;
not sure]
Q11: Please indicate your level of agreement or disagreement with each of the following
statements: it’s okay in this school to discuss feelings, frustrations, and worries with other
teachers; teachers in this school trust each other; teachers in this school respect other
teachers who take the lead in school improvement efforts; teachers in this school respect
those colleagues who are expert in their craft. [strongly disagree; disagree; agree; strongly
agree; not sure]
Q12: Please indicate your level of agreement or disagreement with each of the following
statements: my approach to teaching fits with this school; my professional interests are
the same as those of other teachers in this school; I identify with other teachers in this
school; my professional goals are the same as those of other teachers in this school; I
matter to other teachers in this school; other teachers in this school matter to me.
[strongly disagree; disagree; agree; strongly agree; not sure]
Q13: Please indicate the proportion of teachers in this school who do the following: help
maintain discipline in the entire school, not just their classrooms; take responsibility for
helping one another do well; take responsibility for improving the overall quality of
teaching in the school; feel responsible for helping students develop self-control; set high
expectations for academic work; feel responsible for ensuring that all students learn.
[none; less than half; about half; most; all]
Q14: If you would like to make any additional comments in response to the questions about
your perceptions of your school, please use this space. [open-ended response]

134

Q15: Please indicate your level of agreement or disagreement with each of the following
statements: my approach to teaching fits with this department; my professional interests
are the same as those of other teachers in this department; I identify with other teachers in
this department; my professional goals are the same as those of other teachers in this
department; I matter to other teachers in this department; other teachers in this department
matter to me. [strongly disagree; disagree; agree; strongly agree; not sure]
Q16: Please indicate your level of agreement or disagreement with each of the following
statements: it’s okay in this department to discuss feelings, frustrations, and worries with
other teachers; teachers in this department trust each other; teachers in this department
respect other teachers who take the lead in department improvement efforts; teachers in
this department respect those colleagues who are expert in their craft. [strongly disagree;
disagree; agree; strongly agree; not sure]
Q17: If you would like to make any additional comments in response to the questions about
your perceptions of your department, please use this space. [open-ended response]
Q18: Please indicate your level of agreement or disagreement with each of the following
statements: I am teaching with adequate resources and materials to do my job properly;
Administrative duties/paperwork do not interfere with my teaching; my workload is
manageable; I feel energetic and enthusiastic about teaching; I have time and energy for
friends and family. [strongly disagree; disagree; agree; strongly agree; not sure]
Q19: Are you a member of a teachers’ union or an employee association similar to a union?
[yes; no]
Q20: If you would like to make any additional comments in response to the questions about
your work experience, please use this space. [open-ended response]
Q21: If you go back to your college days and start over again, would you become a teacher or
not? [certainly would become a teacher; probably would become a teacher; chances about
even for and against; probably would not become a teacher; certainly would not become a
teacher]
Q22: How long do you plan to remain in teaching? [as long as I am able; until I am eligible for
retirement; will probably continue until something better comes along; definitely plan to
leave teaching as soon as I can; undecided at this time]
Q23: Please indicate your level of agreement or disagreement with each of the following
statements: I would prefer to continue teaching in this school next year; I could see
myself teaching in this school in five years; I would prefer to continue teaching the
courses I am presently teaching next year; I could see myself teaching the courses I am
presently teaching in five years; I would prefer to continue teaching in this
district/diocese next year; I could see myself teaching in this district/diocese in five years.
[strongly disagree; disagree; agree; strongly agree; not sure]

135

Q24: Which of the following best describes your career plans? [I plan to pursue another
education-related career at some point; I plan to pursue another career outside of the field
of education at some point; I plan to have children and stop teaching at some point; I am
going to see if I like teaching before I make plans; other]
Q25: If you would like to make any additional comments in response to the questions about
your career decisions, please use this space. [open-ended response]
Q26: What is your gender? [female; male]
Q27: Are you of Hispanic or Latino origin? [yes; no]
Q28: What is your race? [American Indian or Alaska Native; Asian; Black or AfricanAmerican; Native Hawaiian or Pacific Islander; White; other]
Q29: Which of the following best describes your formal preparation for teaching? [an
“alternative” program designed to expedite the transition of non-teachers to a teaching
career (e.g., a state, district/diocesan or university alternative program); a bachelor’s
degree granting program (B.A. or B.S.); a fifth-year program (not leading to a master’s
degree); a master’s degree granting program (M.A., M.S., M.Ed., M.A.T.); individual
courses (not part of a program leading to a degree); I did not enter teaching through any
of the routes described above; other]
Q30: Did your preparation for teaching include the following?: coursework in how to select
and adapt instructional material; coursework in learning theory or psychology appropriate
to the age of students you teach; your observation of other classroom teaching; formal
feedback on your teaching. [yes; no]
Q31: Which of the following describes the teaching certificate(s) you currently hold?:
certification from the National Board for Professional Teaching Standards (NBPTS);
regular or standard state certificate; probationary certificate (issued after satisfying all
requirements except the completion of a probationary period); provisional or other type of
certificate given to persons who are still participating in what the state calls an
“alternative certification program”; temporary certificate (requires some additional
college coursework, student teaching, and/or passage of a test before regular certification
can be obtained); waiver or emergency certificate (issued to persons with insufficient
teacher preparation who must complete a regular certification program in order to
continue teaching); regular or full certification by an accrediting or certifying body other
than the state; I do not have any of the above certifications in this stats; other]
Q32: Please identify the areas in which you are certified (e.g., secondary math). [open-ended
response]
Q33: If you would like to make any additional comments in response to the questions about
your background or preparation for teaching, please use this space. [open-ended response]

136

Q34: Please describe your post-secondary degrees in the chart below: associate’s, bachelor’s
#1, bachelor’s #2, master’s #1, master’s #2, ed specialist, doctorate, other BY year, major,
minor, college/university. [open-ended response]
Q35: So that we can record your participation in this research study, please provide your full
name in the box below. Please recall that your name, the name of your school, and the
name of your district/diocese will not be used in any reports or publications. [open-ended
response]
Q36: If you would like to make any additional comments in response to any of the questions in
this survey, please use this space. [open-ended response]
Q37: Please fill-in today’s date. [open-ended response]

137

APPENDIX D

SUPPLEMENTAL TABLES FOR CHAPTER TWO
Table D1
Hierarchical Form of Equation 2.1
Level 1
(SCH_COMMIT_SPR)ij

=

β0j + rij,

=

γ00 + u0j.

Level 2
β0j

Table D2
Hierarchical Form of Equation 2.2
Level 1
(SCH_COMMIT_SPR)ij

=

β0j + β1j(SCH_COMMIT_FALL)ij
+ β2j(MET_EXPECTATIONS)ij+ β3j(TEDEG_YN)ij + rij,

Level 2
β0j

=

γ00 + γ01(OVER40SCH_YN)j + u0j,

β1j

=

γ10,

β2j

=

γ20,

β6j

=

γ60.

138

Table D3
Hierarchical Form of Equation 2.3
Level 1

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) % = β0j,
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij
Level 2

!

β0j = γ00 + u0j,

Table D4
Hierarchical Form of Equation 2.4
Level 1

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) % = β0j + β1j(PRO_COMMIT_FALL)ij
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij
+ β2j(MET_EXPECT_B)ij
+ β3j(TEDEG_YN)ij
+ β4j(FIRSTYEAR_YN)ij,

!
Level 2

β0j = γ00 + γ01(OVER1200SCH_YN)j + u0j,
β1j = γ10,
β2j = γ20,
β3j = γ30,
β4j = γ40.

139

APPENDIX E

APPENDIX E: SUPPLEMENTAL TABLES FOR CHAPTER THREE
Table E1
Hierarchical Form of Equation 3.1
Level 1
(SCH_COMMIT_SPR)ij

=

β0j + rij,

=

γ00 + u0j.

Level 2
β0j

140

Table E2
Hierarchical Form of Equation 3.2
Level 1
(SCH_COMMIT_SPR)ij

=

β0j + β1j(SCH_COMMIT_FALL)ij + β2j(MENT_YN)ij
+ β3j(MENT_FREQ)ij + β4j(COLL_YN)ij
+ β5j(COLL_FREQ)ij + β6j(TEDEG_YN)ij + rij,

Level 2
β0j

=

γ00 + γ01(OVER40SCH_YN)j + u0j,

β1j

=

γ10,

β2j

=

γ20 + u2j,

β3j

=

γ30,

β4j

=

γ40,

β5j

=

γ50,

β6j

=

γ60.

141

Table E3
Hierarchical Form of Equation 3.3
Level 1
(SCH_COMMIT_SPR)ij

=

β0j + β1j(SCH_COMMIT_FALL)ij + β2j(MENT_CHAR)ij
+ rij,

Level 2
β0j

=

γ00 + u0j,

β1j

=

γ10,

β2j

=

γ20.

Table E4
Hierarchical Form of Equation 3.4
Level 1
(SCH_COMMIT_SPR)ij

=

β0j + β1j(SCH_COMMIT_FALL)ij + β2j(MENT_CHAR)ij
+ β3j(TEDEG_YN)ij + rij,

Level 2
β0j

=

γ00 + u0j,

β1j

=

γ10,

β2j

=

γ20,

β3j

=

γ30.

142

Table E5
Hierarchical Form of Equation 3.5
Level 1

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) % = β0j,
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij
Level 2

!

β0j = γ00 + u0j,

Table E6
Hierarchical Form of Equation 3.6
Level 1

" Pr ob(PRO _ COMMIT _ SPR _ YN = 1) % = β0j + β1j(PRO_COMMIT_FALL)ij
log$
'
# Pr ob(PRO _ COMMIT _ SPR _ YN = 0) &ij
+ β2j(NETWORK_YN)ij
+ β3j(NETWORK_FREQ)ij,

!

Level 2
β0j = γ00 + u0j,
β1j = γ10,
β2j = γ20,
β3j = γ30.

143

REFERENCES

144

REFERENCES
Becker, H. S. (1960). Notes on the concept of commitment. American Sociological Review,
66(1), 32-42.
Becker, H. S., & Carper, J. (1956). The elements of identification with an occupation. American
Sociological Review, 21(3), 341-348.
Bidwell, C. E. (2000). School as context and construction: A social psychological approach to
the study of schooling. In M. T. Hallinan (Ed.), Handbook of sociology in education (pp.
15-36). New York: Kluwer Academic Publishers.
Bidwell, C. E., Frank, K. A., & Quiroz, P. A. (1997). Teacher types, workplace controls, and the
organization of schools. Sociology of Education, 70(4), 285-307.
Bidwell, C. E., & Yasumoto, J. Y. (1999). The collegial focus: Teaching fields, collegial
relationships, and instructional practice in American high schools. Sociology of
Education, 72(4), 234-256.
Bond, T. G., & Fox, C. M. (2007). Applying the Rasch model: Fundamental measurement in the
human sciences (2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates.
Bourdieu, P. (1986). The forms of capital (R. Nice, Trans.). In J. G. Richardson (Ed.), Handbook
of theory and research for the sociology of education (pp. 242-255). Westport, CT:
Greenwood.
Breaugh, J. A. (1983). Realistic job previews: A critical appraisal and future research directions.
Academy of Management Review, 8(4), 612-619.
Brewer, D. J. (1996). Career paths and quit decisions: Evidence from teaching. Journal of Labor
Economics, 14(2), 313-339.
Bruning, R. H., Schraw, G. J., & Ronning, R. R. (1999). Cognitive psychology and instruction
(Third ed.). Upper Saddle River, NJ: Prentice-Hall Inc.
Bryk, A. S., & Schneider, B. (2002). Trust in schools: A core resource for improvement. New
York: Russell Sage Foundation.
Burt, R. S. (2001). Structural holes versus network closure as social capital. In N. Lin, K. S.
Cook & R. S. Burt (Eds.), Social capital: Theory and research.
Campbell, D. T., & Stanley, J. C. (1966). Experimental and quasi-experimental designs for
research. Boston: Houghton-Mifflin.
Chan, W.-Y., Lau, S., Nie, Y., Lim, S., & Hogan, D. (2008). Organizational and personal
predictors of teacher commitment: The mediating role of teacher efficacy and
identification with school. American Educational Research Journal, 45(3), 597-639.

145

Chatman, J. A. (1989). Improving interactional organizational research: A model of personorganization fit. The Acamedy of Management Review, 14(3), 333-349.
Cliff, N. (1996). Ordinal methods for behavior data analysis. Mahwah, NJ: Lawrence Erlbaum
Associates.
Coburn, C. E. (2003). Rethinking scale: Moving beyond numbers to deep lasting change.
Educational Researcher, 32(2), 3-12.
Coleman, J. S. (1961). Adolescent society. New York: Free Press.
Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal of
Sociology, 94(Supplement 1988), S95-S120.
de Leeuw, E. D. (2005). To mix or not to mix data collection modes in surveys. Journal of
Official Statistics, 21(2), 233-255.
Dillman, D. A. (2007). Mail and internet surveys: The tailored design method. Hoboken, NJ:
John Wiley & Sons, Inc.
Ebmeier, H. (2003). How supervision influences teacher efficacy and commitment: An
investigation of a path model. Journal of Curriculum and Supervision, 18(2), 110-141.
Education Week. (2010). Fresh course, swift current: Momentum and challenges in the new
surge toward common standards. In Quality Counts 2010. Washington, DC: Editorial
Projects in Education.
Feiman-Nemser, S. (2001). From preparation to practice: Designing a continuum to strengthen
and sustain teaching. Teachers College Record, 103(6), 1013-1055.
Firestone, W. A., & Rosenblum, S. (1988). Building commitment in urban high schools.
Educational Evaluation and Policy Analysis, 10(4), 285-299.
Frank, K. A. (1995). Identifying cohesive subgroups. Social Networks, 17, 27-56.
Frank, K. A. (1996). Mapping interactions within and between cohesive subgroups. Social
Networks, 18, 93-119.
Frank, K. A. (1998). Quantitative methods for studying social context in multilevels and through
interpersonal relations. Review of Research in Education, 23, 171-216.
Frank, K. A., Zhao, Y., & Borman, K. (2004). Social capital and the diffusion of innovations
within organizations: The case of computer technology in schools. Sociology of
Education, 77(2), 148-171.
Garson, G. D. (2011). Logistic regression. Statnotes: Topics in Multivariate Analysis. Retrieved
06/06/2011, from http://faculty.chass.ncsu.edu/garson/pa765/statnote.htm.

146

Gouldner, H. P. (1960). Dimensions of organizational commitment. Administrative Science
Quarterly, 4(4), 468-490.
Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 13601380.
Granovetter, M. (1983). The strength of weak ties: A network theory revisited. Sociological
Theory, 1, 201-233.
Green, S. B. (1991). How many subjects does it take to do a regression analysis? Multivariate
Behavioral Research, 26(3), 499-510.
Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (2004a). The revolving door. Education Next, 4(1),
77-82.
Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (2004b). Why public schools lose teachers. The
Journal of Human Resources, 39(2), 326-354.
Hom, P. W., Griffeth, R. W., Palich, L. E., & Bracker, J. S. (1998). An exploratory investigation
into theoretical mechanisms underlying realistic job previews. Personnel Psychology, 51,
421-451.
Hom, P. W., Griffeth, R. W., Palich, L. E., & Bracker, J. S. (1999). Revisiting met expectations
as a reason why realistic job previews work. Personnel Psychology, 52, 97-112.
Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression. New York: John Wiley &
Sons, Inc.
Ingersoll, R. M. (2001). Teacher turnover and teacher shortages: An organizational analysis.
American Educational Research Journal, 38(3), 499-534.
Ingersoll, R. M., & Alsalam, N. (1997). Teacher professionalism and teacher commitment: A
multilevel analysis. Washington, DC: U.S. Department of Education, National Center for
Education Statistics.
Ingersoll, R. M., Merrill, L., & May, H. (2011). What are the effects of teacher education and
preparation on beginning math and science teacher attrition? Paper presented at the
annual meeting of the American Educational Research Association, New Orleans, LA.
Johnson, S. M., & Birkeland, S. E. (2003). Pursuing a "sense of success": New teachers explain
their career decisions. American Educational Research Journal, 40(3), 581-617.
Kaufmann, J. (2007). Teaching quality/induction programs and mentoring: Defining the terms
and identifying state policies. Denver, CO: Education Commission of the States.
Kennedy, M. (2005). Inside teaching: How classroom life undermines reform. Cambridge:
Harvard University Press.

147

Kirby, S. N., Berends, M., & Naftel, S. (1999). Supply and demand of minority teachers in
Texas: Problems and prospects. Educational Evaluation and Policy Analysis, 21(1), 4766.
Kirby, S. N., & Grissmer, D. W. (1993). Teacher attrition: Theory, evidence, and suggested
policy options (No. P-7827). Santa Monica, CA: The RAND Corporation.
Kristof, A. L. (1996). Person-organization fit: An integrative review of its conceptualizations,
measurement, and implications. Personnel Psychology, 49(1), 1-49.
Kristof-Brown, A. L., Zimmerman, R. D., & Johnson, E. C. (2005). Consequences of individuals'
fit at work: A meta-analysis of person-job, person-organization, person-group, and
person-supervisor fit. Personnel Psychology, 58, 281-342.
Kushman, J. W. (1992). The organizational dynamics of teacher workplace commitment: A
study of urban elementary and middle schools. Educational Administration Quarterly,
28(1), 5-42.
Lankford, H., Loeb, S., & Wyckoff, J. (2002). Teacher sorting and the plight of urban schools: A
descriptive analysis. Education Policy Analysis Archives, 24(1), 37-62.
Lin, N. (1999). Building a network theory of social capital. Connections, 22(1), 28-51.
Liu, E. (2002). New teachers' experiences of hiring in New Jersey. Paper presented at the Annual
Meeting of the American Educational Research Association, New Orleans, LA.
Liu, E., & Johnson, S. M. (2006). New teachers' experiences of hiring: Late, rushed, and
information-poor. Educational Administration Quarterly, 42(3), 324-360.
Long, J. S. (1997). Regression models for categorical and limited dependent variables (Vol. 7).
Thousand Oaks, CA: Sage Publications.
Lortie, D. C. (1975). Schoolteacher: A sociological study. Chicago: University of Chicago Press.
Meglino, B. M., Ravlin, E. C., & DeNisi, A. S. (2000). A meta-analytic examination of realistic
job preview effectiveness: A test of three counterintuitive propositions. Human Resource
Management Review, 10(4), 407-434.
Mowday, R. T., Steers, R. M., & Porter, L. W. (1979). The measurement of organizational
commitment. Journal of Vocational Behavior, 14, 224-247.
Murnane, R. J., Singer, J. D., & Willett, J. B. (1989). The influences of salaries and "opportunity
costs" on teachers' career choices: Evidence from North Carolina. Harvard Educational
Review, 59(3), 325-346.
National Center for Educational Statistics. (1997). Teacher professionalization and teacher
commitment: A multilevel analysis. Washington, DC: National Center for Education

148

Statistics, Office of Educational Research and Improvement, U.S. Department of
Education.
National Center for Educational Statistics. (2002). Schools and staffing survey, 1999-2000:
Overview of the data for public, private, public charter, and bureau of Indian affairs
elementary and secondary schools. Washington, DC: Institute of Education Sciences,
U.S. Department of Education.
National Center for Educational Statistics. (n.d.). Schools and Staffing Survey. Retrieved May
22, 2011, from http://nces.ed.gov/surveys/sass/index.asp
Norušis, M. J. (2008). SPSS Statistics 17.0: Statistical procedures companion. Upper Saddle
River, NJ: Prentice-Hall Inc.
Phillips, J. M. (1998). Effects of realistic job previews on multiple organizational outcomes: A
meta-analysis. Academy of Management Journal, 41(6), 673-690.
Planty, M., Provasnik, S., Hussar, W., Snyder, T. D., Kena, G., Hampden-Thompson, G., et al.
(2007). The condition of education 2007 (No. NCES 2007-064). Washington, DC:
National Center for Education Statistics, Institute of Education Sciences, U.S.
Department of Education.
Porter, L. W., & Steers, R. M. (1973). Organizational, work, and personal factors in employee
turnover and absenteeism. Psychological Bulletin, 80(2), 151-176.
Portes, A. (1998). Social capital: Its origins and applications in modern sociology. Annual
Review of Sociology, 24, 1-24.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data
analysis methods (2nd ed. Vol. 1). Thousand Oaks, CA: Sage Publications.
Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R., & du Toit, M. (2004). HLM 6:
Hierarchical linear and nonlinear modeling. Lincolnwood, IL: Scientific Software
International, Inc.
Rickman, B. D., & Parker, C. D. (1990). Alternative wages and teacher mobility: A human
capital approach. Economics of Education Review, 9(1), 73-79.
Riehl, C., & Sipple, J. W. (1996). Making the most of time and talent: Secondary school
organizational climates, teaching task environments, and teacher commitment. American
Educational Research Journal, 33(4), 873-901.
Rosenholtz, S. J., & Simpson, C. (1990). Workplace conditions and the rise and fall of teachers'
commitment. Sociology of Education, 63(3), 241-257.
Roux, A. V. D. (2004). A glossary for multilevel analysis: Part III. Epidemiological Bulletin,
25(1), 14-16.

149

Rumberger, R. W. (1987). The impact of salary differentials on teacher shortages and turnover:
The case of mathematics and science teachers. Economics of Education Review, 6(4),
389-399.
Smith, T. M., & Ingersoll, R. M. (2004). What are the effects of induction and mentoring on
beginning teacher turnover? American Educational Research Journal, 41(3), 681-714.
Somech, A., & Bogler, R. (2002). Antecedents and consequences of teacher organizational and
professional commitment. Educational Administration Quarterly, 38(4), 555-577.
Theobald, N. D. (1990). An examination of the influence of personal, professional, and school
district characteristics on public school teacher retention. Economics of Education
Review, 9(3), 241-250.
Veenman, S. (1984). Perceived problems of beginning teachers. Review of Educational
Research, 54(2), 143-178.
Wanous, J. P., Poland, T. D., Premack, S. L., & Davis, K. S. (1992). The effects of met
expectations on newcomer attitudes and behaviors: A review and meta-analysis. Journal
of Applied Psychology, 77(3), 288-297.
Weiss, E. M. (1999). Perceived workplace conditions and first-year teachers' morale, career
choice commitment, and planned retention: A secondary analysis. Teaching and Teacher
Education, 15, 861-879.

150