CLASS SIZE AS A FACTOR OF TEACHERS
SENSE OF SELF EFFICACY
Leslie Jamieson Sharp
B. A., Rockford College, 1970
M.S. Ed., Northern Illinois University, 1980
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Administration, Curriculum, and Supervision
1993 by Leslie Jamieson Sharp
All rights reserved.
This thesis for the Doctor of Philosophy
Leslie Jamieson Sharp
has been approved for the
Steve Del Castillo
Sharp, Leslie Jamieson (Ph.D., Administration, Curriculum, and
Class Size as a Factor of Teachers Sense of Self Efficacy
Thesis directed by Professor Michael J. Murphy
This study explored interactions between class size, teachers sense
of self efficacy, ideal class size, student characteristics, and
instructional/management methods using a mixed design of qualitative
and quantitative methods.
The three-phased study involved screening, interviewing, and
surveying elementary teachers (grades 1-3) in nine school districts in
western Colorado. The phase one questionnaire contained demographic
questions which elicited background information. Two questions from
Rand studies identified high and low efficacy teachers. Phase two
included interviews with seven self-selected teachers using a semi-
structured interview guide and analysis of responses using interpretive
techniques. Phase three employed a survey based on analysis of the
interviews. Efficacy was determined by the Gibson and Dembo Efficacy
Scale. Appropriate statistical procedures were used including ANOVA,
correlation, multiple regression, factor analysis, and Cronbachs alpha
Interviews indicated efficacy was situational and influenced by many
variables. One variable which affected the efficacy of some teachers was
large class size. Teachers defined their sense of efficacy in terms of
student outcomes. While interviews indicated class size could affect
efficacy, there were no significant differences between high, moderate, or
low efficacy groups for class size.
Neither the interviews nor the survey results yielded consensus
among teachers on ideal class size. Rather, teachers concepts of class
size were based upon classroom experience and student characteristics.
The survey revealed that student characteristics were an important
component of the class size issue. Analyses supported the interview
findings that the concept of class size was influenced by subjective
Large classes added a constraint to preferred instructional/
management methods which lower efficacy teachers were not able to
overcome, whereas higher efficacy teachers persisted using preferred
instructional methods. All teachers believed that students benefited to a
greater degree in smaller classes both academically and affectively.
School administrators are admonished to make procedural decisions
mindful that there are no simple solutions to the complicated class size
and efficacy question. The interaction between variables is situation
specific and subject to systemic and external forces. It is recommended
that workplace conditions be consistent with maintenance of efficacy and
both student and teacher needs be considered.
This abstract accurately represents the content of the candidates thesis.
I recommend its publication.
This thesis is dedicated to my husband, Carl.
Carissimo amico, meo marito, maximas
gratias propter exhortationes tuas,
sententias optimas, et amorem constantem, ago.
Forsan et haec olim meminesse iuvabit."
1. INTRODUCTION............................................. 1
Operational Definitions for the Study.................... 9
Outline of the Preliminary Phase...................... 11
Outline of the Interview Phase........................ 12
Outline of the Survey Phase......................... 13
Implications of the Study............................... 14
Structure of the Study.................................. 15
2. LITERATURE REVIEW....................................... 16
Literature Supporting the Relationship
Between Class Size and Student Achievement.............. 16
Literature Not Supporting the Relationship
Between Class Size and Student Achievement.............. 25
Efficacy ............................................... 32
Class Size and Teacher Efficacy......................... 40
Implicit Psychology..................................... 43
3. THE RESEARCH METHODS
Research Questions Guiding the Study..................... 48
Description of the Research Setting ..................... 49
Description of Interview and Survey Populations ......... 52
Interview Sample ..................................... 52
Teacher One, Sally ................................. 55
Teacher Two, Joe ................................... 56
Teacher Three, Alta .................................. 56
Teacher Four, Beth .................................. 57
Teacher Five, Meg.................................... 58
Teacher Six, Joan .................................... 58
Teacher Seven, Ann.................................... 59
Survey Population .................................... 59
Preliminary Phase Questionnaire....................... 60
Interview Phase....................................... 61
Survey Phase ......................................... 61
Survey Response....................................... 63
Nonresponse Bias...................................... 64
Research Question One: Efficacy Questions............. 65
Research Question Two: Boundary
Conditions of Class Size.............................. 66
Research Question Three:
Student Characteristics.................................. 66
Research Question Four: Instructional/Management
Methods ................................................. 68
Data Collection: Preliminary Phase ......................... 69
Data Collection: Interview Phase........................... 71
Data Collection: Survey Phase .............................. 73
Analysis of Interview Data ................................. 74
Classifying and Coding Data.............................. 75
Insuring Trustworthiness................................. 77
Additional Analysis of Interview Data....................... 79
Cultural Meaning ...................................... 79
Data Analysis of the Survey Instrument..................... 81
Initial Data Analysis and Reduction of Survey Response 82
Efficacy Groupings....................................... 83
Analysis of Variance (ANOVA) Procedure................... 85
Limitations ................................................ 86
4. FINDINGS .................................................... 91
Research Question One: Class Size and Efficacy.............. 91
Teacher Interviews....................................... 91
Teachers Definition of Self Efficacy.................... 92
Class Size and Efficacy.................................. 95
Survey Results........................................... 98
Discussion of Class Size and Efficacy.............105
Research Question Two: Boundary Conditions of
Class Size ..............................................107
Teacher Interviews ....................................107
Discussion of Boundary Conditions
of Class Size .........................................117
Research Question Three: Student Characteristics ........119
Teacher Interviews ....................................119
Discussion of Student Characteristics..................133
Research Question Four: Instructional/
Teacher Interviews ....................................136
Discussion of Instructional/
Summary of Results ......................................149
5. SUMMARY OF FINDINGS, CONCLUSIONS,
IMPLICATIONS, AND RECOMMENDATIONS........................153
Research Questions and Design ...........................153
Summary of the Findings..................................154
Class Size and Teacher Self Efficacy ..................155
Boundary Conditions of Class Size......................156
Student Characteristics and Class Size.................158
Class Size and Efficacy.........................161
Boundary Conditions of Class Size...............162
Recommendations for Further Research.............168
A. CONFIDENTIAL QUESTIONNAIRE ON CLASS SIZE ........173
B. CLASS SIZE AND TEACHER EFFICACY
C. SURVEY FOR FINAL PHASE...........................177
D. CONSENT FORM.....................................184
E. DATA SHEET.......................................187
F. SALLYS VERBATIM STATEMENTS .....................190
G. JOES VERBATIM STATEMENTS........................195
H. ALTAS VERBATIM STATEMENTS.......................200
I. BETHS VERBATIM STATEMENTS.......................206
J. MEGS VERBATIM STATEMENTS .......................213
K. JOANS VERBATIM STATEMENTS.......................218
L. ANNS VERBATIM STATEMENTS........................223
M. SPSS VARIABLE LABELS .......................227
3.1 Respondent Profiles.................................53
3.2 Table of Means, Early Versus
3.3 EFFICACY Groupings .................................84
4.1 Descriptive Statistics for
All Survey Variables............................ 100
4.2a Partial Correlations: Efficacy and
Class Size, with Actual Class
Size Removed.................................... 104
4.2b Partial Correlations: Efficacy and
Actual Class Size, with Class Size
Removed ........................................ 104
4.3 ANOVA: Efficacy Groups and
Demographic/Baseline Information................ 105
4.4a Cross Tabulation of High, Moderate,
and Low EFFICACY Groups and Use Less
Cooperative Learning (DIFFER) .................. 113
4.4b Cross Tabulation of High, Moderate,
and Low EFFICACY Groups and Do More
Homogeneous Grouping (DOMORE)................... 114
4.4c Cross Tabulation of High, Moderate,
and Low EFFICACY Groups and
Individualize Instruction (INDIV)............... 115
4.4d Cross Tabulation of High, Moderate,
and Low EFFICACY Groups and Use More
Direct Instruction (DIRECT)..................... 116
4.5a Descriptive Statistics of Student
Characteristics: Few Difficulties
in Teaching..................................... 123
4.5b Descriptive Statistics of Student
Characteristics: Many Difficulties
in Teaching....................................... 124
4.6 EFFICACY Correlations with Demographics/
Baseline Information and Student
Characteristics; Section A........................ 125
4.7 ANOVA: EFFICACY Groups and Percent of Class
Hard to Teach..................................... 126
4.8 EFFICACY Correlations with Difficult to
Teach Students.................................... 127
4.9a ANOVA: EFFICACY Groups and Difficult to Teach
Student Characteristics........................... 128
4.9b ANOVA: EFFICACY Groups and Student
Characteristics of Current Class.................. 129
4.10 ANOVA: Class Size Groups and Student
Characteristics .................................. 130
4.11 Factor Analysis of Student
Characteristics Variables ........................ 131
4.12 Descriptive Statistics for Instructional/
Management Methods: Preferred
Instructional Methods............................. 142
4.13 Correlations Between EFFICACY and Instructional/
Management Methods................................ 143
4.14 ANOVA: EFFICACY Groups and Instructional/
Management Methods................................ 144
4.15 Descriptive Statistics for Instructional/
Management Methods: Changes in Management
with Large Classes ............................... 145
4.16 Descriptive Statistics for Instructional/
Management Methods: Grouping in Larger
Classes .......................................... 145
4.17 Descriptive Statistics for Instructional/
Management Methods: Limitations of
Small Classes..................................... 146
4.18 ANOVA: EFFICACY Groups and Instructional/
Management Methods................................ 146
The national outcry for educational reform is pervasive. In virtually all
parts of the country educational reform is a topic of concern. Various
organizations have different agendas for reform. Federal and state
policymakers view reform as an opportunity to "trim the fat" from
educational bureaucracies and hold education accountable for the
massive expenditures of public monies. Teachers associations consider
it an opportunity to empower teachers. Some professional educators
and researchers see reform as an opportunity to make education more
socially relevant (English & Hill, 1990, p. 1). By contrast, institutes such
as the Carnegie Institute envision a changed curriculum based on the
humanities (Boyer, High School. 1983).
The impetus for reform is due to solicitude over low student
achievement. Poor student performance and the failure of our schools is
the subject of national reports such as A Nation at Risk (1983) as well as
the popular press and television. More recently, our schools have been
compared to the Japanese system and have been found wanting
(Rothman, 1991, pp 6-7). This educational malaise has been linked to
the United States overall economic decline and loss of stature as a world
leader as well as the current economic recession. Business, realizing
that,"... a shrinking labor supply, poorly prepared workers, and
heightened international competition will still exist when the recession
ends ..have formed over 73,000 school-business partnerships across
the country to reform and improve education (Weisman, 1991, p. 19).
Student achievement is a complex equation and every interest group
has a strong opinion about it. Student achievement is influenced by a
number of factors including students backgrounds and neighborhoods,
teaching behaviors, school organization, and socio-economic factors.
Although student achievement is influenced by these variables, many of
which are outside the schools control, a variable with a long history in
achievement research is class size. The precise relationship of class size
to achievement has not been completely revealed, but it appears there is
a modest relationship between the two.
Class size has been researched in depth for many years. Teachers,
a formidable force in the class size debate, have long believed that the
relationship between class size and student achievement is strongly
negative and that class size should be reduced to enhance student
achievement. They continually stress the importance of small classes in
polls conducted by the National Education Association (NEA) ("Teacher
Opinion Poll: Professional Satisfaction," 1975, & Teacher Opinion Poll,"
1975). Teachers are willing to sacrifice pay increases in order to hold the
number of students constant or reduce the number (Gibbs, 1991, p. 32)
demonstrating the obvious importance of class size to them. As Bain
and Achilles stated,
Class size has been a continuing issue in negotiations between
teachers and school boards, and the need to attend to class size
remains a popular topic for discussion in education organizations.
(1986, p. 662)
McLaughlin and associates stated,
Problems related to the composition of classes-particularly class
size and the increased academic and emotional needs of
students--head the list as a source of teacher dissatisfaction and
concern. (Bain & Achilles, 1986, p. 421)
Class size has been largely left out of the recent national debate on
reform, but states and school districts continue to address it. Thirty
states have adopted class size limits (Webster & McMillin, 1991, p. 81).
Individual district contracts, too, often contain class size targets. Two
states, Tennessee and Indiana, have implemented large-scale studies to
learn if student achievement can be improved by reducing class size
(Swan, Stone, & Gilman, 1987; Word et al., 1990).
A difficulty with class size research and policy is that there is little
evidence on how teachers think about class size and its effect on student
achievement to construct their teaching reality. Based on implicit
psychology, it has been shown that people construct their realities based
on past experiences and use these experiences to predict the future
(Wegner & Vallacher, 1977). "In this sense, everyone is his own
psychologist, sociologist. .. and physicist" (Argyris & Schon, 1974, p. 8).
Because teachers work with students in the classroom on a daily basis,
logic would dictate that they would be the people to consult regarding
class size and student achievement. The difference between research
and practice is reflective of the historical tension between teachers
beliefs concerning class size and the conclusions of some research
which finds that class size does not make a major difference in student
One way of exploring beliefs of what can be accomplished by a
teacher is through the concept of self efficacy. Self efficacy is the belief
that one can accomplish his/her goals in spite of obstacles. Research
has indicated that teacher self efficacy is one factor of student
achievement (Ashton & Webb, 1986). One way to view how teachers
think about their teaching reality, then, is through self efficacy.
Efficacy allows a person to persist in spite of perceived difficulties
until s/he succeeds. Bandura defines efficacy as "the conviction that one
can successfully execute the behavior required to produce outcomes"
(Dusek, 1985, p. 23). Fuller, Wood, Rapoport, & Dornbuch (1982)
expand the view of teacher efficacy to include more than academic
goals. They define efficacy as the "... individuals perceived expectancy
of obtaining valued outcomes through personal effort" (p. 7). They
identify it as a psychological precursor to outcomes such as student
achievement and effective instruction and draw a distinction between
organizational/performance and personal/self efficacy.
Research by Gibson and Dembo (1984) and Ashton and Webb
(1986) developed multidimensional models of teacher efficacy which were
influenced by Banduras theory of self-efficacy.
Bandura (1977) argued that although locus of control is primarily
concerned with causal beliefs about action-outcome
contingencies or a persons estimate that a given behavior will
lead to certain outcomes, personal efficacy is concerned with the
conviction that one can successfully execute the behavior
required to produce the outcomes. Outcome and efficacy
expectations are differentiated because individuals can believe
that certain behaviors will produce certain outcomes, but if they
do not believe that they can perform the necessary activities, they
will not initiate the relevant behaviors, or if they do, they will not
persist. (Gibson & Dembo, 1984, pp. 569-570)
Ashton and Webbs (1986) multidimensional model of teachers
sense of efficacy is hierarchical. They draw a distinction between
teaching efficacy, "expectations that teaching can influence student
learning and have an effect on student performance" and personal
teaching efficacy, "individuals assessment of their own teaching
competence to bring about student learning" (p. 4). Their construct
includes both Rand efficacy factors of personal and organizational
efficacy (p. 5). The distinction is important because it suggests that if
teachers believe they have an effect on students in spite of obstacles,
they will be willing to make the necessary changes and/or learn
strategies to improve themselves and thereby help students.
In addition to the above research, teacher efficacy and student
achievement have been linked in research by Filby, Cahen, McCutcheon,
and Kyle (1980) and Glass, Cahen, Smith & Filby, (1982). These studies
consider teachers curricular modifications, amount of individualization
with students, lower stress, increased teacher-pupil interactions/relations,
and elevated morale as positive indicators of small class benefits. One
interpretation of these studies is that teacher efficacy affects student
A variety of situational features have been shown to affect teachers
efficacy including class size. Teacher attitudes of workload, morale,
perceptions and satisfaction, expectations for performance, absences,
and professional growth were quantitatively described and found
favorable in smaller classes (Glass et al., 1982).
Bandura's description of bidirectional determinism emphasizes the
interaction/interdependence between perceived self-efficacy and
behavior. "Sense of self-efficacy influences behavior and the
consequences of that behavior alter perceived self-efficacy through a
continual bidirectional determinism" (Ashton & Webb, 1986, p. 12).
Similarly, one can hypothesize the interaction between teachers sense of
efficacy and perceived class size as being bidirectional. Thus, it may be
advantageous to investigate class size in terms of its interdependence
Existing research demonstrates a linkage between class size and
student achievement. There is much research to show that student
achievement and teacher efficacy are related. However, there is a lack of
definitive research to demonstrate a relationship between class size and
teacher efficacy. By examining these two variables, further
understanding of what leads to increased student achievement may
The purpose of this study is to explore the interaction of teachers
sense of class size being too large and teachers sense of efficacy. The
understanding that results will enable researchers, educators, and
policymakers to make informed decisions regarding class size and the
allocation of resources in schools.
Class size and teacher efficacy have both been shown to be related
to student achievement, but little is known about how these two
enigmatic variables interact. Drawing upon implicit psychology (Wegner
& Vallacher, 1977), it may be the case that the way teachers construct
this interaction affects their actions and how they perceive their ability to
be effective. Further, because efficacy and class size have been shown
to be related to student achievement, an examination of the two variables
class size and teacher efficacy is pertinent as teachers are advocates of
the research that says class size has an impact on student learning, at
least in the lower grades.
There is a complex network of influences at work in the school
organization. Unraveling these influences will aid in understanding the
interaction or interdependence between class size and efficacy which
may help clarify the relationship between class size and student
achievement. Student achievement is the critical outcome of the
This study examines teachers perception of personal efficacy and
class size. Its focus is on how teachers think about their teaching load,
particularly as it is expressed in the number of students with whom they
interact, the characteristics of those students, and its impact on their
sense of their capacity to bring about the student learning they seek
using preferred instructional/management methods.
The research questions that guide this study are:
1. How do class size and teachers perceived sense of efficacy interact?
2. What do teachers perceive as ideal class size?
3. How do student characteristics affect teachers sense of class size?
4. What are the interactions between class size and instructional/
management methods as perceived by teachers?
Operational Definitions for the Study
Classroom management strategies a teacher uses to organize and
manage students. It includes a variety of grouping techniques and
disciplinary methods use with disruptive students.
Class size the number of students a teacher works with on a daily
basis. As this study will be concerned with teachers perceptions of
large/small classes and the concomitant effects on their feelings of
efficacy, no a priori definitions of large and small will be used.
Efficacy perceived ability of a teacher to personally cause learning in
students and teacher belief that teaching itself can overcome inherent
Instructional methods individualized/one-on-one instruction, cooperative
learning, Readers and Writers Workshop, or direct instruction.
Student achievement although nationally, student achievement refers to
standardized achievement tests, for the purposes of this study teachers'
assessment of student achievement will dominate the study. None of the
schools involved in the study uses standardized tests below grade four.
Student characteristics special needs, gifted/talented, from
dysfunctional families, low socio-economic status (SES), behavior
problems, culturally diverse, high affective needs, mobile, truant, English
as a Second Language (ESL), and from single parent families.
Because this study examined teachers perceptions of interactions
between efficacy and class size, a combination of qualitative and
quantitative methods was used, forming a mixed design. Class size and
efficacy interactions are complex; therefore, a three-phase study was
conducted. In the preliminary phase, teachers were asked about their
efficacy via a questionnaire and invited to self-select for an in-depth
interview. The next phase involved interviewing seven teachers and
analyzing the results using interpretive/ qualitative methods. In the final
phase, all teachers in the research setting were surveyed in reference to
the variables of sense of class size, student characteristics, self efficacy,
and instructional/management methods.
Outline of the Preliminary Phase
The study consisted of three phases and was conducted over two
school years. To address the research questions, the researcher sent a
questionnaire to 161 elementary school teachers (grades one through
three) in Western Colorado to identify high/low efficacy teachers, their
definition of ideal class size and additional background data (Appendix
A). This geographic area, composed of nine school districts, which will
be described in greater detail in Chapter 3, was chosen because the
researcher worked with the districts and had access to the sample.
The two questions from the two Rand studies which were developed
to determine teacher efficacy were pervasive throughout the literature
(Ashton, Doda, & Webb, 1983; Ashton & Webb, 1986). They were
chosen as a screening devise for the preliminary phase of this study to
determine high and low efficacy teachers.
1. When it comes right down to it, a teacher really cant do
much because most of a students motivation and
performance depends on his or her home environment.
2. If I really try hard, I can get through to even the most
difficult or unmotivated students. (Ashton & Webb, 1986,
A true-false forced-choice format was chosen to identify high and low
efficacy teachers. Informational questions relating to years of experience,
gender, school, current and ideal class size were included to aid in
developing a respondent profile.
Outline of the Interview Phase
Seven teachers were selected from those 38 respondents who
volunteered to be interviewed as a result of the preliminary screening
process. This group was chosen based on the teachers diversity of
experience, their current class sizes, their statement of ideal class sizes,
and their dispersion throughout the study area. The responding teachers
scored high on the efficacy portion of the questionnaire, there was a
diversity of experience ranging from six months to 24 years, and a variety
of actual and ideal class sizes. Further, the school districts were diverse
demographically and geographically.
These seven teachers were interviewed at their schools using a
semi-structured interview based on the Ashton and Webb (1986)
interview format as a guide (Appendix B). They were interviewed in order
to establish a relationship between class size and efficacy in actual
practice. Although an interview guide was used, the interviews were
open-ended and respondents were encouraged to elaborate.
Responses were tape recorded with their permission and later
transcribed using a word processing program and a personal computer.
Interpretive analysis techniques suggested by Miles and Huberman
(1984) were used to analyze the transcriptions based on the following
four categories established by the research questions:
1. How do class size and teachers' perceived sense of efficacy
2. What do teachers perceive as ideal class size?
3. How do student characteristics affect teachers sense of class size?
4. What are the interactions between class size and instructional/
management methods as perceived by teachers?
Outline of the Survey Phase
To explore the variables quantitatively, a survey was developed
(Appendix C). It was a means of gaining more precise information to
answer the question whether, based on their degree of efficacy, teachers
changed their preferred instructional/management methods because of
class size and/or student characteristics. The entire population of 164
primary teachers in grades one through three in the nine original districts
was surveyed. The survey established further consistency by presenting
all of those canvassed with a standard set of questions. Additionally, this
survey enhanced confidentiality to a greater degree than the interviews,
which may have given respondents greater confidence in their anonymity
and thus freed them from constraints when being interviewed (Fowler,
Quantitative methods of correlation, partial correlation, cross
tabulation, factor analysis, and ANOVA were chosen to analyze the
survey data. A reliable and valid instrument, the Gibson and Dembo
Teacher Efficacy Scale (1984) was available to measure efficacy in
section B, questions one through sixteen. Extending this scale,
additional questions were directly developed from the interviews which
considered the interactions of the variables of class size and efficacy,
class size and instructional/management methods, and class size and
Implications of the Study
In this era of educational reform and demands for fiscal
accountability, it is apropos to examine the interaction between class size
and teachers' sense of efficacy. It is the responsibility of school
administrators to take cognizance of teachers' effectiveness as teachers
are an integral part of the educational process. It is also the
administrators' responsibility to appraise student progress and hold
expenses to an acceptable level. Therefore, it would be of benefit to
educators and the public to understand the interdependence among
class size, student characteristics, and teachers sense of efficacy rather
than considering improving student achievement in isolation.
Further, policymakers should develop a better perspective of the
forces at work in the school setting. They will then not be as subject to
the whims of capricious interest groups if they understand what will be
gained and what will be lost as class size is increased at the expense of
teacher efficacy. Knowledge of the effects of class size or other student
characteristics on teachers will also allow administrators to establish
policies that enable student achievement to be maximized without
Structure of the Study
This study is written in standard thesis format. Chapter 1 introduces
the research topic and rationale. Chapter 2 reviews the related literature.
Chapter 3 describes the methods used in the three phases of the study.
Chapter 4 details and analyzes the findings relating to the research
questions for phases two and three. Chapter 5 evaluates the major
findings of the research questions, compares the results of the research
questions, and makes recommendations for implementation and further
research based on the findings.
The literature cited in this study concerns research which supports or
refutes the relationship between class size and student achievement,
research on teacher efficacy as it relates to class size, and literature
which relates positively and negatively to student achievement and
teacher efficacy. Finally, implicit psychology is reviewed briefly as it
relates to teacher conceptions regarding class size.
Literature Supporting the Relationship Between Class Size and Student
Class size is a hotly debated issue that has been studied extensively
since the early 1900s. The most recent period of class size research is
often credited to Blakes 1954 review of 267 class size studies done prior
to 1950. From these studies, he analyzed 85 which produced conflicting
results on class size effects. His study was a landmark as he established
criteria to determine scientific validity (Word, Johnston, Bain, Fulton,
Zaharias, Achilles, Lintz, Folger, & Brede, 1990, p. 200).
Class size studies most often focus on student achievement effects
because assessment tools are readily available (Bernstein, 1973; Cahen
et al., 1983; Filby et al., 1980; Glass et a!., 1982; Mortimore, Sammons,
Stoll, Lewis & Ecob, 1988; Sindelar, Rosenberg, Wilson, & Bursuck,
Cahen et al. (1983) conducted a study of early primary grades in a
school in rural Virginia and a school in urban California. In this study,
students were initially placed in classes of varying sizes, or the classes
were changed during the school year. The California classes were
reduced from 35 to 22 students. The Virginia classes were reduced from
nineteen to thirteen students. The achievement tests from the Beginning
Teacher Evaluation Study were used as the basis of comparison. A
large percentage of students in the reduced classes scored higher on
the post-testing than had been predicted from the pre-tests. The smaller
classes produced an increase in student-teacher interaction, less waiting
time for students to get help from the teacher, and increased student
attention to the instructional task.
Filby et al. (1980) also found that attention rates for students
increased from 56 percent to 72 percent as class size decreased.
Additionally, there were fewer class disturbances and students did not
have to wait as long to get teacher attention in smaller classes.
Sindelar et al. (1984) found that reduced class size maximized factors
related to achievement. There was increased time on task, teacher
feedback, and teacher monitoring of student performance. They
concluded that the smaller class provided greater opportunity for teacher
engagement with students.
Glass and Smith (1978) developed meta-analysis, a statistical
treatment which allows unlike metrics to be compared. They applied this
technique to 77 class size studies which yielded over 700 comparisons of
achievement in classes in different sizes. They found that the greatest
gains in achievement came in classes of fewer than fifteen students. The
difference in being taught in a class of 20 as opposed to a class of 40
had an advantage of 6 percentile ranks (Glass & Smith, 1978). "A clear
and strong relationship between class size and achievement has
emerged .. [to show that] other things equal, more is learned in smaller
classes" (Geary, 1986, p. 70). "The overall strength of the relationship
would not be judged as small by most persons who study educational
production functions (Glass, 1980, p. 243).
In a follow-up study Smith conducted in 1979, similar achievement
gains were found, but the emphasis was on the effects smaller classes
had on teachers and the classroom environment. The results
revealed a substantial relationship between class size and
teacher and pupil attitudes as well as instruction.... Smaller
classes are associated with greater attempts to individualize
instruction and better classroom climate. (Smith, & Glass, 1979, p.419)
Bay City, Michigan, conducted a year long study of the achievement
effects of reduced class size on student achievement. In this study,
class size was "reduced" by using teacher aides in teacher-directed
classes of at least 45 students. Elementary classes with an aide made
greater academic progress than students in smaller classes with no aide
(Park, Carroll, Chase, Hymes, McCuskey, Rast, & Rulon, 1956).
Ballow (1969) conducted a three year longitudinal experimental
primary-grade reading program in Riverside, California. Using scores
from the Metropolitan Readiness Tests, the California Short Form Test of
Mental Maturity, the Metropolitan Achievement Test, and School and
College Aptitude Test in grades one through four. The treatment group
was composed of 656 students and the control group had 602 students.
The sample remained constant. Small classes had fifteen students and
large classes had 30 students. The researcher discovered that
achievement improved when students remained in small classes for two
consecutive years. He determined that small classes were decisive in
improving reading achievement in the first two grades, but by third grade
the effect was diminished (Ballow, 1969).
A More Effective Schools Program was conducted in Cleveland, Ohio,
for three years in two selected schools with the goal of improving
achievement of disadvantaged inner-city children. The program intended
to raise the achievement levels of all of the children in all grades by
altering organizational and instructional patterns school-wide. In the first
two years of the study, students demonstrated significantly higher
achievement gains than the control group, although the gains were not
as high during the third year (Taylor & Fleming, 1972).
Furno and Collins (1967) conducted a five year longitudinal study in
Baltimore, Maryland, on 16,449 students. They examined the effects of
class size which ranged from one to 25, 26-31, 32-37, and 38 or more,
on mathematics and reading achievement. The students were tested at
the end of grade three. The researchers also cross-classified students
by IQ score, whether the student was in special education or not, the
mothers occupation, and race. They found that smaller classes yielded
higher achievement gains in math and reading, with even higher gains for
students in special education classes and minority students (Educational
Research Services, 1978, pp. 20-23).
Longitudinal research projects conducted indicate that class size and
student achievement are interdependent. Beginning in 1981, Indiana
embarked on a two-year investigation to study the effects on
achievement of reducing class size. After a years study, the increase in
student achievement test scores induced the State Department of
Education to conduct the statewide PRIME TIME project. The project
began in 1984-1985 with first grade classes by reducing class size to
eighteen, expanded in 1985-1986 to include second grade, included third
grade classes in 1986-1987, and in 1987-1988 reduced class size for
grades kindergarten through three. As nearly all classes in the state
were able to reduce class size because of state funding, it was not
possible to conduct a scientific study with established control and
treatment groups. Since some of the schools already had small class
sizes, it was possible to compare their achievement test results to the
actual reduced class size results (McGiverin, Gilman, & Tillitski, 1989).
The results of studies from ten selected districts in Indiana yielded a
total of 24 comparisons (3,967 scores) as reported by McGiverin et al.
(1989) in their "Meta-analysis of the Relation between Class Size and
Achievement." Statistical results, using Fishers inverse chi-square
procedure, demonstrated that students in the PRIME TIME study, where
class size was held at eighteen, had higher achievement in basic skills
after two years than students in larger classes and that primary children
learned more efficiently in smaller classes.
The authors caution that meta-analysis might not be the ideal method
of analyzing combined test results. They note that research continues to
indicate that the relationship between class size and student achievement
is complex and that reducing class size by itself will not necessarily
improve student achievement. They suggest additional research to
investigate the factors in small classes that combine to increase student
A 1983 yearly report concluded that three major results emerged
from the study, students in the program had higher achievement test
scores than those not in the program, teachers felt they increased their
effectiveness, and there were fewer discipline problems in smaller classes
(Project Primetime: 1982-83 Report, 1983).
Mueller, Chase, and Walden (1988) studied 29 out of 304 school
districts in Indiana to determine results of the PRIME TIME program.
They collected and compared pre and post treatment achievement test
data. Tests of twelve selected, representative districts indicated
significant (30 percent) gains in mathematics and reading (50 percent) in
the first grade and 10 and 20 percent gains respectively in the second
grade. Further, surveys and interviews conducted with students, parents,
and teachers in the representative districts revealed that 80 percent of
the teachers felt that both below average and above average students
were achieving more than they were before the program.
Teachers reported more interaction with students with more time for
individualizing instruction, better attention, and more immediate feedback
with reduced class size. Additionally, pupils were on task for a greater
length of time, the classroom was quieter and less frenzied, and teachers
were more enthusiastic about their jobs.
Project STAR in Tennessee is another statewide effort to improve
student achievement by reducing class size. In the spring of 1984, the
Tennessee State Legislature adopted a comprehensive educational
reform plan known as the Better Schools Program. It was an opportunity
for higher learning institutes to improve public education. The director of
the program, Bain, proposed statewide adoption of a reduced class size
model similar to that in Indiana. Because the literature, particularly the
Glass and Smith meta-analysis (1978) recommended class size be
lowered to fifteen and the costs of this reduction would be so high, the
legislators wanted a well-designed study conducted (Word et al., 1990).
The study involved 6,500 kindergarten students in 75 schools who
to small (13-17), regular (22-25), and regular-with-aide classes, and
kept the students in the same type of class through 3rd grade.
Standardized test scores in reading and math showed that small
classes were one month ahead of regular classes by the end of
kindergarten, and two months ahead by the end of 1st grade. These
effects were maintained, though not increased in 2nd grade. This
sustained impact is greater than can be expected from many
educational initiatives. ... (Willis, 1990, pp. 1 & 5)
Further, of the classes that scored in the top ten percent on the SAT
Total Reading test, 18 of the top 33 classes were in small kindergarten
classes, 22 of the top 34 classes were in small first grade classes, 23 of
the top 34 classes were in small second grade classes, and 25 of the top
32 classes were in small third grade classes.
Inner-city minority students who were placed in small classes
consistently out-scored inner-city students in regular and regular with
aide classes. These students also made the greatest gains on the SAT
Relating to the socio-economic factors, non-free lunch minorities in
suburban small classes performed as well as non-free lunch whites.
Also, in every grade in all types of locations and in every class type,
non-free lunch students outperformed free lunch students (Word et al.,
The Project Star Technical report concludes:
The design and magnitude of Tennessees randomized class size
experiment allow researchers to make, with high level of confidence,
statement about class size effects.... This research leaves no
doubt that small classes have an advantage over larger classes in
reading and mathematics in their early primary grades.... This
experiment yields an unambiguous answer to the question of the
existence of a class size effect as well as estimates of the magnitudes
of the effect for early primary grades. (Word et al., 1990, p. 196)
Literature Not Supporting the Relationship between Class Size and
Although class size is one of the most highly researched topics, to
date, empirical research has not generated consistent findings in spite of
the quantity of research that has been conducted. As a result, the
literature is often complex, contradictory, and inconclusive.
In the early 1970s, a three year long study was conducted by the
Cleveland Public Schools at two elementary sites. This "More Effective
Schools Program" had as its goal improving achievement for inner city
disadvantaged students. They hoped to individualize instruction by
reducing class size to no larger than 25, providing more teachers and
equipment, staff development, and involving parents. During the first two
years of the study, students in the treatment schools exhibited higher
achievement compared to the control groups, but in the third year,
overall achievement gains were not as great (Taylor & Fleming, 1972).
Cacha (1982) strongly recommends exploring alternatives such as
innovative ways to groups students in different curricular areas and
improved teacher training before lowering class size. Her review of the
research literature found that "in classes of 25 to 34 pupils above the
primary grades, there is no evidence to conclude that smaller classes will
result in greater academic achievement" (p.14). She also raised
questions about the Glass and Smith meta-analysis because their results
were not accompanied by statements of statistical significance and data
were homogenized during the meta-analysis process obscuring many
Rather than being sophisticated methods of statistical analysis, the
meta-analyses are too insensitive to distinguish subtle relationship
among variables. This type of data analysis leads to
oversimplifications of the results. (1982, p. 15)
Cacha cites the 1978 review of the class size literature by The
Educational Research Service (ERS) just prior to the Glass and Smith
study. ERS determined that reducing class size by itself would not raise
student achievement. Results were inconclusive, relationships between
class size and student achievement complex, and much of this literature
Robinson (1990) concludes in his "Synthesis of Research on the
Effects of Class Size" after analyzing over 100 studies there is little
research to support small classes improving student achievement itself.
He concedes that most positive effects are in grades K-3, that student
attitudes and behavior are affected, and that "smaller classes can
positively affect the academic achievement of economically
disadvantaged and ethnic minority students" (p. 82).
He reviews the four most common approaches that have been taken
by researchers in the past in class size studies, namely descriptive
analysis, meta-analyses, best-evidence synthesis, and related cluster
Descriptive analyses, such as that by the Educational Research
Service in 1978, concluded "that class size has little impact on the
academic achievement of most pupils, in most subjects, above the
primary grades" (p. 80).
Another approach, meta-analyses, which researcher Glass introduced
in the late 1970s as a means of providing statistical rather than
descriptive analyses, calculated the effect size between treatment and
control groups in each study that was reviewed. Glass and Smith
examined 76 class size studies in 1978 and found "only a 6 percentile
rank difference in the mean scores of pupils taught in classes of 20
versus those taught in classes of 40" (p. 80). Nevertheless, a negative,
curvilinear relationship was established which indicates major benefits
result when class size is reduced below 20 students.
The third approach, best-evidence synthesis, introduced by Slavin
and Madden in 1989, combined elements of meta-analysis with
descriptive analysis. Of the eight studies which met his criteria, Slavin
found small, but positive effects on student achievement when class size
was reduced from 27 to sixteen.
Finally, related cluster analysis, which Robinson and Wittebols
applied "to all class size research studies conducted between 1950 and
1985 in K-12 classes containing five or more pupils" (1986, p. 82)
clustered the studies around such factors as grade levels, subject areas,
student characteristics, student achievement, student behavior and
teaching practices. The advantage of this method was the sorting out
from a large body of research results on class size the findings that
related to specific areas. They concluded the most positive effects of
small classes are on grade K-3 in reading and mathematics, although
these effects might not be sustained in later years. Student attitudes and
behavior were most positive in smaller classes in primary grades.
Economically disadvantaged and ethnic minority students could be
positively affected by smaller classes. However, in some classes there
was no effect.
[For] the midrange of 23 to 30 pupils, class size has little impact on
the academic achievement of most pupils in most subjects above the
primary grades . [t]he positive effects of class size on student
achievement decrease as grade levels increase ... [and] little if any
increase in pupil achievement can be expected from reducing class
size if teachers continue to use the same instructional methods and
procedures in the smaller classes that they used in the larger classes.
Supporting similar findings, The U. S. Department of Education in its
policy paper Class Size and Public Policy: Politics and Panaceas
contends that reducing class size is too costly for the small achievement
gains and suggests efforts should be directed towards teacher
recruitment and training to improve educational quality (American School
Board Journal. 1988, p. 12).
Robinson and Wittebols (1986) found that there is no optimum class
size encompassing all types of students, all curricular areas, or all
grades. They advised policymakers to examine research that relates to
their specific areas of concern, such as mathematics and reading, and
focus class size decisions in order to achieve academic goals.
Odden (1990) in an article on class size and achievement reviews the
related research and suggests policy alternatives to reducing class size.
He is critical of the Glass and Smith meta-analysis of 1978.
Meta-analysis gives equal weight to ail study findings, whether they
are from well or poorly designed studies . meta-analyses often
combine studies that are on different topics while ostensibly
addressing the same topic ... [and] meta-analysis [relies] on
statistical interpolations, (pp. 215-216)
Odden asserts that the 1989 Slavin re-analysis of the Glass and Smith
data relies on increased achievement results for classes of less than
twenty students based on statistical interpolations of the findings of the
studies not actual examples of the impact of classes with as few as five
or ten students.
It seems that the impacts of small classes in the Glass and Smith
(1978) study were driven by small group and one-to-one tutoring....
If these effects had been eliminated from their analysis of the 14
studies, class size reduction even down to 15 or 20 would have
shown essentially little or no impact, (p. 216)
Odden cites research from two recent longitudinal studies that were
conducted statewide in Indiana and Tennessee. The Indiana PRIME
TIME project resulted in 0.34 standard deviation over a two year period,
but first year gains were often eroded in the second year.
A derivative of the Indiana statewide study was examined by Tillitski,
Gilman, Mohr, and Stone (1988) in the North Gibson School Corporation.
They used a repeated measures cohort study comparing students in
large class instruction with those in small class size instruction. Initially,
the results indicted statistically significant increases in student
achievement. However, the study has yielded unexpected results over
The gains that were evident during the first two years and attributed
to the state-sponsored reduced class size program had largely
disappeared by the end of grade three.... These results suggest
that gains as a result of class size reduction may be limited to the
early primary grades, (p. 38)
According to Odden (1990), the Tennessee research undertaking
called Project STAR also shows almost no difference in achievement for
third graders who have been in small (less than seventeen) classes since
kindergarten. However, Achilles, a member of the consortium, disagrees
(personal communication, March 22,1991). Achilles believes that the
gains in student achievement are substantial and will be maintained over
Odden (1990) suggests reducing class size for some elementary
students to allow for tutoring or small group work 20 to 30 minutes per
day. He cites the recent success of the Reading Recovery Program in
Ohio which is grounded in the New Zealand early grade individual
tutoring program. He recommends cross-grade and within-class ability
grouping for certain subjects and proposes parallel scheduling to reduce
class size at certain periods of the day. Some other strategies he
suggests include early childhood education, extended day kindergarten,
continuous progress programs, cooperative learning, peer tutoring, and
computer-assisted instruction (pp. 222-223).
Glass, Cahen, Smith, and Filby (1982) reported that a myriad of class
size studies (Fleming, 1959; Haskell, 1964; Shapson, 1980; Shaver &
Nuhn, 1971; Whitney & Willey, 1929-30) reached no unequivocal
conclusion as to the optimum class size. It is difficult to generalize class
size studies to other populations because of the disparity between what
is meant by small or large classes and (Berger, 1982). Populations in
many studies are biased initially by the very nature of the classes
selected and the length of time of the treatment varies widely from study
to study. The Coleman et al. (1966) study did ascertain that at the
national level the average number of students reported was 30.2 for
elementary teachers of white students and 30.5 for teachers of black
students (p. 163).
Glass et al. (1982) reviewed previous studies including the Coleman
study and said,
[these studies] were what research methodologists call
non-experimental studies.... The comparison of what naturally
constituted large and small classes is non-experimental; the
achievement differences might be attributed to class size or to other
differences between the large and small classes .... But huge
studies and multinational data pools can not compensate for the
fundamental deficiencies of these studies, namely that they are not
experiments, and hence may not be a reliable guide to what could be
expected if one particular class were suddenly made smaller.
There has been considerable research in the area of teacher efficacy
and its effect on teacher performance in the classroom. Efficacy is "the
individuals perceived expectancy of obtaining valued outcomes through
personal effort" (Fuller et al., 1982, p. 7). Research identifies two
components to efficacy, organizational efficacy and personal or self
"Organizational efficacy" refers to an organizational actor feeling
efficacious in gaining valued outcomes by influencing another person
in a different level of the organization.... "Performance efficacy"
indicates the perceived efficacy in performing ones own work tasks,
independent of social interaction with other staff members of the
school organization. (Fuller et al., 1982, p. 9)
The model of efficacy that many researchers have developed has its
antecedents in Rotters social learning theory. Rotters "social learning
theory (SLT) was the first attempt to develop a systematic theory of
human behavior that used the expectancy construct" which so many
others such as Bandura have employed (Dusek, 1985, p. 18).
At the simplest level, Rotters fundamental concern was the concept
of internal versus external locus of control, not the related concept of
efficacy, and the effect this belief of control has on the individual
acquiring and performing new skills. A person with an internal locus of
control would compare to a person with a high sense of efficacy. Rotter
characterized the person with an internal locus of control by saying,
... the individual who has a strong belief that he can control his own
destiny is likely to: a) be more alert to those aspects of the
environment which provide useful information for future behavior; b)
take steps to improve his environmental condition; c) place greater
value on skill or achievement reinforcements and be generally more
concerned with his ability, particularly his failures; and, d) be resistive
to subtle attempts to influence him. (1966, p. 25)
One of the factors determining whether a person will acquire new
skills and knowledge is dependent upon whether a person perceives the
outcome as being dependent on his efforts (internal control) or others
(external control). The reinforcement of behavior is not "a simple
stamping-in process, but depends whether or not the person perceives a
causal relationship between his own behavior and the reward" (Rotter,
1966, p. 1).
Rotter hypothesized that the variable of control (internal/external) was
significant in understanding individual differences in learning and the
learning process itself. He developed tests of individual differences in a
generalized belief in internal-external control and validated them in
experimental studies. Based on his social learning theory, as a person
grows, s/he determines if an outcome or reward is dependent on his/her
behavior or an outside force. When it is not based on his/her behavior it
will not increase expectancy of reward in future situations as much as
when it is perceived as being dependent on him/her.
Some predecessors and contemporaries of Rotter provide additional
theoretical perspectives concerned with efficacy including Weiners
(1979) attribution theory, de Charms (1968,1976) personal causation
theory, and Deci and Ryans (1985) intrinsic motivation theory. Weiner
(1979) maintained people are motivated to make attributions about
outcomes in order to make judgments of their self efficacy. De Charms
(1968, 1976) suggested that personal causation, a belief that one can
originated his/her own actions and bring about a desired goal, was a
basis of motivation. Deci and Ryan (1985) found that external pressures
negatively influence motivation and personal efficacy.
Other early motivational researchers considered the concept of
expectancy but did not resolve some theoretical issues. Vroom (1964)
concluded that a person would be highly motivated if s/he had high
expectancy of achieving a valued outcome. Lewin proposed that task
difficulty influenced expectations (Deci and Ryan, 1985).
Rather than relying on Rotters or others theories, though, Ashton
and Webbs situation-specific attitude is based on Banduras self efficacy
model, which Bandura contends is a cognitive mechanism that regulates
behavior and emerges as people develop a feeling of personal
competence (Bandura, 1977, p. 81). Self efficacy is not global, rather it
is specific to a situation.
Efficacy in dealing with ones environment is not a fixed act or simply
a matter of knowing what to do. Rather, it involves a generative
capability in which component cognitive, social, and behavioral skills
must be organized into integrated courses of action to serve
innumerable purposes. (Bandura, 1982, p. 122)
The person gathers information on self efficacy from four sources,
performance accomplishments, vicarious experiences, verbal
persuasions, and physiological states. The most important source of
efficacy is found in the actual accomplishments of the person which have
provided success (Bandura, 1982). "Sense of self-efficacy influences
behavior and the consequences of that behavior alter perceived
self-efficacy through a continual bidirectional determinism" (Ashton &
Webb, 1986, p. 12).
Ashton and Webb (1986) explained their conception of self efficacy:
A sense of self-efficacy is a cognitive mechanism that regulates
behavior. A sense of self-efficacy develops as an individual acquires
a conviction of personal competence.... The strength of an
individuals sense of self-efficacy determines whether he or she will
initiate and sustain a behavior in the face of difficulties. (Ashton &
Webb, 1986, p. 8)
Ashton and Webb (1986) constructed a framework composed of the
microsystem, the mesosystem and exosystem to classify antecedents to
a teachers sense of efficacy. The microsystem in the teachers
classroom. Things in this system that affect efficacy are student
characteristics, teacher characteristics, teacher beliefs, class size, and
role definitions. The mesosystem consists of interacting factors affecting
efficacy such as school size, school norms, student-teacher relations,
and collegial relations. Finally, the exosystem includes the "many formal
and informal social structures external to the school environment11
(Ashton & Webb, 1986, p. 22) which can potentially affect a teachers
sense of efficacy.
In their study, Ashton and Webb (1986) showed a clear relationship
between teacher efficacy and student achievement both statistically and
qualitatively, contending that efficacy is an important component of
The extent to which teachers feel capable of affecting student
achievement influences both the effort they exert in classroom
instructions and their willingness to persist in working with their most
difficult students. (1986, p. 159)
They found that teachers sense of efficacy is influenced by many
factors including school climate, personal, student, economic,
administrative, and collegial. These influences bring about changes in
teacher behaviors as well as student achievement. Efficacy also
fluctuates within the individual depending on the following conditions:
Uncertainty, isolation, and a sense of powerlessness [which] threaten
teachers sense of self esteem, and the lack of adequate economic
rewards and societal recognition increase teachers feelings of
self-doubt. (Ashton & Webb, 1986, pp. vii & ix)
In an earlier study, Ashton concluded schools tend to contribute to
teachers feelings of powerlessness and lack of satisfaction and that one
of the factors teachers have little control over is class size (1984,
Gibson and Dembos research earlier has further supported
Banduras theory that "ones behavior is determined by both a general
outcome expectancy as well as a sense of self efficacy" (Gibson &
Dembo, 1984, p. 574). High efficacy teachers in contrast to low efficacy
teachers spent more time in whole group instruction, were able to
monitor children more effectively, and led children to correct answers
(not telling them the answers) through questioning (1984, pp. 576-579).
Conley, Bacharach, and Bauer (1989) supported Gibson and
Dembos and maintained that school reform efforts neglected the work
environment where teachers feel "they can fulfill their intrinsic motivations"
(Conley et al., p. 59). Teachers perceived that their professionalism was
violated by the bureaucratic organizations in which they work. The
feeling that teachers did not have decision-making authority lead to "a
sense of powerlessness and dissatisfaction" (p. 61). The researchers
cited large class size as an obvious component in teachers feeling they
cannot perform as professionals (p. 63).
Other researchers, namely Darling-Hammond and Wise, found that
educational policies designed to eliminate incompetent teachers had a
negative effect on highly motivated, competent teachers. These
prescriptive policies removed the teachers sense of professionalism and
decision-making. The teachers became de-personalized and their sense
of efficacy deteriorated as bureaucratization increased (1983, pp. 68-69).
Hawley and Rosenholtz concurred with these findings and determined
that teachers needed to play a role in determining school
curricula/policies, to share the values of the system, and to have
adequate training and support in the use of new techniques (1984, p.
More control over professional decisions and training for teachers
were suggested by Ashton and Webb (1986) and Gibson and Dembo
(1984) to reform education.
It is not likely that dramatic changes in teacher attitudes or in the
educational system will result from microsystem experiments
designed to develop teachers sense of efficacy by training them to
behave like teachers with high efficacy. (Ashton & Webb, 1986,
Teachers must be involved in the decision-making process of the school
(Ashton & Webb, pp. 169-171).
Ashton and Webb (1986) and Gibson and Dembo (1984) contend
there are some measures that can be taken to change a persons feeling
of efficacy. Factors which can be changed include working conditions
such as class size and student type.
Class Size and Teacher Efficacy
If one looks more closely into the class size research, there is
evidence to show that class size matters in many ways to teachers.
Referring to the original Glass and Smith meta-analysis, Glass contended
"Our study of class size and affective outcomes shows that increasing
class size from 20 to 40 will take its toll on pupil attitude and interest,
teaching practices, and teacher morale" (1980, p. 242).
In a companion piece of the original meta-analysis which analyzed 59
studies, Smith and Glass concluded,
reducing class size has beneficial effects both on cognitive and
affective outcomes and on the teaching process itself. These
relationships have not in the past been apparent because of an
inability to deal with either the class sizes or the effects.,.. (Smith &
Glass, 1980, p. 432)
Repeatedly in research opinion polls, teachers cite class size as one
of their overriding concerns. In a nationwide survey conducted by the
NEA in 1974 concerning class size, teachers reported that "they believed
small classes were extremely important in improving the academic
achievement of pupils" (79.7 percent); they "considered small classes
extremely important for the social and personal development of pupils"
(approximately 64 percent); and they considered small classes "extremely
important for job satisfaction for the teacher" (74.1 percent) Teacher
Opinion Poll," (Today's Education. 1975, p. 109). In another NEA
nationwide survey, teachers were asked "If you could make one change
that you think would improve your own morale or professional
satisfaction as a teacher, what would that change be?" The largest
specific response was received was to "lower class size" "Teacher
Opinion Poll," (Today's Education. 1975. p. 14).
Harup studied large class size and concluded that
[It] seems to be among the most potent factors that shape the
teachers attitude towards his job.... If our data have any force,
then the large class does more to destroy the confidence of the
teacher than anything else. (1959, p. 56)
Lipskys work in 1980 concluded that large classes as well as other
factors cause teachers to compromise their ideals and sense of efficacy
(Ashton & Webb, 1986, p. 42).
A study of a single school in Illinois presented teacher satisfaction as
one of the benefits of reducing class size from 25 to fifteen. Teachers
felt they spent less time disciplining students, were better able to
integrate language arts and reading, had fewer parent conferences to
discuss problems, and that student self esteem increased (Hawkinson,
Filby et al. (1980) found that with smaller classes teachers felt they
were able to individualize more for students and help them more quickly
when they had difficulties. When they had large classes, they felt their
workload was heavy. With smaller classes, teachers felt less stress and
frustration. They also believed the climate of the class was more positive
and they felt more enthusiastic about their jobs.
Although primarily designed as a class size/student achievement
study, the Project STAR study in Tennessee incorporated effects on
teachers, such as use of planning time, staff development for teachers,
use of teacher aides, varying levels of teacher experience, and "the
differential effects of small classes on students from various
socioeconomic backgrounds" (Wood et al., 1990, p. 4). Johnston (1989)
reported on the effects of work related problems that emerged from the
The study found that all first grade teachers, regardless of class size
or the addition of a full-time aide, related that problems related to time
were more frequent and bothersome than other types of problems.
Efficacy is a concept which has been linked to increased student
achievement. Ashton and Webbs study supports the hypothesis that11.
.. teachers sense of efficacy is related to student achievement... [and
that]... teachers' efficacy attitudes are situation-specific." (Ashton &
Webb, 1986, pp. 3-4). Fuller et al. (1982) offer research that efficacy is
significant when attempting to improve either individual or organizational
performance. They concluded,
... individual teacher priorities, though often distinct from the
interventions goals and priority implementation tasks, may be
successfully adapted to the organizational change approach.
Research has shown that people actively construct their own reality
guided by experiences (Wegner & Vallacher, 1977). Because people do
this unconsciously and not in any apparent manner, their conceptions of
people and their world are implicit rather than explicit. However,
psychologists have observed this constructed reality "is not simply a
conglomerate of disconnected thoughts and perceptions but instead is
an orderly and structured system" (Wegner & Vallacher, 1977, p. 4).
Implicit psychologists take the philosophical stance that,
... every person is a scientist.... [People] are scientists in the
sense that they approach the world through a process of reality
construction that is analogous to the theory construction process by
which the scientist attempts to understand the world. (Wegner &
Vallacher, 1977, pp. 8-9)
Argyris and Schon (1974) support this view when discussing their
theories of action.
... Everyone is his own psychologist, sociologist, anthropologist,
engineer, and physicist.... In the course of intellectual history, much
formal academic knowledge has emerged through making explicit the
informal knowledge of everyday life. (p. 8)
People are constantly interacting with one another. These
interactions affect their construction of reality by causing them to change
their view of reality, by causing them to build detailed sets of information
about others, and by helping them understand how others think. "Acting
is testing, and the practitioner is an experimenter" (Argyris & Schon,
1974, p. 159). All of this data collection is an attempt by the person to
control his world and predict the future. Cognitive structure is both a
storehouse of past memories and also a guide to the future.
People form hypotheses about themselves and others based on past
experiences and gradually construct theories about their world.
Theories are vehicles for explanation, prediction, or control. An
explanatory theory explains events by setting forth propositions from
which these events may be inferred, a predictive theory sets forth
propositions from which inferences about future events may be
made, and a theory of control describes the conditions under which
events of a certain kind may be made to occur. (Argyris & Schon,
1974, p. 5)
At times one cannot test a theory, but a person "may well go on
believing in the theory despite the fact it cannot be tested" (Wegner &
Vallacher, 1977, p.21).
Further, noted psychologists such as Piaget and Kelly
have stressed the cyclic nature of the thought process, suggesting
that there is a continuous cycle of interaction among our cognitive
structures, our actions, and our experiences. (Wegner & Vallacher,
1977, p. 290)
Argyris and Schon (1974) advocate that professionals as well as
scientists communicate with one another and consider what they have
learned from experience (p. 144). As they state,
The professionals knowledge of these structured environments, his
certification to practice in them, his ability to understand the language
spoken in them and to negotiate in them constitute a great part of his
technical expertise and authority in relation to laymen, (p. 150)
If we consider teachers to be "scientists." as Wegner and Vallacher
and Argyris and Schon contend, then teachers are the people to ask
about class size and its effect on their sense of efficacy. Further,
teachers may well have constructed an implicit theory that large classes
limit their ability to teach students effectively. For these teachers, large
classes pose a threat to preferred instructional methods, including
individualization strategies, and their closeness to students.
Theoretically, large classes impede efficacy.
Although there is continuing disagreement over the effects of class
size on student achievement, several themes run through the literature.
Smaller classes may result in more teacher-student interaction as well as
increase teacher-student bonding. Economically disadvantaged and
lower ability students appear to produce greater achievement gains in
small classes. Very small classes or one-on-one tutoring produce the
highest achievement gains. Finally, smaller classes seem especially
beneficial to students in the early primary grades, with the caveat that the
preponderance of research has been at that level. The effect on
teachers is that they are able to individualize instruction to a greater
extent and teacher morale and enthusiasm is higher.
The literature on class size is often contradictory as well as
complicated. Efficacy, too, has been shown to be complex, situation
specific, and a determining factor of persistence. It is rooted in
expectancy theory and linked to personal causation A conclusion can be
drawn, however, that although class size and teacher efficacy have not
been directly studied, there are at least cursory indications that the two
may be interdependent. There are indications from research that
teachers may feel better in a small class and their efficacy may be
influenced by class size.
By understanding how teachers think about and react to class size, a
deeper understanding of the class size-student achievement controversy
will be accomplished. Therefore, the focus of this study is on the effect
class size has on teacher efficacy. Additionally, it investigates the role
class size plays on teachers sense of self-efficacy and ascertains what
teachers mean by boundary conditions of class size. Further, it
examines how exceptional student characteristics affect teachers sense
of class size, what instructional/management methods teachers prefer to
use, how large classes affect teaching style, and how teachers define self
THE RESEARCH METHODS
This chapter describes the research methods used in the three
phases of the study. In addition to detailing the research processes, a
description of the research setting and the teachers interviewed is
presented. Instrumentation, data collection, and data analysis
procedures are discussed. The limitations of this research will conclude
Research Questions Guiding the Study
This study examined the relationship between class size and teacher
efficacy. It also investigated student characteristics which affected sense
of class size being too large, the concept of boundary conditions of
class size, and conditions which caused teacher to change preferred
instructional/management methods. The research questions that guided
the study were:
1. How do class size and teachers perceived sense of efficacy
2. What do teachers perceive as ideal class size?
3. How do student characteristics affect teachers sense of class
4. What are the interactions between class size and instructional/
management methods as perceived by teachers?
This study utilized both qualitative/interpretive and quantitative
research methods appropriate to each of the three phases of the study.
Because the researchers questions in the interview phase were
concerned with teachers perceptions of their ability to effectively teach
large classes of students, interpretive methods were used for the second
phase of the study. Erickson (1986) described interpretive research as
reflecting the participants viewpoint bounded by the researchers partially
preconceived parameters for the research already established. The final
phase, by contrast, employed a survey which was designed to be
analyzed using statistical procedures on a larger sample of teachers.
Description of the Research Setting
The districts ranged in size from approximately 200 to almost 3,500
students. The districts were diverse economically, geographically,
politically, and culturally. By way of example, three of the districts were
in ski resort communities while two others were primarily agricultural/
ranching communities. One community had depended on the mining
industry until about ten years ago when the mine closed. This same
community experienced a sudden economic decline from which it had
never fully recovered, although it was attempting to become a year-round
tourist attraction. As people left this district, one school had to be closed
due to lack of students and the number of inexperienced teachers was
reduced. Consequently, this district had the oldest, most experienced
faculty in the state.
Another school district was in an area that has been subject to
"boom or bust" since its inception over 100 years ago. It was primarily
agricultural, then "alternative fuels" were discovered in the region in the
early 1900s. This discovery brought an influx of people to the area, but,
when the bottom fell out of the energy market, as it had several times
during the past 60 years, they egressed. This pattern continued. The
district served as a "bedroom" community for a resort town nearby.
There were several new school buildings that were near capacity, and
some new staff members, although the majority of staff members
grew-up or had been in the area for a long time.
The faculties of the three ski resort area school districts tended to be
younger and had more advanced degrees than the faculties of the
agricultural/ranching districts. Reflective of the community members,
teachers tended to stay a few years in the area and then move
elsewhere due to the high cost of living. In general, these districts had
more money to spend on salaries and their constituents demanded
quality schools. Based on standardized test scores, low dropout rates,
and the number of students who went on to college, the quality of
education for all of these districts was generally high.
The school districts were not without their problems, which were
regularly documented in the local newspapers. The cost of living in the
resort communities was high; rental housing was extremely hard to find;
there was a great deal of stress on the people who commuted to the
resort communities to work in the service industries. Incidences of
spouse and child abuse were reported among these people; illegal drugs
were readily available in any of the communities; and the transience rate,
especially among the workers, was very high. It was not unusual for a
student to attend four or five different schools in vastly different
geographic areas during one school year.
Teachers, who were predominantly Anglo, had to contend with these
problems in addition to a widely diverse student population. Most of the
students were Anglo, however, many of the other students were Latino,
either recently from countries south of the border or from the small
mining towns where their families had lived for 150 or more years when
the railroads and the mines were developed. The students came from
families with varying degrees of affluence and parental expectations
Description of Interview and Survey Populations
The teachers who formed the population of this study worked in
selected school districts in Western Colorado serviced by an educational
organization whose primary purpose was to provide special educational
services for small districts (under 3,500 students). Small, rural districts
were selected because the teachers were more likely to experience small
class sizes in contrast to teachers working in the Front Range of
Colorado. Since these teachers experienced smaller than average class
sizes, their responses were enlightening regarding what teachers say and
what they do when they actually have smaller classes.
Initially, the entire population of first through third grade teachers
received the preliminary questionnaire. In the interview phase, a sample
of seven teachers was chosen. The final phase surveyed the entire
population of first through third grade teachers during the second year of
From the respondent profiles of the 38 teachers who agreed to be
interviewed in the preliminary phase, seven teachers were selected for
the in-depth interviews (Table 3.1).
Table 3.1. Respondent Profiles.
NAME SCHOOL EXPER. #3 #4 SIZE IDEAL
LB PK 16 F F 12 18
Joan GL 19 F/T T 23 <25
JEN A 13 F T 21 18-20
Sally S 8 F T 19 16
KB A 18 F T 21 20-22
KBR L 11 F T 16 16
DC NC 2 F T 19 15-18
BC SI 19 F T 20 18-20
Ann u 7 F T 16 24
JDEM c 26 F T 24 22
JD NC 7 F T 53* 20
JF B 3 F F 26 18-20
CF A 10 F T 19 <20
JF PK 2 F T 15 20-22
PF A 13 F T 19 15-20
LF L 14 F T 22 18
JG NC 13 F T 22 22
JGR S 20 F T 22 18
TH C .5 F T 22 20
KH S 13 F T 22 16
Joe w 22 F T 24 22
TJ F 17 F T 32* 17-20
Meg PK 3 F T 14 16-18
AK C 3 F T 22 18-20
LK NC 9 F T 22 22
DK L 17 F F 24 18
DM U 12 F T 28 20-22
KM SI 24 F T 27 18-20
LO L 23 F F 25 15-20
PP U 7 F F/T 20 17-20
CS WP 10 F F/T 21 15
FS DV 21 F T 24 15-20
SS S 15 T F/T 17 15
Table 3.1. (contd.)
NAME SCHOOL EXPER. #3 #4 SIZE
Beth c 2 F T 22 <20
Alta WP 28 F F 21 20
MT u 16 F F 17 18-20
LW BR 10 F T 23 <20
CW T~ _"n U .25 F i F 19 15-20
These interviews were analyzed to answer the research questions of
class size and the effect on teacher efficacy. The methods suggested by
Miles and Huberman (1984) were used to analyze the interview data.
The researcher proposed to conduct a "detailed probing of an instance
in question rather than mere surface description of a multitude of cases"
(Lincoln & Guba, 1985, p. 358).
Further, qualitative data are more likely to lead to "serendipitous
findings and to new theoretical integrations; they help researchers go
beyond initial preconceptions and frameworks" (Miles & Huberman, 1984,
p. 15). As efficacy beliefs are essentially attitudes, one of the best
sources of attitudes, values, and beliefs is self-report in the interview
(Sproull, 1988). An interview allows the opportunity to probe deeply into
peoples thoughts and/or beliefs. The interviews were then used to
focus and shape the survey questions. By way of example, because
teachers indicated during the interviews that factors such as exceptional
student characteristics influenced some of their perceptions, a section on
the survey attempted to probe that area.
Although surveys are a convenient, efficient way of collecting
information with a large sample, the questions may be limiting.
Information on actual experience with various class sizes and types of
students provide a richer description of the complex forces at work in the
classroom. For these reasons, the combination of interview and survey
was chosen for this study and the study was conducted in the classical
inductive to deductive manner.
Teacher One. Sallv
Sally had been teaching for eight years, three years in New
Hampshire and the rest in a resort community in Colorado. She taught
in a school that was built within the last ten years and the rooms were
large compared to some older elementary buildings. She shared the
area with another first grade teacher and they were able to partition the
area off for a more self-contained setting. Her room was alive with work
the students had created. In a central area of prominence gingerbread
houses which the students had constructed were displayed. She had
nineteen first graders at the time of this research. Her ideal class size
reported on the survey was sixteen.
Teacher Two. Joe
The second teacher, Joe, had taught a total of 22 years, two years in
California as a special education teacher, and the remainder in Colorado
as a special education and a regular classroom teacher. He had been in
this Colorado district nineteen years. At the time of the interview, he had
24 second grade students in a relatively new building that was close to
capacity, but indicated 22 was his ideal size on the initial survey. He has
had classes as high as 42 students.
Due to a change in the schedule because of holiday program
rehearsals, we met in his classroom during an art period. Rather than
having a break, he conducted the class instead of the art teacher. The
students were very involved with their projects, moving around the
classroom as necessary to get more materials, conferencing with other
students, or coming to Joe for help.
Teacher Three. Alta
The third teacher, Alta, had been teaching 28 years. All but two of
those years were in the same community. The school where she taught
was approximately twenty years old. The district was attempting to
provide materials and resources for the teachers, but the current
economic situation dictated only the bare necessities. She began
teaching fifth grade, her least favorite grade, but since then had taught
first, second, and third grades. In contrast to the other teachers rooms,
this classroom was organized in a traditional, orderly pattern
with no learning centers and more teacher-designed bulletin boards than
displays of student work. At the time of the interview, she had 21 third
grade students, although she had had classes with as many as 30
students. Her stated ideal size was 20 students on the initial survey.
Teacher Four. Beth
The fourth teacher, Beth, who was in her mid-thirties, had only been
teaching two years. She worked at home as bookkeeper for her
husbands construction company and cared for their two children. Now
that her children were in school and the family was financially secure, she
stated she had returned to her first love, teaching. She substituted for a
year before becoming a full time second grade teacher in this rapidly
growing district. Her classroom was decorated with the students work
as well as with commercial posters in English and Spanish. Everywhere
there was evidence that this was an active class. There were learning
centers and the rug in the center of the room showed evidence of recent
student activity. At the time of the interview, Beth had 22 students, but
thought that less that 20 was ideal.
Teacher Five. Mea
Meg had been teaching three years in a small community which
relied on ranching and mining for its income. As both of these markets
were depressed, times were hard economically, and many families had
left the area. Her school was about two decades old, and although it
was in good repair, the classrooms were not designed for the types of
programs now popular. Nevertheless, there were signs of contemporary
teaching techniques such as learning and publishing centers, displays of
students work, and posters. At the time of this interview, Meg had
fourteen third grade students, but expressed sixteen to eighteen as being
her ideal class size on the initial survey.
Teacher Six. Joan
Joan taught in a large (1,000 students), old elementary school
building which had recently been repaired and updated to accommodate
the changing style of elementary teaching. Teachers were beginning to
work in teams and integrate their programs to a greater extent than they
had in this conservative, yet forward-looking district. Joans room was
small by current standards, but brightly decorated with bulletin boards
depicting student work. Joan had taught for nineteen years, all but two
of them in her current district. At the time of the interview, she had 23
second grade students and her ideal class size, according to the survey,
was less than 25.
Teacher Seven. Ann
Ann taught in a relatively new building in a small district which was
experiencing economic hardships. Funds were limited but her room
reflected current instructional techniques and contained many displays of
student work. Ann had taught seven years, four in her current district,
three in Denver, and had substituted on a long term basis there. Her
school was small and the staff appeared to be very cohesive. At the
time of this interview, she had sixteen third graders, but indicated 24 was
her ideal class size on the initial survey.
The survey was conducted in 1992, the second year of the study.
The survey population consisted of all the teachers in grades one
through three in nine school districts in western Colorado. Kindergarten
teachers were eliminated because of the generally unstructured nature of
Instruments were developed which were appropriate for each stage
of the study. The preliminary phase employed a seven-item
questionnaire, the interview portion used a modified interview guide, and
the final phase used a 57-item survey developed for this study.
Preliminary Phase Questionnaire
The questionnaire in the preliminary phase (Appendix A) was used to
develop a respondent profile. It ascertained current class size, what
number teachers considered to be ideal class size, years teaching
experience, and gender. The two Rand questions were included to
establish high or low efficacy.
1. When it comes right down to it, a teacher really cant do much
because most of a students motivation and performance depend
on his or her home environment.
2. If I really try hard, I can get through to even the most difficult or
Additionally, it requested the respondent to volunteer to be interviewed at
a later date.
The semi-structured interviews used a format that was modified from
the 1986 study conducted by Ashton and Webb (Appendix B). During it,
teachers were asked to describe their teaching career, talk about their
school and the teachers they worked with, and state their sense of their
current class size. Teachers compared their current perception of
efficacy to when they first started teaching. They were asked what made
them feel efficacious. Finally, they were asked if class size affected their
perception of efficacy, whether they used different instructional/
management techniques with various class sizes, and what student
characteristics made it hard for them to teach. Although the same
questions were asked of all teachers, they were encouraged to add their
own thoughts and expand upon the guide.
The survey phase, conducted in 1992, the second year of the study,
was the outgrowth of the semi-structured interviews of seven primary
grade teachers. During the interview process, teachers stressed that
issues generated by class size were not only a by-product of numbers of
students, but also types of students in classes. They demonstrated a
concern for affective growth of children as well as academic growth.
They indicated they could accomplish their goals more effectively in small
classes. The survey represented an efficient way to get a large number
of responses from the population of teachers. Although the researcher
knew who the respondents were, the survey assured greater
confidentiality to the teachers than did the interviews because the
teachers were not directly interviewed, thus preserving their anonymity to
a greater degree.
The questions on this survey elicited demographic information,
established degree of efficacy, and finally focused on teacher
conceptions of class size being too large or too small, student
characteristics, and instructional/management methods. These
components were then analyzed for interaction using the statistical
procedures of Pearson Product-Moment Correlation Coefficient, partial
correlation, multiple regression, factor analysis, and Analysis of Variance
(ANOVA). The statistical criterion of p = .10 was set for all analyses
unless otherwise indicated. All data were presented as mean plus or
minus standard deviation unless otherwise noted.
As background information, demographic questions were included to
ascertain the number of years teaching experience and gender. Several
questions contained reversed scales to ensure that teachers were
reading each question carefully.
Teachers completed questions on Section A, which contained
demographic questions and questions relating to student characteristics
(Appendix C) with numbers they felt appropriate. The exception was
question six, which asked them to indicate on a scale of one to ten
whether their sense of their current class size was "too small" (one) or
"too large" (10). Sections B and C, which contained the Gibson and
Dembo efficacy scale (1984) and instructional/management questions,
used a six point scale. One (1) indicated the teacher strongly disagreed
and six (6) indicated the teacher strongly agreed with the statement.
The final Survey response was 70 percent, 115 returns of 164
sampled teachers in the survey phase. Initial distribution of the survey
resulted in a 59 percent (n = 64) return. One week after the survey was
distributed, a follow-up was conducted by telephone, which increased the
rate of return to 70 percent. It was determined that an additional
follow-up would not be pursued in order not to alienate people. A 70
percent response was deemed very acceptable, or "very good" (Babbie,
1983, p. 165).
As with any survey, there is bias due to nonresponse on the survey.
For mail surveys, bias due to nonresponse can be studied by
comparing those who respond immediately with those who respond
after follow-up steps are taken. One clear generalization that holds
up for most mail surveys is that people who have a particular interest
in the subject matter or the research itself are more likely to return
main questionnaires than those who are less interested. (Fowler,
1988, pp. 48-49)
The surveys returned after initial distribution were kept separate from
those returned after the follow-up telephone calls. A t-test compared
early responders versus follow-up responders was performed on all
variables. There was little difference in efficacy scores (Table 3.2).
Table 3.2. Table of Means, Early Versus Follow-up Respondents.
Mean of early respondents = 57.98 6.6
Mean of follow-up respondents = 56.06 9.6
(P = 23).
No significant differences between the groups were noted for the
majority of the variables. Significant differences were noted for the
following variables: Small classes can be too small for a diversity of
opinion (p = .008), change instructional methods to maintain discipline in
larger classes (p = .001), group heterogeneously for instruction in larger
classes (p = .001), number of students currently (p = .001), years
teaching experience (p = .029), and sense of class size ( p = .001).
Analysis of internal consistency reliabilities yielded Cronbachs alpha
coefficient of .70 for the efficacy variables. This coefficient corresponded
closely with research of Gibson and Dembo (1984) which yielded a
coefficient of .79.
Research Question One: Efficacy Questions
In the preliminary phase, the two Rand questions were used, however
interviews made it apparent that there were degrees of efficacy, which
those questions failed to discern. Therefore a more extensive instrument
was chosen for survey phase of the study. Section B surveyed efficacy
with questions one through sixteen from Gibson and Dembos (1984)
Efficacy Scale. This instrument ascertained the teachers overall degree
of efficacy and placed the teachers into one of three groups, high,
moderate, and low, based on efficacy scores.
Factors identified in the Gibson and Dembo (1984) study
corresponded to the previous Rand studies (Armor et al., 1976; Berman
& McLaughlin, 1977). Since both the Rand and the Gibson and Dembo
measures were developed independently with corresponding items,
some congruent validity was established (Rhodes, 1988).
Research Question Two: Boundary Conditions of Class Size
Four questions in section A were included to measure teachers
conceptions of ideal class size. As background information, question
one in section A (Appendix C) requested the number of students
currently enrolled in the teachers class, questions four and five asked for
the greatest and the fewest number of students the teacher had ever had
in a class.
Question six asked the teacher to judge whether his/her class was
large or small. The interviews had indicated that class size was a highly
individualized perception often based on previous classroom experience.
It was necessary to ascertain how the teachers surveyed felt.
During the interviews, several teachers indicated the necessity to
change from more individualized instructional methods to more direct
instructional methods as class size increased. Questions B24 and 25
related to both class size and preferred methods of instruction. These
questions asked them to specify a class size at which they changed
methods of instruction and how a class of fifteen versus a class of
twenty-five impacted instructional methods.
Research Question Three: Student Characteristics
Section A, questions seven through twenty asked teachers at what
class size students with exceptional characteristics could be added to
their classes made it difficult to teach. During the interviews, teachers
identified exceptional student characteristics such as students with
behavior problems, special needs (learning disabled), students from
dysfunctional families, or otherwise hard to teach students. Some of the
teachers interviewed suggested it depended on the type of student who
was added to the class and that a gifted students were perceived
differently than students with problems. Student characteristics
contributed to boundary conditions of class size (" Cause you could
have 20 and if fifteen of them are hard to work with, thats a lot harder
than the person up the hall who has 30 and has a lot of good students").
Question A21 asked teachers what percentage of students in their
current classes they considered hard to teach. Questions A22 and A23
asked them to state reasons for this percentage and what type of
student would cause them to change preferred instructional methods.
Section C was designed to elicit information about the composition of
teachers current classes. The six questions concerned characteristics of
current students. Teachers were asked if they had students with
behavior problems, unmotivated students, those with special educational
needs, or culturally diverse students. It was felt that this information
would provide a clearer picture of the types of students teachers actually
experienced in their classes.
Research Question Four: Instructional/Management Methods
During the interviews, the teachers repeatedly stated they preferred to
individualize instruction and cited several methods or programs they
used such as Readers and Writers Workshop, cooperative learning, and
working one-on-one with students. They also stated that classroom
management and organization was easier in small classes. Questions B
seventeen through 28 related to instructional methods and classroom
management. Teachers indicated they adjusted assignments or lessons
during the interviews and question nineteen addressed individualizing
Question B23 was developed when several teachers mentioned that
classes could be too small for diversity of ideas ("Its nice to have a few
more because you can get more interactions and ideas ... a few more
would be nice") during interviews. They felt that a "critical mass" of
students was necessary to keep the class going. This question was
included as a management issue.
Other management questions were B21 and B24 asked the teacher
to specify whether it took the teacher longer to physically manage
students and whether s/he varied instructional methods to maintain
discipline as classes got larger or not. During the interviews, several
teachers mentioned they varied instructional techniques because of
crowd control issues.
Questions B26 through B28 asked them how they grouped for
instruction as class size increased. During the interviews, some teachers
mentioned they preferred to group heterogeneously, whereas some
pulled students of similar abilities for special help. These questions were
designed to see if management strategies changed as class size
In summary, the survey contained items pertinent to this study,
namely, efficacy, class size, student characteristics, and instructional
/management methods. The survey and subsequent quantitative
analyses provided general availability of the instrument to all teachers,
the possibility of a larger response than was provided by the interviews, a
less-threatening nature of the survey, and greater consistency.
Data Collection: Preliminary Phase
Districts were contacted to obtain their consent and enlist their
support. Permission to distribute the anonymous questionnaire was
obtained from the superintendents and the principals of elementary
schools in the nine districts after obtaining consent to conduct the study
from the Human Research Committee. The questionnaires were coded
by building and distributed by the building principals after the researcher
determined the number of first through third grade teachers in each
building. Questionnaires were returned via district mail or through the
When the questionnaires were returned, those who volunteered to be
interviewed were separated from the others, after stamping the date on
all of them. A tally was kept as to how many questionnaires were
returned from each school. Responses from the 38 volunteers were
tabulated and and a profile was developed from which the seven
teachers were selected. Teachers who were not chosen received a letter
that thanked them for their willingness to participate and requested to
remain available in the event additional participants were needed.
The variety of districts chosen make the data more theoretically
relevant. As Glaser and Strauss (1967) noted,
Comparison groups provide ... control over the two scales of
generality: first, conceptual level, and second, population score.
Third, comparison groups also provide simultaneous maximization or
minimization of both the differences and the similarities of data that
bear on the categories being studied, (p. 55)
Of the 161 questionnaires distributed, 139 or 86 percent were
returned. This high return could suggest teachers have intense interest
in the topic as well as the principals and the teachers acquaintance with
the researcher. It was determined by the researcher that a follow-up
contact was not necessary because of the return rate. Of the 139
responses, 38 or 24 percent of the teachers volunteered to be
Data Collection: Interview Phase
Interviewing was chosen as one method of data collection for this
phase of the study because the researcher wanted to explore the
complexity of teachers sense about class size being too large or too
small and teacher efficacy. Patton (1980) stated,
The purpose of the interviewer is not to put things in someones mind
... but rather access the perspective of the person being
interviewed.... The assumption is that that perspective is
meaningful, knowable, and able to be made more explicit, (p. 196)
Interviews were held in the teachers classrooms. This familiar
atmosphere was reassuring to teachers and comfortable. After
thoroughly briefing participants on the objectives and procedures of the
project, teachers were asked to sign a consent form (Appendix D) giving
permission for their participation in the study. Teachers were asked
questions that elicited their perceptions of class size and its effect on
their feelings of efficacy (Appendix B). The questions were modeled after
Ashton and Webbs questions (1986, pp. 184-187). The questions were
phrased to see if efficacy was connected with class size in teachers'
minds and they were asked to expand on that linkage, if any. Similar
questions were posed to each teacher in order to compare responses.
Each interview generated additional questions pertinent to individual
teachers. The initial interviews were tape recorded, with permission, and
transcribed as soon as possible after the interview by the researcher on
a word processing program. Follow-up telephone interviews were
conducted as necessary to verify information or ask for clarification.
The opening questions called upon teachers to describe themselves
in order to put them at ease and gather demographic information.
Subsequent questions probed their perceptions about their their sense of
efficacy. These questions were non-threatening and evoked extensive
responses. Questions were also asked regarding the size of their
current class compared to previous class sizes and the effect class size
had on their teaching. Two questions regarding the use of different
teaching strategies for different class sizes and in-service training
experiences were posed because the literature indicated teachers
generally did not change teaching strategies unless or until they received
staff development (Cahen et al., 1983; Filby et alM 1980; McGiverin et al.,
1989; Robinson, 1990).
Questions required teachers to consider what student characteristics
affected their efficacy and which factor, class size or student
characteristics, weighed more heavily on their sense of efficacy.
Additionally, teachers were asked how their current senses of efficacy
compared to their initial teaching experiences and how it related to class
size. Finally, they were asked what made them feel efficacious.
Most of the questions were open-ended and all were non-judgmental.
Although questions three, four, and eight (see Appendix C) appear to be
dichotomous, allowing a yes or ho response, they, too, evoked
descriptive responses from the teachers. The teachers were eager to
discuss the questions at length, which the researcher encouraged.
Data Collection: Survey Phase
Two tests were conducted prior to the administration of the
59-question survey. First, a review panel of seven experts determined
readability and content validity. Second, a pilot study of nineteen
teachers was conducted and data were analyzed. Cronbachs alpha was
used to analyze internal consistency reliabilities. Appropriate changes
were made after each test.
For the survey phase of the study, conducted the second school
year, individual schools were telephoned to ascertain the number and the
names of the teachers in grades one through three. Surveys were then
distributed to 164 primary teachers by the United States Postal Service or
by district mail with stamped return envelopes included. Codes which
identified the teachers and their schools were placed on each survey to
facilitate nonresponse follow-up. Code lists and individual teachers
surveys were kept confidential. Nonrespondents were contacted by
telephone to encourage return of the completed survey.
As surveys were returned, they were date-stamped. The surveys
returned the first week after release were kept separate from the other
respondents who were telephoned and encouraged to return the
surveys. The surveys were tabulated and the data entered into the SPSS
program on the personal computer for analysis.
Analysis of Interview Data
Interpretive research leads from individual instances to the complete
picture of a situation. Data analysis followed the process outlined by
Miles and Huberman (1984) to analyze data. Data analysis occurred
during and after data collection and included data reduction, and data
display decisions, as well as conclusion drawing and verification. It was
the crucial process of winnowing the irrelevant from the relevant data, still
maintaining the subjects perspective.
Classifying and Coding Data
The categories established by the nature of the study prior to data
collection included class size, teacher efficacy, student characteristics,
instructional/management methods, and the effect of class size on
teacher efficacy. Each category helped the researcher organize
transcripted narrations into more meaningful and manageable "chunks" of
data (Miles & Huberman, 1984, p. 56).
As data accumulated, incidents applicable to each category from
different interviews were compared/contrasted and interpreted.
This constant comparison of the incidents very soon starts to
generate theoretical properties of the category. The analyst starts
thinking in terms of the full range of types or continua of the category,
its dimensions, the conditions under which it is pronounced or
minimized, its major consequences, its relation to other categories,
and its other properties. (Glaser & Strauss, 1967, p. 106)
Many methods of coding incidents exist, but Miles and Huberman
(1984) suggest keeping codes simple and using words rather than
numbers because they convey more meaning. For this reason, a Data
Sheet (Appendix E) was developed by this researcher. A combination of
color-coding and brief descriptive notations was developed. Initially,
color-coding allowed for easy comparisons of responses among the
participants and notes on the data sheets highlighted phrases indicative
of the categories, cross references to other participants, and memos
regarding follow-up questions. By way of example, Beths transcription
was color coded in pink highlighter and labeled for class size (cs). Her
response "Anything over 25 is exponential" could easily be compared to
Sallys similarly coded response "Even with just three more, its not as
Copies of the transcriptions were made with prenumbered lines which
facilitated easy access to original narrations. The codes were applied to
"units of analysis". "[T]he unit of analysis [is] a sentence or
multi-sentence chunk" (Miles & Huberman, 1984, p. 63). Transcripts
were cut into blocks and attached to note cards for storing and retrieval
as suggested by Levine (Miles & Huberman, 1984).
Following Miles and Hubermans advice, the data were reduced,
weighted, summarized, and selected excerpts included (1984).
In order to draw and verify conclusions, a matrix was designed for each
teacher and a summary matrix across all teachers. The matrix
presents] information in a compressed, ordered form, so that the
user can draw valid conclusions and take needed action. . Our
experience tells us that narrative text alone is an extremely weak and
cumbersome form of display. (Miles & Huberman, 1984, p. 79)
Lincoln and Guba (1985) coined the terms credibility, transferability,
dependability, and confirmability to match the more traditional and
quantitatively rooted terms internal validity, external validity, reliability, and
objectivity. This study employed many of the methods they suggested to
insure trustworthiness including the following:
1. To insure credibility (internal validity), the researcher used a form
of persistent observation, termed here persistent interviewing. There was
a focus on those aspects of each interview which were most relevant to
the research questions.
2. Trianoulation in the classical sense was limited by the fact that
only one researcher was working on this study. However, information
from the deliberately drawn sample of teachers served to cross-check
interpretations and conclusions therefore providing a form of
3. The researcher utilized member checks or "peer debriefers11
suggested by Lincoln and Guba (1985). that is, interviews and
interpretations were continuously reviewed by members of a group of
doctoral students who were conducting their own research using
methods similar to those employed in this study.
4. Transferability (external validity) was assured by means of
purposive sampling derived from the initial questionnaire, which was
intended to maximize the scope and range of information gathered.
From the respondents, a group of teachers with varying amounts of
experience, varying class sizes, and different schools was chosen.
Purposive sampling illuminated the contextual factors which had to be
considered before final conclusions could be reached.
5. Reliability, or a dependability audit, that was an audit trail, was
maintained by the researcher which could be examined by an external
auditor to determine whether the research process used was acceptable.
Transcriptions and the original tape recordings were available as were
the original questionnaires which contained coded names of schools and
names of teachers who self-selected participation.
6. Finally, with respect to confirmability (objectivity), a member check
provided an ongoing source of external review and served as an external
audit (Lincoln & Guba,1985). Throughout the research process, the
group met on a regular basis to consider the progress of this study. As
transcriptions became available, members were asked to read them and
comment on the researchers coding, categorizing, and conclusions.
Methods of presenting the material and suggestions for follow-up were
Additional Analysis of Interview Data
Spradley (1979) defines domains as large units of cultural meanings.
Upon analyzing the interviews, it became evident that underlying the
teachers conversations facets of their culture were being described.
Spradley further defines culture as "the acquired knowledge that people
use to interpret experience and generate social behavior" (p. 5). Much of
what people know about their culture is at the tacit level, therefore, the
researcher must make deductions based upon what s/he has seen or
heard (p. 188).
The interviews were re-analyzed using techniques suggested by Miles
and Huberman (1984), Van Maanen (1979), and Spradley (1979). This
in-depth analysis led to deeper understandings of the research
One of the strategies Spradley suggested using to understand cultural
meaning was asking contrast questions to get beneath the data to the
underlying theories that can be formed (Spradley, 1979). Similarly, Van
Maanen (1979) suggested identifying the distinction between operational
data, that which was observed, and presentational data, that which was
produced to maintain an appearance in order to draw and verify
conclusions. The researcher could often be misled if s/he did not detect
the difference between the two types of data and allowed these
"deceptions" to be reported as facts (p. 540).
Further, the informant might unknowingly lie or deceive.
That is not to say, however, that his deceptions, evasions,
conjectures, and so on are categorically disregarded by the
ethnographer. To the contrary, false and misleading information is
exceedingly valuable to the fieldworker when it is regarded as false...
. A central postulate of the ethnographic method is that people lie
about the things that matter the most to them. (Van Maanen, 1979,
This perspective was also supported by Miles and Huberman (1984)
who suggested that the researcher weight evidence to generate
... The idea that, regardless of the degree of trust a fieldworker may
feel has been developed, people in field sites nearly always have
some reasons for omitting, selecting, or distorting data, and may have
active reasons for deceiving the fieldworker (not to mention
themselves). If the fieldworker has actively entertained such a view of
particular respondents, and of a particular set of data from them ...
more confidence is justified, (p.236)
With these caveats in mind, the researcher examined the
transcriptions of the seven teachers and noted the discrepancy between
what some teachers said they did and what their other statements
indicated they really did.
Data Analysis of the Survey Instrument
The Statistical Package of the Social Sciences (SPSS) for the
Macintosh (1990) was used to compute all statistical analyses for this
phase of the study. Nonresponse bias between the early and the later
returns was determined by use of a t-test. Internal validity of the efficacy
portion of the survey was established by Cronbachs alpha coefficient.
A distribution of scores from the first sixteen questions in section B
(Appendix C) was used to establish three groups of teachers based on
efficacy. Similar to procedures employed by Gibson and Dembo (1984),
in this study teachers were grouped into high, moderate, and low efficacy
Once efficacy was determined and the teachers grouped, according
to differences in instructional/management methods, student
characteristics, and class size, exclusive of questions A22 and 23, which
provided anecdotal information, and were assessed using ANOVA. The
analysis of variance, ANOVA,
tests the hypothesis that the group means of the dependent variable
are equal. The dependent variable is interval level and one or more
categorical variables define the groups. (SPSS, 1990, p. 62)
Initial Data Analysis and Reduction of Survey Response
Initially, means and standard deviations were calculated for all of the
questions. The Pearson product-moment correlation coefficients (r) were
also calculated to determine the relationship between student
characteristics, instructional/management methods, sense of class size,
and demographic/baseline information.
Statistics, including means, standard deviations, and ranges were
calculated for all of the variables initially before grouping for efficacy
scoring. The responses in the survey which had been negatively
phrased to ensure careful reading by the teachers were recoded (1 = 6,
etc.). The variables in the table are arranged by demographics, including
class size, student characteristics, efficacy, instructional/management
methods, and teacher perception of student characteristics in their
current class. The order of the variables presented corresponds to the
survey order (Appendix C). As indicated in the instrumentation section,
class size used a scale from one to ten. From MATHCON (math
concepts) throughout the rest of the survey, the numerical scale was
from one to six.
After correlations were performed, teachers were assigned to one of
three efficacy groups (Table 3.3), high, moderate, or low, based on their
total efficacy score. This assignment was similar to the methods used by
Gibson and Dembo "in an attempt to address both factors [personal
teaching efficacy and teaching efficacy] simultaneously" (1984, p.572).
Table 3.3. EFFICACY Groupings.
Value Frequency Percent Cum %
36.00 1 .9 .9
37.00 1 .9 1.7
39.00 3 2.6 4.3
40.00 1 .9 5.2
42.00 3 2.6 7.8
Low 43.00 2 1.7 9.6
45.00 2 1.7 11.3
47.00 1 .9 12.2
48.00 1 .9 13.0
50.00 4 3.5 16.5
51.00 3 2.6 19.1
52.00 2 1.7 20.9
53.00 6 5.2 26.1
54.00 7 6.1 32.2
55.00 9 7.8 40.0
56.00 4 3.5 43.5
57.00 5 4.3 47.8
Moderate 58.00 7 6.1 53.9
59.00 10 8.7 62.6
60.00 4 3.5 66.1
61.00 3 2.6 68.7
62.00 3 2.6 71.3
63.00 8 7.0 78.3
64.00 7 6.1 84.3
65.00 4 3.5 87.8
High 66.00 2 1.7 89.6
67.00 3 2.6 92.2
68.00 1 .9 93.0
69.00 2 1.7 94.8
70.00 2 1.7 96.5
71.00 1 .9 97.4
72.00 2 1.7 99.1
75.00 1 .9 100.0
Total 115 100.0