J“.

BIASED TEACHER REFERRALS AND THEIR EFFECTS
UPON SCHOOL PSYCHOLOGISTS’ DIAGNOSTIC
' JUDGMENTS AND EXPECTANOIES

Dissertation for the Degree of Ph D
MICHIGAN STATE UNIVERSITY
DARRELL D ELDERS

1977 ' '

 

' “V ' ' " ‘ ' ' ‘ ‘ ”0“ 'Q.‘ ‘
i— 7 ‘.'t.:o<.‘; “a." ' ‘L * 5‘ szJYS w ‘ ‘ a a..’
U
s Q
‘7 v4

$.12: y 1; I‘- -L v, «v r N E.
. A; L L ~ A L

Pfigphln°flflf.‘gi E

Ullé NIVAD L ‘I/ 6"”;

 
 
  

 

This is to certify that the

thesis entitled

BIASED TEACHER REFERRALS AND THEIR EFFECTS
UPON SCHOOL PSYCHOLOGISTS' DIAGNOSTIC
JUDGMENTS AND EXPECTANCIES

presented by

DARRELL D . ELDERS

has been accepted towards fulfillment
Of the requirements for

PH. D. degree m EDUCATIONAL PSYCHOLOGY
(SCHOOL PSYCHOLOGY)

A [ C
' 'L “:52 \I

I V I l
/ MAiOr/érofessor O

 

 

 

Date ~;//(7//’7 I7

0-7639

 

at; M . $-

Aium ‘— “S‘H'fffléj/‘l

Wt?”

. A”? Lillipui

 

 

 

ABSTRACT

BIASED TEACHER REFERRALS AND THEIR EFFECTS
UPON SCHOOL PSYCHOLOGISTS' DIAGNOSTIC
JUDGMENTS AND EXPECTANCIES

BY

Darrell D. Elders

Two experiments were conducted to test the effects
of biased teacher referrals upon diagnostic judgments and
treatment recommendations of Michigan School Psychologists.
School psychologists were exposed to one of three types
of teacher referral reports, each describing the same
child, but differing in the language used for this
description. One referral described the child in nega-
tive terms such as "hyperactive, aggressive, and impul-
sive"; while another, positively oriented referral,
described him in terms like "energetic, assertive, and
spontaneous." A third behaviorally oriented referral
reported only the child's observable behavior. The sub-
jects studied identical information packets which contained
the history and test protocols of an active six-year-old
boy in first grade. All subjects viewed a video—taped
recording of the boy in his classroom. Subjects rated

the child's functioning on nine continua designed to

Darrell D. Elders

measure pathological severity. They also selected diagnos-
tic labels from two lists taken from the Michigan "Man-
datory Special Education Act" and clinical diagnostic
classification systems, each of which included a psycho-
logically healthy category. The subjects also selected

a treatment from a list of special education interventions
ranging from no special help at all to full-time placement
in special education classrooms. Several independent
variables were measured including the subjects' theoreti-
cal orientation (psychodynamic versus behavioral), length
of professional experience, and type of graduate training
(school psychology versus clinical psychology).

Four hypotheses were advanced, each predicting
that the negatively biased teacher referral report would
influence school psychologist subjects to over—estimate
the child's patholoqy on one of four dependent measures:
the "Severity of Pathology" scale, "Mandatory Special
Education" diagnosis, clinical diagnosis, and the special
education treatment the subjects decided to recommend.

The results of the main experiment, with 48 school
psychologists from eight Michigan school districts, failed
to show significant differences among the three types of
teacher referrals on any of the four dependent measures.
The same experiment, administered to a group of 12 school
psychologists all employed by a medium-sized city school

district, however, did result in significant (p < .05)

Darrell D. Elders

differences in the subjects' estimates of pathological
severity. The negative referral condition produced sig-
nificantly (p < .02) higher estimates of pathology than
the positive referral condition, but not higher than the
behaviorally oriented referral condition. This result
could have been a function of the particular behaviorally
oriented referral included in these experiments, or it
could have been an indication that school psychologists
experienced difficulty in evaluating behavioral data.
This trend could also have been an indication that "halo
effect" was slightly more powerful than negative expec-
tancy effect.

Further analysis showed that the subjects' length
of experience and type of graduate training were not
highly related to their responses on the four dependent
measures. These school psychologists' theoretical orien—
tation showed moderate overall correlations with the
dependent measures, but the values were lowest, not
highest, in the negative teacher referral condition.
Although these experiments were not specifically designed
to test for school district effects, the judgments of
school psychologists were significantly different among
the districts sampled (p < .007).

It was concluded that Michigan school psycholo-

gists are unlikely to be inappropriately influenced by

Darrell D. Elders

biased teacher referrals, at least in so far as their
judgments would substantially affect the life and school
experience of children. The effects of the school psycho—
logists' school district upon their diagnostic judgments
and treatment recommendations appeared to be worthy of
further study. Discovery of which variables, within a
school district, might account for these effects was
deemed to be particularly important. Further study of

the effects of behaviorally oriented pre—evaluation

information was also recommended.

BIASED TEACHER REFERRALS AND THEIR EFFECTS UPON
SCHOOL PSYCHOLOGISTS' DIAGNOSTIC JUDGMENTS

AND EXPECTANCIES

BY

Darrell D. Elders

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Counseling, Personnel Services, and
Educational Psychology

1977

DEDICATION

To Pat, my wife and friend, and to my children,

Randy, Barbie, Jeannie, and Becky, who have so lovingly

sustained and encouraged me.

ii

ACKNOWLEDGMENTS

The members of my committee have not only given me
the guidance and friendship necessary for my professional
development, but their influence has and will be felt in
my personal development as well. I am honored to
acknowledge them here with my gratitude.

--To Harvey Clarizio, my chairman and teacher,
whose training has changed my life, and whose
creative thought, knowledge of the field, and
personal example are evident throughout this
study.

—-To Robert Craig, who, despite his heavy respon-
sibilities and demanding schedule as department
chairman, unselfishly gave many hours of study
and guidance in the final preparation of this
study.

--To J. Edwin Keller, for his many helpful sug-
gestions in the planning stages of this study
and for his encouragement and warm friendship
during its completion.

--To William Hinds, whose ideas and knowledge were
especially helpful in the early stages of this

study.

iii

TABLE OF CONTENTS

Chapter
I. THE PROBLEM . . . . . . .
Need for the Study . . . .
Purpose of the Study . . .
Overview of the Study . . .

II. REVIEW OF THE LITERATURE. . .

III.

Teacher Expectation Studies . . . .

Experimenter Expectancy Effects. . .

Sex . . . . . . . . . . .
Prestige. . . . . . . . . .
Experimenter Experience. . . . .
Anxiety and Need for Approval. . .
Experimenter-Subject Acquaintance .
Task Variables. . . . . . . .
Mediating Variables . . . . . .

Psychometric and Clinical Experiments.

Intelligence Test Expectancy Studies
with Examiner-Examinee Interaction

Intelligence Test Expectancy Studies
without Examiner-Examinee Inter-
action . . . . . . . . .

Expectancy Effects in Personality
Testing . . . . . . .

Interview, Observational and Assessment

of Therapy Experiments . . . .
Labeling Bias in Clinical Studies .

Summary and Implications for this Study

METHODOLOGY 0 O O O O O O O O O

The Sample. . . . . . . . . .
Experimental Stimuli . . . . . .

iv

Page

12

13
24

26
27
27
28
29
30
31

33

35

40
47

52
55

61
66

66
68

Chapter Page

Teacher Referral Report . . . . . . 70
Background Information Sheet . . . . 71
Revised Wechsler Intelligence Scale
for Children Protocol . . . . . . 72
Bender Visual Motor Gestalt Test
Protocol . . . . . . . . . . 72
Kinetic Family Drawing . . . . . . 73
Wide Range Achievement Test Protocol. . 73
Video-Taped Behavior Observation . . . 74
Dependent and Independent Measures . . . 75
Severity of Pathology Scale. . . . . 75
Diagnostic Classification . . . . . 77
Degree of Confidence Scale . . . . . 78
Special Education Treatment Recommen-
dation . . . . . . . . . . . 78
Referral for Pediatric Determination
of Medication Possibilities . . . . 79
Theoretical Orientation . . . . . . 79
Experience and Training . . . . . . 80
Procedures . . . . . . . . . . . 80
Sampling Procedure. . . . . . . . 80
Experimental Procedure . . . . . . 81
Experts' Judgments. . . . . . . . 84
Hypotheses . . . . . . . . . . 86
Design and Analyses . . . . . . . . 89
IV. ANALYSIS AND DISCUSSION OF RESULTS. . . . 93
Analyses for Hypotheses. . . . . . . 93
Analyses of Relationships and Associ-
ations. . . . . . . . . . . . 95
Length of Experience . . . . . . . 99
Theoretical Orientation . . . . . . 101
Confidence . . . . . . . 102
"Severity of Pathology" . . . . . . 103
Special Education Treatment Recom-
mendation . . . . . . . . . . 104

"Classical" Diagnosis, "Mandatory"
Diagnosis and Referral for

Medication. . . . . . . 105
Graduate Training in SchoOl Versus
Clinical Psychology. . . . . . . 107

Chapter

School District . . . . . . .

Grand Rapids Study. . . . .
Summary and Discussion of Results. .

V. SUMMARY, CONCLUSIONS, AND IMPLICATIONS.
Summary . . . . . . . . . .
Conclusions . . . . . . . . .
Implications for Practice and Future

Research . . . . . . . . .
APPENDICES
APPENDIX

A. TEACHER REFERRAL REPORTS . . . . .

B. BACKGROUND INFORMATION SHEET . . . .

C. WECHSLER INTELLIGENCE SCALE FOR
CHILDREN-REVISED PROTOCOL . . . .

D. BENDER VISUAL MOTOR GESTALT TEST PRO-
TOCOL O O O O O O O O O O C

E. KINETIC FAMILY DRAWING . . . . . .

F. WIDE RANGE ACHIEVEMENT TEST PROTOCOL .

G. SEVERITY OF PATHOLOGY SCALE . . . .

H. RATING SHEET: DIAGNOSES, CONFIDENCE,
SPECIAL EDUCATION TREATMENT
RECOMMENDATION . . . . . . . .

I. RATING SHEET: REFERRAL FOR MEDICATION,
THEORETICAL ORIENTATION, TRAINING
EXPERIENCE O O O O O O O O 0

J. RAW DATA FREQUENCY DISTRIBUTIONS. . .

K. RANDOM GROUP TESTS OF ASSOCIATION . .

REFERENCES I O O O O O O O O O O 0

vi

Page

109

111
117

132

132
137

142

149

152

153

159
161
162

163

164

165
166
168

172

Table

3.1

4.1

LIST OF TABLES

Sample Characteristics . . . . . . .

Random Group Univariate Analysis of
variance I O I O O O U C O O 0

Random Group Cell Means and Standard
Deviations . . . . . . . . . .

Random Group Correlations, Associations,
and Significance Levels with all Treat-
ment Groups Combined . . . . . . .

Random Group Correlations, Associations,
and Significance Levels within the Nega-
tive Treatment Condition. . . . . .

Random Group Correlations, Associations,
and Significance Levels within the
Positive Treatment Condition . . . .

Random Group Correlations, Associations,
and Significance Levels within the
Behaviorally Oriented Teacher
Referral Condition . . . . . . .

Analyses of Variance for Graduate Training
School District Groups Analyses . . . .

Grand Rapids Group Univariate Analyses of
Variance . . . . . . . . . . .

Grand Rapids Group Cell Means and Standard
Deviations . . . . . . . . . .

Grand Rapids Group Correlations, Associa-
tions, and Significance Levels with all
Treatment Groups Combined . . . . .

Grand Rapids Group Fisher's Exact Tests with

all Treatment Groups Combined . . . .

vii

Page

69

94

95

96

97

98

100
108

110

112

113

114

115

Table Page
J-l Raw Data Frequency Distributions . . . . . 166
K-l Random Group Chi Square Tests of Associa-

tion with all Treatment Conditions

Combined . . . . . . . . . . . . 168

K—2 Random Group Fisher's Exact Tests Within
the Negative Treatment Condition . . . . 169

K-3 Random Group Fisher's Exact Tests Within
the Positive Treatment Condition . . . . 170

K—4 Random Group Fisher's Exact Tests Within

the Behaviorally Oriented Teacher
Referral Condition . . . . . . . . . 171

viii

LIST OF FIGURES

Figure Page

3.1 Experimental Design . . . . . . . . . 90

ix

CHAPTER I

THE PROBLEM

Since its inception as a basic diagnostic service
for the purpose of determining levels of mental retardation
in school children, school psychology has expanded and
refined its expertise to a level of considerable sophisti-
cation. Special education remedial programs have con-
currently broadened and specialized so that both diagnosis
and treatment are available to more children than ever
before. As of 1970, The Bureau of Education for the
Handicapped reported over 1,700,000 children being served
in special education programs for the retarded, emotionally
handicapped, visual and hearing impaired, and learning
disabled. Continued expansion is assured by the estimated
prevalence of these disorders in over four million school
age children (Hobbs, 1975, pp. 50-79).

Along with increased numbers of children coming
under psychological scrutiny has come growing professional
and lay concern with the issue of labeling and classifi-
cation. It is feared that labels can have a stigmatizing

effect upon their recipients. It is commonly believed

that labels tend to produce expectancy sets on the part of
those dealing with the child so that areas of opportunity
can become limited or closed. Perhaps the most well-known
experiment purporting to demonstrate expectancy effect

was reported by Rosenthal and Jacobson (1966) in which
teachers were led to expect that randomly labeled "intel-
lectually blooming" students would increase in their
ability. The authors' results showed that these same
students made more gains on post IQ tests than students
not so labeled. Many studies have demonstrated that
labels such as mentally retarded are associated with low
expectations and negative stereotypes and attitudes
(Boekel, 1969; Guskin, 1963a and b; Hollinger & Jones,
1970; Jones, 1972; Meyers, Sitkei, & Watts, 1966; Shotel,
Iano, & McGettigan, 1972). These stereotypes can be so
powerful for teachers that when children exceed the expec-
tations set for them, they are often seen as having
problems with social adjustment (Rosenthal & Jacobson,
1968). Expectancy effects can be quite resistant to
educational measures taken to extinguish them. Foster,
Ysseldyke, and Reese (1975) found that even when their
sample of graduate and undergraduate students in special
education were exposed to the work of Rosenthal and
JacObson, as well as other expectancy and bias studies,
they continued to hold negative, stereotyped expectations

for the children they observed.

The Rosenthal-Jacobson findings have been called
into serious question on methodological grounds (Elashoff
& Snow, 1971; Snow, 1969; Thorndike, 1968) and for failures
to produce like results in subsequent replications (Clai-
born, 1969; Fleming & Anttonen, 1971a and b; Jose & Cody,
1971). As will be shown later, however, there remains
enough evidence in the area of negative teacher expecta-
tions to create apprehension among most educators, includ-
ing school psychologists. In fact, many school psycholo-
gists, fearing expectation effects, attach statements such
as "he shows higher intellectual potential," or "this
should be considered a minimal estimate of her ability"
to their evaluation reports of lower ability children.

Although the practice of labeling is a source of
concern, experience confirms that it continues to exist.
In fact, diagnostic labeling has become more prevalent
than ever before. Part of the reason for the ubiquity of
labels most certainly accrues from the fact that they are
an efficient means of transmitting and conceptualizing
large amounts of information. Indeed, much of the infor-
mation accumulated regarding handicapped children has
been organized around such categories. Even more impor-
tant, however, is the fact that needed help often cannot
be provided children without the application of a label
of some kind. The federal government has appropriated

funds for the education of handicapped children (Education

for all Handicapped Children Act, Public Law 94-142, 1975),
but requires that such children be formally labeled. This
law defined handicapped children as

. . . mentally retarded, hard of hearing, deaf,

speech impaired, visually impaired, or otherwise

health impaired children, or children with specific

learning disabilities who by reason thereof require

special education and related services. (Section 602,

paragraph 1)
In Michigan, the Mandatory Special Education Act (Public
Act 198, 1971) not only provides funds for the education
of handicapped children, but makes it mandatory that all
handicapped children in the state receive a diagnosis and
special education services. Since federal law (P.L. 94-
142) requires the states to implement a state-wide plan
for the delivery of special education services, it is
likely that all states will enact legislation similar to
Michigan's within the next few years. Indeed, several
states have already done so. As a consequence, the prac—
tice of diagnostic labeling is almost certain to continue
for some time, at least as it is practiced in the schools.

Given the fact that one will apparently live with

the possible negative consequences of labeling in order
to obtain needed help for exceptional children, the
occurrence of misdiagnosis or inappropriate labeling is
especially abhorrent. To suffer the stigma attached to
an accurate diagnosis is one thing, but to endure the

stigma attached to misdiagnosis is quite another.

Increased quality and quantity of professional training

have been the preferred approaches to a cure for the
problem of misdiagnosis. Unfortunately, even the best
diagnostician is only as good as the information he
receives. Furthermore, it is possible that the diagnos—
tician is just as susceptible to expectancy set produced
by pathological labels as are the consumers of the

diagnoses he makes.

Need for the Study

 

Diagnostic error resulting from biased referral
information can be a problem in psychology and psychiatry.
Although several hundred studies have been reported in
the general area of expectancy set or bias, the literature
is quite silent with regard to clinical diagnostic error
resulting from bias in psychological or psychiatric evalu-
ations. If one were to delete those studies which have
used graduate school students as subjects and those studies
which have focused upon intelligence or personality test
scoring as the dependent variable, published data in this
area of interest would be very sparse indeed.

In school psychology, published expectancy studies
are confined completely to test scoring and examiner-
examinee test behavior experiments. Despite the fact
that school children are now probably the largest single
group of recipients of psychological services in the
United States, they receive these serviCes from a group

of professionals who have been given little attention

in the experimental literature and about whom little is
known. The judgments of school psychologists can have a
profound effect upon school children throughout their
educational careers and the evidence is that ever-increas-
ing numbers of children are exposed to these judgments.
The possible effects of bias upon these judgments is
therefore of critical concern.

In the mental health fields diagnosis has generally
focused on the client‘s negative characteristics, or
deviant behavior, rather than emphasizing the individual's
positive characteristics. The clinician practicing with
the illness orientation, then, tends to diagnose pathology
when he is in doubt, rather than risk overlooking a pos-
sible disease condition which may be harmful to his
patient. Despite various attacks made upon the medical
model (Sarbin, 1967; Szasz, 1961), school psychologists,
no less than other mental health professionals, have been
trained with materials espousing this same orientation

and thus continue to focus most of their attention upon

pathology. Underlying most diagnoses, then, there appears
to be a general bias toward pathology.

In addition to this general pathological bias,
school psychologists are exposed to certain unique
pressures which may serve to increase the strength of
this bias. With the proliferation of special education

programs and the need to have full classrooms in order to

qualify for reimbursing funds, school psychologists are
under frequent pressure from their administrations to fill
these special classrooms quickly. Classroom teachers have
also been known to exert pressure upon school psychologists
to diagnose handicapping conditions so that troublesome
children can be removed from their classrooms. Mandatory
Special Education legislation in Michigan and some other
states has also alerted many parents to the variety of
special help possibilities for their children and these
laws have allowed parents the vehicle to exert considerable
pressure upon special education administrators and school
psychologists to obtain this help, even for mild learning
problems. The area of learning disabilities has been
especially problematic in this regard. Parents of children
with learning disabilities have formed powerful and vocal
groups with influence at all levels of school administra-
tion. They are well informed with regard to schools'
legal responsibilities and are not hesitant to press for
special services. This pressure, coupled with the nebulous
quality of the handicap, has often produced situations in
which it is easier for the school psychologist to diagnose
learning disability than to avoid the label.

The referral agent for most school psychological
evaluations is the classroom teacher. It is the teacher
who makes the first judgments concerning a child's

behavior, and who often transmits these to the school

psychologist within the referral. The psychological evalu-
ation is not an isolated activity, as Towbin (1960) points
out; it consists of a triadic group of individuals, the
patient, the referring person, and the examiner, who
mutually influence each other. The referral is made
because the referring person expects that the psychological
evaluation will serve his own purpose in some way. Within
the bias and expectancy literature, appallingly little has
been published concerning the influence the classroom
teacher has upon the school psychologist's diagnostic

judgments.

Purpose of the Study

 

The purpose of this study was to add knowledge
to existing and somewhat meager information concerning
the effects of biased referrals upon the psychological
evaluation of children. Specifically, this experiment
focused upon judgments made by school psychologists rela-
tive to the diagnosis of learning disabilities in children.
The study was designed to enhance knowledge in this area
by going a step further than previous studies which have
measured differences in test scoring as a dependent
variable to bias induction. It could be argued that
even moderately large deviations in test scores resulting
from bias may not be enough to influence a child's final
diagnosis and what will subsequently happen to him. Con-

sequently, this study attended primarily to school

psychologists' judgments concerning diagnostic classifi—
cation, severity of the pathology, and special education
treatment as dependent variables with the test scoring
element held constant.

To add to the generalizability of the findings,
this study used actual practicing school psychologists
as subjects, rather than students in testing courses as
has been done in many previous experiments. The reactions
of these subjects were studied in relation to their theo-
retical orientation and length of professional experience.

The effects of biased teacher referrals in this
experiment were studied in the context of evaluating a
possible learning disabled youngster. Attention was paid
to all aspects of his functioning, not just his intelli-
gence. It is hoped that the results of this study will,
therefore, have broad applicability and that its variables
will have appropriateness to the current practice of
special education referral and school psychology.

This study was designed to test the hypothesis
that positively and negatively biased teacher referrals
differentially affect judgments made by school psycholo—
gists in their evaluation of a possible learning disabled
boy. It was expected that these effects could be observed
in school psychologists' willingness to label the child
with a pathological classification, their estimate of

the severity of this pathology, and, as a consequence,

10

their judgment concerning the amount of special education
intervention necessary to remediate or manage the problem.
Formal statements of four specific hypotheses are pre-
sented in Chapter III.

Further, since any school psychologist brings to
his work an orientation toward diagnosis that is dependent
in part upon his training and experience, it might be
expected that these variables are related to his product:
judgments concerning adjustment, diagnosis, and necessary
treatment. Some evidence has suggested that theoretical
orientation and length of experience may be related to
biasability. Behaviorally oriented psychologists may be
less susceptible to bias than their psychodynamically
oriented colleagues (Langer & Abelson, 1974). Less
experienced clinicians may also be influenced somewhat
less by pre-diagnostic information than experienced
workers in clinical situations (Temerlin, 1968). It was
anticipated, therefore, that the experience and theoreti-
cal orientation of this study's subjects would show sig-
nificant positive correlations and associations with their
diagnosis, their estimates of pathological severity, and
their recommendations of special education intervention
when they were exposed to negatively biased information
concerning the subject of the evaluation. Although this
study was not designed to provide an experimental test of

these expectations, data were collected from the school

11

psychologists sampled which allowed for an exploration of
relationships between the above variables. These relation-
ships, while being descriptive of the sample only, were
analyzed for the purpose of determining whether the

variables would merit further study.

Overview of the Study

 

Chapter II, next, will begin with a discussion of
the types of studies previously done in the general area
of expectancy. Several hundred experiments have been
carried out in this general area, many of which are only
tangentially related to the variables of interest in this
experiment. For this reason, only those studies which
have a rather direct bearing upon the process of diagnosis
as it relates to expectancy and bias will be reported and
analyzed in depth.

The experimental design, including sample descrip-
tion and sampling method, instruments, procedure, statis-
tical hypotheses, and plan for statistical analysis will
be presented in Chapter III. The statistical analysis
and discussion of the results of the experiment will be
the focus of Chapter IV. Lastly, Chapter V will contain
a summary of the study and will present the reader with
a discussion of conclusions and implications for practice.

Suggestions for future research will also be offered.

CHAPTER II

REVIEW OF THE LITERATURE

The experimental literature pertaining to bias
and expectancy is truly voluminous. In his review of
the literature on expectancy effects, Rosenthal (1976)
observed that there were well over 300 experiments
reported in the field. Most of these studies were con-
cerned with the effects of expectations in general exper-
imental situations and were focused primarily upon dif-
ferential subject behavior as a result of experimentally
induced expectations given to their experimenters. Manip—
ulations in experimental situation, task, experimenter,
and subject variables comprised the subject of most of
these studies.

Besides the above array of literature, special
applications of bias and expectancy experiments have
built up a formidable volume of research in their own
right. Of particular relevance to this study are those
experiments reported in the field of education with
teacher expectations, and in the field of psychology

with examiner bias studies in testing situations. The

12

13

review of related literature in this chapter begins with

a summary of the major findings in the area of teacher
expectancy, together with citations of representative
experiments, and following, a discussion of some pertinent
knowledges gleaned from the broader area of experimenter
expectancy research. The chapter will conclude with a
more detailed review of reported experiments in the areas

of psychometric and clinical evaluation.

Teacher Expectation Studies

 

In 1966, Robert Rosenthal and Lenore Jacobson
published the results of an experiment in which 20% of
the children in each of 18 classrooms were randomly
reported to their teachers as showing unusual potential
for intellectual gains. This potential was said to be
shown on a test designed to predict academic "blooming,"
but was really the Flanagan Tests of General Ability
(1960), a nonverbal intelligence test. Eight months
later, these "blooming" students showed significantly
greater gains on the same IQ test, administered a second
time, than did the remaining children in the control
group. Although the study included children in grades
one through six, these expectancy effects were significant
only for the first two grades.

Two years later, Rosenthal and Jacobson (1968)

published a more complete discussion of the 1966 "Oak

14

School" experiment together with a presentation of sup-

porting research and implications. This book, Pygmalion

 

in the Classroom: Teacher Expectation and Pupils' Intel-

lectual Development, swiftly became one of the most con-

 

troversial works in educational literature and touched
off a flurry of subsequent teacher expectation experiments.
The Rosenthal and Jacobson (1968) findings have
been subjected to severe criticism on methodological
grounds. Thorndike (1968) attacked the "Pygmalion" study
on the basis of reported pre- and post-test scores. He
showed that many of these scores could not possibly have
occurred, and therefore the entire study is based upon an
analysis of faulty and valueless data. In his short but
scathing review, Thorndike (1968, p. 708) says of the
study, "Alas, it is so defective technically that one
can only regret that it ever got beyond the eyes of the
original investigators!" Jensen (1969) showed that
Rosenthal and Jacobson's use of the same IQ test for
pre- and post-measures, use of teacher—administered
tests, and use of the child as the unit of analysis
rather than the classroom, all work together to cast
considerable doubt upon their results. Claiborn (1969)
criticized the Rosenthal-Jacobson data analysis for
failure to correct for known pre—test differences, for
practicing partial data analysis, and for probability

pyramiding through multiple significance tests.

 

15

Although Rosenthal and Rubin (1971) have answered
many of the above criticisms, the debate concerning the
Rosenthal and Jacobson research continues, and, in fact,
is the subject of a full-length book (Elashoff & Snow,~
1971). Despite the criticisms left unanswered and the

questionable validity of Pygmalion in the Classroom, its

 

central hypothesis, that teachers' expectations for stu-
dents can become a self-fulfilling prophecy, appears to
have real merit. In addition, many reviewers have pointed
out that the work of Rosenthal and Jacobson should be
valued for its effect in prompting other, more sound,
research in this area.

The number of studies published in the area of
teacher expectancy is now in excess of 60. Even the
firmest believer in teacher expectancy effects will be
forced to admit that none of these studies has unequivo-
cally replicated the original Rosenthal and Jacobson
study and findings. To be sure, expectancy effects have
been found, but not in exactly the same way or with
exactly the same measures as were used in the "Oak
School" experiment. An exhaustive review of the more
than 60 published teacher expectation studies will not
be presented here. A more thorough review can be found
in Baker and Crist (1971) and, more recently, in Brophy

and Good (1974) and Dusek (1975).

16

One of the greatest obstacles to demonstrating
teacher expectancy effects is the fact that the induction
of expectancy is not always successful in an experiment.
Brophy and Good (1974) have analyzed a number of failures
to replicate Rosenthal and Jacobson's findings and con-
clude that many of the experimenters were simply not
successful in getting their teachers to believe or remem-
ber what they were supposed to believe about their stu-
dents. In support of this conclusion, Brophy and Good
cite several examples where large percentages of the sub—
jects failed to remember which of their students were
"bloomers," or disregarded the expectancy induction
because they guessed the purpose of the experiment.
Further, if one tallies the number of positive findings
in experimentally induced teacher expectation studies
versus those in naturalistic studies, where teachers have
already formed their own expectancies, the majority of
positive findings in the naturalistic experiments shows
the importance of successful induction.

The variables of professional experience and
teacher—student acquaintance appear to be important to
the success of teacher expectation experiments. Most
studies which have demonstrated expectancy effect in
teachers have used graduate or undergraduate students
with little or no experience in teaching. Rubavits and

Maehr's (1971) study using undergraduate volunteer

17

teachers and Carter's (1969) experiments with University
of Indiana students are examples of studies which have
shown powerful teacher expectation effects with inexper-
ienced teachers. The Rothbart, Dalfen and Barrett (1971)
study with teacher trainees also showed differential
treatment of students depending upon teacher expectations.
In some studies the expectation that certain stu-
dents were of "high" ability simply did not match the
teacher's perceptions of that student. This is perhaps
one of the reasons why studies begun late in the school
year have failed to produce the experimentally desired
expectations in teachers. In these situations the
teachers have had enough time to form their own impres-
sions of students before the beginning of the experiment.
The negative findings of the Jose and Cody (1971); Conn,
Edwards, Rosenthal, and Crowne (1968); Fielder, Cohen,
and Feeney (1971); and Claiborn (1969) studies are good
examples of experimentally induced teacher expectations
which did not occur until the second semester or later.
On the positive findings side of the ledger, studies in
which the student and teacher did not know each other
seem to find the process of induction easier. The Carter
(1969); Rothbart, Dalfen, and Barrett (1971); and Ruba-
vits and Maehr (1971) studies, cited earlier, all showed
expectation effects with teachers who were not acquainted

with their students. Brown's (1969) doctoral dissertation

18

also showed expectation effects with subjects and students
who were not acquainted. Rosenthal and Jacobson's (1966,
1968) original study showed significant IQ gains for the
first two grades only (p = .002 for the first grade and

p = .02 for the second) and with more gains for the first
graders than the second. Presumably the first grade
teachers were acquainted with fewer of their students
than teachers in the later grades. The most spectacular
expectancy effects are also observed in tutorial situ-
ations where students and their tutors have not been
acquainted before the experiment. Unfortunately, most
studies of tutorial teaching situations also involve
inexperienced teachers. Thus, with the variables of
experience and acquaintance confounded, it is difficult
to separate their effects and significance.

In their review of the teacher expectancy lit-
erature, Brophy and Good (1974) distinguish between pro-
cess and product measures used as criterion or dependent
variables. Product variables are those which show the
outcome of the self-fulfilling prophecy such as differ—
ences in IQ or achievement test scores. Process variables
are those intervening teacher behaviors which are sus-
pected of causing differences in learning. When one
examines the literature with the distinction of these
two types of variables in mind, abundant evidence of

differential treatment, or process results appear as a

19

consequence to different teacher expectancies. The exper-
iments of Medinnus and Unruh (1971); Rothbart, Dalfen,
and Barrett (1971); and Rubavits and Maehr (1971) all
suggest that teachers tend to interact more frequently
and more positively with high expectation than with low
expectation students. Brophy and Good (1970) found

that teachers communicated their expectations of stu-
dents by demanding more of high expectation students

and rewarding them more frequently. Low expectation
students, on the other hand, were the butt of more
frequent criticism, received less feedback for answers,
and were given fewer chances to respond. In another
example of process variable expectation effect, Chaikin,
Sigler, and Derlega's (1972) sample of undergraduate
tutors smiled three times more often, nodded their heads
two and one-half times more often, looked their students
in the eye more often, and leaned toward them longer when
they thought they were working with bright students than
when their expectation was of either normal or dull
intelligence.

The successful induction of differential expec-
tations and even the differential treatment of students
as a result of this induction does not necessarily lead
to differential product measures such as the tested
learning of a child or his IQ. Brown's (1969) doctoral

dissertation is a good example of this phenomenon. In

20

this tutoring situation involving the learning of states
and their capitals, Brown found that his high expectation
students learned no more than controls, but the teachers
with high expectations did attempt to teach more than did
teachers with low expectations. Kester and Letchworth
(1972), in their study using seventh grade students, also
found no expectancy advantage in product measures (achieve—
ment, IQ, and aptitude tests), but process observations
showed that teachers spent more time with high expectancy
students, talked with them more often, and were generally
more supportive toward them.

The length of time that teachers and students are
together in an experiment seems to have some bearing on
the outcome. Generally, those studies which have spanned
less than a full academic year have been more successful
in showing teacher expectancy effect than those studies
lasting for longer periods of time. It is suspected that
this is true because longer periods of exposure to their
students allows teachers a chance to form their own inde-
pendent notions regarding the ability of these students
and allows for more time to forget the experimenter's
induced expectancies. Perhaps the Conn, Edwards,
Rosenthal, and Crowne (1968) experiment provides the
most dramatic example of this observation. The Conn
et a1. study was a partial replication of the original

Rosenthal and Jacobson (1966, 1968) "Oak School"

21

experiment using the same basic paradigm. Two sets of
data were taken, one at four months and another at three
semesters after expectancy induction. While the first

set of data showed small differences in favor of the

high expectancy students, the second set actually reversed
this state of affairs so that in the end, the control
group showed the advantage to a degree very near statis-
tical significance. The self-fulfilling prophecy hypothe-
sis is most powerfully supported by the experiments of
Meichenbaum, Bowers, and Ross (1969) and Beez (1968).

Both of these experiments showed significant expectancy
effect on both product and process measures. The Meichen-
baum et a1. experiment lasted for two weeks and Beez's
subjects tutored students for 15 minutes.

As Brophy and Good (1974) suggest, teacher expec-
tancies are not necessarily harmful. Most expectancies
formed are accurate and help the teacher adjust his
instruction to the individual needs of the child. Beez's
(1968) study, for example, showed that "low ability"
children were given more examples and explanations and
were taught at a slower pace than "high ability" students.
If those children were actually low ability students,
their instruction would have been entirely appropriate
for them. Even if the teacher's initial expectations are
in error, most professional teachers quickly adjust these

expectations after having worked with the child. It is

22

only when teachers rigidly hold to their initially false
expectations for students that potentially harmful con-
ditions ensue. This kind of rigidity is usually the
result of poor teaching or poor mental health. Despite
the fact that inappropriate expectancy set is an example
of bad teaching, it does continue to exist as is so con-
vincingly demonstrated by Rist (1970) who cites anecdotal
evidence of the inappropriate "caste system" placement of
children whose expectations were formed arbitrarily by
kindergarten teachers and maintained by later primary
grade teachers.

In summary, the findings of the original Rosenthal
and Jacobson (1966, 1968) study have not been replicated
by any full academic year experiment using IQ or achieve-
ment as the dependent variable in an experimentally
induced expectancy design. In fact, the experimental
literature has not generally produced convincing evidence
that teachers' expectations have a differential effect
upon the actual learning or intellectual development of
children. Examples of the difficulties involved in
expectancy induction were presented which showed that
the expectancy must be believable and must not be at
variance with impressions the teacher has already formed
of the child in question. Therefore, experiments which
focus on students and teachers who are not acquainted

with each other generally show more positive results.

23

This is also perhaps one of the reasons why positive
results have been found with younger children and in
tutorial situations where teachers and students meet for
the first time. Experiments of short duration also

tend to show more positive results, again perhaps because
the induced expectation remains accepted by the teacher
who has less time to interact with the child and to form
conflicting opinions regarding his abilities. Studies
which use novice or relatively inexperienced teachers
also report positive expectation results more frequently
than those using experienced professionals. Presumably
the inexperienced teachers are more open to suggestion
and are less experienced in detecting student behavior
that is contradictory to the induced expectation.

That teachers do form expectations for students
and that these expectations can produce differential
teaching behavior is convincingly demonstrated by the
majority of studies which focus on process variables in
addition to, or instead of, the product measures of IQ
and achievement test scores. Often, however, children
are resistant to the differential treatment of their
teachers; several studies have shown teacher expectation
effects on process variables such as warmth, frequency
of interaction, and reinforcement patterns, but have
failed to show the effects of these behaviors on objec—

tive tests of achievement or IQ.

24

Although the literature on experimentally induced
teacher expectation effects is sometimes contradictory,
ambiguous, and confusing, studies which have been con-
cerned with the effects of teachers' own, naturalistically
formed expectations are very often cited as the most con-
vincing evidence of the self-fulfilling prophecy. Most
naturalistic studies are contained in the literature on
tracking which has shown, quite universally, that
teachers' expectations result in the differential
treatment of children. High expectancy children are
usually treated appropriately while low expectancy chil—
dren are not, and are, therefore, adversely affected.
Moreover, once a child is placed within a track or
expected ability group, his chances of remaining in that
group, regardless of his true ability, are surprisingly
high. Unfortunately, it is difficult to understand
clearly the results of tracking since outcome or product
measures in these naturalistic studies may have been
true measures of the students' abilities and their
teachers may have responded to real student differences

rather than false or biased expectations.

Experimenter Expectancy Effects

 

The classic paradigm for studying experimenter
expectancy effects is the "person perception task"
designed and standardized by Robert Rosenthal and first

reported by Rosenthal and Fode (1961). This task

25

involves the presentation of 10 previously judged neutral
photographs of faces to subjects who are asked to rate
the people shown in these photographs as experiencing
success or failure on a scale from ~10 to +10. In the
initial experiment using the person perception task
(Rosenthal & Fode, 1961), five experimenters were led

to expect an average rating of +5 from their subjects
while another five were led to expect average ratings of
-5. The results indicated that experimenters anticipating
an average rating of +5 in their subjects actually
obtained a mean rating of +.4 while the five experi-
menters who expected an average of -5 actually obtained

a mean rating of -.08 (p < .007). As part of his
master's thesis, Kermit Fode replicated this experiment
and published his results with Rosenthal in 1961 and
again in 1963 (Rosenthal & Fode, 1961, 1963). His
results were more impressive and showed mean ratings

of +2.27 and +.48 for the +5 and -5 expectancy treatment
conditions (p < .0003).

Since the original Rosenthal and Fode experiments,
over 100 studies have been published using the person
perception task as the vehicle for examination of various
aspects of experimenter expectancy. In his most recent
examination of the literature, Rosenthal (1976, p. 442)
found that person perception studies yielded significant

(p < .05) results 25 to 30% of the time. By simple head

26

count, therefore, one would be forced to conclude that
the balance of evidence weighs most heavily toward a
judgment that experimenter effects in this area are not
significant and are difficult to demonstrate. Further,
Rosenthal's (1976, p. 442) analysis of over 300 studies
including experiments in reaction time, ink blot tests,
animal learning, laboratory interviews, psychophysical
judgments, learning and ability, person perception, and
everyday situations yielded significant (p < .05) expec-
tancy effects in only 35% of the cases. However, Rosen-
thal (1976, pp. 441-442) argues that significant findings
in the approximately one-third of published expectancy
studies is a significant finding overall since this pro-
portion is about seven times what would be expected by
chance alone.

Most experimenter expectancy studies have not
really concentrated upon the question of whether or not
these effects occur, but rather have been concerned with
questions such as under what circumstances and with whom
are these effects most powerful, and how are the effects

mediated? Several of these variables are discussed below.

Sex
The literature does not clearly settle the
question of sex influences in experimenter expectancy.

First, the studies are conflicting, some showing greater

male influence, some female, and some showing no difference.

27

Further, there seems to be a difference in experimenter
effect depending upon the sex of their subjects and the
type of task in which they are involved. Finally, the
experimenter's sex variable has often been confounded
with other variables such as the experimenter's prestige
or dominance. Rosenthal (1969a, pp. 240-242; 1976,

pp. 225-233), however, believes that there is some
overall evidence to suggest that male experimenters

tend to produce the strongest expectancy effects.

These effects become more pronounced when male experi-

menters interact with female subjects.

Prestige

The prestige or status of the experimenter has
been a variable of interest in several studies. In
general, experimenters of higher status tend to be
better able to influence their subjects in expected
directions. However, the status of the experimenter
is often confounded with other variables such as age,
"air of authority," and simple interpersonal competence
in most studies. The prestige variable may not be
important in itself, but may depend more upon some of

the reasons why certain people become prestigious.

Experimenter Experience

 

As was the case in the teacher expectation liter-

ature summarized earlier, the greatest effects in general

 

 

28

experimenter expectancy studies are usually found with
less experienced experimenters. The experience variable
may also be indirectly examined in experiments which cover
a long period of time. It could be assumed that in such
experiments, the experimenter gains experience as time
progresses. Sampling of experimenter effects at pro—
gressive intervals has shown that expectancy effects are
more powerful at the beginning of an experiment than

during its later stages.

Anxiety and Need for Approval

 

The anxiety of the experimenter has been found to
be important in several expectancy studies. Anxiety
levels in both experimenter and subject have been shown
to be significantly related to expectancy effects, but
the direction of this relationship has been remarkably
unpredictable. It appears that highly anxious experi-
menters may show their greatest expectancy effects with
subjects who are also highly anxious. Conversely, when
low-anxious experimenters show their greatest expectancy
effects, their most susceptible subjects are likely to
have less than average anxiety as well.

Several studies have shown that expectancy effects
were positively related to the experimenter's need for
social approval, but only in the case of experimenters
whose level of anxiety was about average. When the

experimenter had either high or low anxiety, the

29

relationship between expectancy effects and experimenter
need for approval appears to have reversed itself and
minus correlations were observed. Although it would be
expected that subjects with a high need for approval

would show greatest susceptibility to expectancy influence,
studies have consistently shown that no relationship exists
between these two variables. It appears, then, that need
for approval has more to do with a person's ability to
influence someone else than to their susceptibility to

this influence. This influence occurs in expected
directions when the experimenter's anxiety level is about
average, but when anxiety is either high or low, the
experimenter's influence produces opposite than expected

results.

Experimenter-Subject Acquaintance

Teacher expectation studies, it may be remembered,
have tended to show greatest expectation effects in situ-
ations where the teacher and student were not acquainted.
The suspected reason for this phenomenon is that as the
teacher gained a better knowledge of the student, the
induction of the intended expectancy tended to weaken.
Teachers began to form their own expectancies based on
their experience with the child rather than depending
upon the artificially induced expectation. In the
experimenter expectancy literature this situation

appears to be reversed. Here, greatest expectancy

 

 

 

30

effects have usually been obtained when subjects have
known their experimenters. Examination of the experi-
menter expectancy studies, however, shows this finding
only in the case of male experimenters. When female
experimenters were used, the results were similar to
the findings in the teacher expectancy literature. The
discrepancy between teacher and experimenter expectancy
studies on this variable, then, may be more a reflection
of sex differences than anything else. It is probable
that most teacher expectancy studies used female
instructors while most experimenter expectation studies

have used male experimenters.

Task Variables

 

In answer to criticism by Dana and Dana (1969)
that the evidence for experimenter expectancy effects
comes almost entirely from person perception experiments,
Rosenthal (1969b) reviewed some 40 experiments which
employed other tasks. In general, Rosenthal noted, tasks
other than person perception tend to show the greatest
effects of experimenter expectancy. Sixty-six percent
of his 40 study sample showed significant (p < .10)
experimenter expectancy effects as compared with only
36% significant (p < .10) expectancy effects obtained
in person perception task studies.

In a more complete and recent breakdown of task

variables related to expectancy effects, Rosenthal (1976,

31

p. 443) verified his 1969 findings, this time at the more
acceptable 95% confidence level. Expectancy experiments
in animal learning showed a 64% significant (p < .05)
effects finding, while person perception tasks showed

only 27% significant effects above the .05 chance level.
In fact, of the eight types of experimental tasks reviewed
and compared, person perception task experiments were

tied with learning and ability expectancy studies for

the lowest position of the group.

Several authors have argued that relatively
structured or factual tasks may show less experimenter
expectancy effect than ambiguous or less structured tasks.
This postulation has been shown directly in a study by
Shames and Adair (1967) and is supported by several other

experiments.

Mediating Variables

 

Barber and Silver (1968) have outlined 11 possible
ways experimenters' expectancies can influence the results
of their research. Experimenters may unintentionally
influence their subjects to give expected responses
through paralinguistic cues (e.g. variations in voice
tone), kinesthetic cues (e.g. changes in facial expres-
sion), or verbal reinforcement of desired responses.

They may also intentionally or unintentionally misjudge
or misrecord subject responses. Finally, they can simply

fabricate their data. In Rosenthal's (1966, 1976) reviews

32

of the literature, the focus was primarily upon uninten-
tional influence through paralinguistic and kinesthetic
cues. In a careful review of 12 studies showing apparently
significant effects of expectancy, Barber and Silver (1968)
found that these effects were most commonly produced by
errors in judging, recording, or reporting the experimental
results. Verbal reinforcement as well as paralinguistic
and kinesthetic cues also communicated the experimenters'
desires in several of these studies. Barber and Silver
were led to suspect that when student experimenters test
student subjects on relatively structured tasks, the most
prominent mode of expectancy mediation may be intentional
data misreporting by the experimenters. In situations
where experimenters test subjects on relatively ambiguous
tasks, unintentional transmission of expectancy through
paralinguistic and kinethetic cues may be the most promi—
nent form of mediation.

In summary, it must be admitted that perhaps the
experimenter expectancy literature is more impressive for
its bulk than its incidence of significant findings. A
"box score" of significant expectancy effects ranges from
a high of 64% of studies in the area of animal learning
to a low of 27% of those experiments conducted in human
learning and ability and with person perception tasks.

The phenomenon of expectancy effect is not so simple as

to show itself without complication. Expectancy effects

33

are modified by and show interaction with such variables
as sex, experimenter and subject personality character-
istics, experimenter status and experience, and the degree
to which the experimenter and subject are acquainted. The
type of experimental task appears to make a difference in
expectancy effects and the responses of early subjects
seem to exert an influence upon the expectations of the
experimenter, as well. Although Rosenthal (1976) believes
that experimenter expectancies are transmitted to subjects
through such subtle cues as the experimenter's voice tone
or facial expression, other writers have shown that out-
right verbal reinforcement, intentional or nonintentional
misrecording of data, or misjudgment of subject responses

also frequently occur.

Psychometric and Clinical Experiments

 

The study of expectancy effects in clinical and
psychometric situations is complicated by the unrelia-
bility of psychological testing and the diagnostic pro-
cess. The reliability of psychiatric and psychological
judgment has been recognized as a problem in its own
right for many years. Disagreements among psychiatrists,
for example, have been the butt of jokes and the subject
of many popular press articles designed to produce
reactions ranging from amusement to indignation. The
professional literature, too, has produced evidence of

this problem. Kessel and Shepherd (1962), for example,

34

noted that a clinical judgment may be more revealing of
the clinician than the child. Stuart (1970) suggested
four factors that contribute to the low reliability
associated with diagnostic inquiry. First, differences
in individual clinicians may influence them to ask dif—
ferent questions and therefore elicit different behaviors
from their subjects. These clinician differences also
cause differences in the interpretation of behavior and
differences in reasoning and observation. Secondly,
irrelevant characteristics of the child such as dress,
social class, attitude, and race can sometimes influence
clinical judgment. Also, the category system of diagnosis
is faulty, according to Stuart. Differences between the
categories are based upon assumptions and inferences
rather than direct observations of behavior. It is rare
to find actual examples of the pathological conditions
found in textbook descriptions. Finally, low diagnostic
reliability may result from the context in which judgments
are made. The child may behave differently in an inter-
view or test situation than with his parents, classmates,
or siblings. Thus different judgments could be made
depending upon where and when the child was observed.
Diagnostic unreliability has also been studied
extensively with adult patients. Ash's (1949) sample of
clinicians, for example, agreed less than 50% even with

regard to what major diagnostic category should be used

35

on a given case. His findings have been supported by
many other researchers (Goldfarb, 1959; Grosz & Grossman,
1964, 1968; Katz et al., 1969; Mehlman, 1959).

Reliability studies with both psychiatrists and
psychologists, then, tend to show that certain internal
biases on the part of individual clinicians may be
responsible for some of their differences in diagnosis.
The diagnostic situation may be somewhat similar to a
projective personality test. The more ambiguous the
situation, the more it contains projective aspects for
the clinician and, perhaps, the less it contains true
information regarding the patient.

Intelligence Test Expectancy Studies
with Examiner-Examinee Interaction

 

 

The administration and scoring of intelligence
tests has provided a medium for the largest number of
experiments in the clinical and psychometric expectancy
literature. The results of these experiments are mixed
and appear to depend somewhat upon whether both adminis-
tration and scoring are studied or if scoring only is
the task of the experiment.

Larrabee and Kleinsasser (1967) in a widely
quoted, but unpublished study, found significant expec-
tancy effects on the verbal section of the Wechsler
Intelligence Scale for Children (WISC). The authors

provided fictitious scores to each of five examiners who

36

then administered a full WISC to 12 6th grade children of
average intelligence. The fictitious scores were arranged
so that the examiners expected either above average or
below average performance from their subjects. Each sub-
ject was tested by two examiners giving either the odd
or even numbered items of the test. On the average, sub-
jects whose performance was expected to be superior
obtained full scale IQs 7.5 points higher than subjects
who were thought to be below average. Differences were
found primarily in the verbal section of the WISC with
high expectancy subjects receiving an average of over 10
IQ points more than low expectancy subjects while the
performance section differences were less than three
points.

Hersh (1971) used 28 graduate school students in
a study designed to test the effects of positive or nega-
tive teacher referral reports upon the Stanford-Binet
and Peabody Picture Vocabulary Test scores of headstart
children. He found that the positive referral reports
influenced his testers to give significantly higher IQs
to their subjects so designated than to those children
who received negative referrals. In addition, positively
referred children were rated more favorably on the
Stanford-Binet's "Factors Affecting Test Performance."
The testers also recommended retention and special class

placement more frequently for negatively referred children.

 

 

37

Schroeder and Kleinsasser (1972), in an experi-
mental format similar to the Larrabee and Kleinsasser
(1967) study, reported significant WISC verbal IQ dif-
ferences when their 18 graduate school examiners were
given positive or negative expectancy through bogus pre-
test California Tests of Mental Maturity scores (dull-
normal or superior). Schroeder and Kleinsasser also com-
mented that since their experimental WISC protocols con-
tained few scoring errors, the effects of bias seemed to
have occurred before scoring. They noted that the Infor-
mation, Similarities, and Vocabulary subtests of the WISC
were most susceptible to bias.

In perhaps one of the first studies in expectancy
using intelligence test data, Wartenberg-Ekren (1962) was
unable to show significant examiner expectancy effect on
the Block Design subtest of the Wechsler Adult Intelli-
gence Scale. This failure occurred deSpite the fact that
the examinees reported differential treatment during test-
ing which was related to the examiner expectancy.

Gillingham's (1970) doctoral dissertation in this
area did not show significant expectancy effects. Gil-
lingham induced positive or negative expectancy in his
eight graduate student examiners by exposing them to a
rating sheet containing fictitious California Test of
Mental Maturity and school achievement ratings toqether

with a "predicted WISC score." Each examiner administered

 

38

a full-scale WISC to eight junior high students of
average intelligence. Gillingham suspected that his
failure to produce expectancy effects may have resulted
from a failure of his examiners to retain the expectancy
induction.

In order to determine whether the referral problem
has a biasing effect upon intelligence examiners, Saunders
and Vitro (1971) asked each of six graduate school
examiners to administer the Stanford-Binet Intelligence
Scale to 10 children. Five children were referred for
possible entry into a gifted educational program and the
other five were referred because of possible retardation.
The authors found no significant differences between the
two groups and concluded that clinical cognitive assess-
ment is not as likely to be influenced by examiner bias,
as may be possible in an experimental or nonclinical sit-
uation when the examiners do not feel as much of an
ethical responsibility for their testing.

Harry Dangel (1972) also reported a study of
expectancy effects using a population of 54 mentally
retarded children as examinees. Three graduate student
examiners administered the WISC to these children under
three different expectancy conditions. Bias was induced
by fictitious existing Stanford-Binet scores of 84-93
and favorable classroom comments (positive bias con-

dition), or Stanford—Binet scores of 55-64 and

39

unfavorable classroom comments (negative bias condition).
A control group was obtained by exposing the examiners

to the child's actual pre-test Stanford-Binet score and
classroom comments indicating average behavior. Dangel
found that the WISC scores did not differ significantly
across treatment conditions.

From the foregoing review it is impossible to
reach a conclusion regarding bias effects in intelligence
testing. In all of these studies, the examiner inter-
acted with his subjects and thus the presence of bias
effects may have depended upon the examiner's ability to
influence his subjects, quite apart from the effects this
bias may or may not have had upon his data interpretation.
A study by Lasky, Felice, Moyer, Buddington, and Elliot
(1973) seems to indicate that the influence of the
examiner's expectations upon his subjects' test behavior
can be quite substantial. Lasky et al. used administra-
tion and scoring of the Peabody Picture Vocabulary Test
as the dependent variable in this study with three
examiners. Two pre-test expectancy inductions were
presented, one giving an inflated picture of the sub-
ject's abilities and the other giving a true report of
his abilities. Examiner-examinee interaction was con-
trolled through two reinforcement procedures. For some
subjects, the examiner was instructed to use candy as a

reinforcer for correct responses, for others he was

4O

instructed to use the standard test reinforcement. Eighty
retarded mental hospital patients served as subjects.
Lasky et al. found that subjects in the combined standard
reinforcement-inflated pre-test score treatment condition
achieved an average of 9.6 IQ points more than their
standard reinforcement-true pre-test score counterparts.
However, when candy reinforcement was used for correct
responses, no expectancy effect was observed. Apparently
the primary reinforcement of candy was able to suppress
or mask the negative effects of examiner expectancy that
would ordinarily occur under normal testing conditions.
Intelligence Test Expectancy_8tudies

Without Examiner-Examinee Inter—
action

 

 

Studies which do not include examiner-examinee
interaction provide considerably more support for the
expectancy effects hypothesis than the previous studies
which did include this interaction. In the following
review, only one of the seven studies presented does not
show significant examiner expectancy effects.

Egeland (1969) studied examiner expectancy effects
in the scoring of the Comprehension, Similarities, and
Vocabulary subtests of the Wechsler Intelligence Scale
for Children (WISC). Forty-six graduate students were
asked to "help score some short forms of the WISC"
(verbal section). They were randomly assigned to three

treatment conditions: positive, negative, and neutral.

41

Expectancy induction was accomplished by giving each
scorer a referral form consisting of a reason for testing
(consideration of the subject for either accelerated
classes or classes for the slow learner), falsified
previous IQ (130 or 80), falsified raw scores for the
WISC Information and Arithmetic subtests, and false
standardized achievement test scores which were either
well above or below average. In the neutral condition,
test scores received no referral information or WISC
subtest raw scores. The results showed differences
between the scoring groups on the Comprehension and
Similarities subtests (p < .01), but not on the Vocabu-
lary subtest. A comparison of group means revealed that
significant differences were achieved between the above
average and below average expectancy groups, but not
between the neutral (no information) and below average
groups. Thus the below average expectancy group appar-
ently did not suffer from negative expectation so much
as the above average expectancy group benefited from
halo effect.

William Simon (1969) trained 72 introductory
psychology students to score some 20 WISC vocabulary
items previously selected from the WISC protocols of
children having IQs ranging from 90 to 110. Expectancy
was induced by suggesting that the child tested was

reading far above or below his age level and that the

 

 

 

42

score would be used to help determine whether to hold
back or skip ahead a grade in school. The group means,
analyzed by t test, were shown to be significantly

(p < .05) different.

Sattler, Hillix, and Neher (1970) have published
results from two experiments showing examiner expectancy
effects. In the first experiment, 15 graduate students
scored WISC Vocabulary and Comprehension subtests under
bright, dull, and unspecified expectancy conditions.
Expectancy was induced by supplying the scorers with
bogus scaled scores for the remaining WISC subtests
(6-8 for the dull condition and 12-14 for the bright
condition). Analysis of the data showed that the
examiners' scoring of the Vocabulary, but not the Com-
prehension subtest, was affected by expectancy. The
second experiment was devised to present an experimental
situation more like an actual examination setting. In
this experiment eight graduate students who had the
experience of having administered, scored, and written
reports for at least 26 tests were asked to score the
Comprehension, Similarities, and Vocabulary subtests
from the Wechsler Adult Intelligence Scale. Their
stimulus was audio-recordings of read item response
scripts designed to produce verbal 103 of either 130 or
90. Embedded within these response scripts were four

to five ambiguous responses for each subtest. Only these

 

 

‘W-h

 

 

43

15 ambiguous response scores were analyzed for differ—
ences between the average and superior expectancy con—
ditions. The results showed significant halo effect in
that the superior examinees received more credit than
the average examinees for the same ambiguous responses.
In an exceptionally well-designed set of two
experiments, Sattler and Winget (1970) were able to
simulate actual testing conditions, control for examiner
influence upon examinees and, at the same time, study
differences in examiner behavior as a result of expec-
tancy set. Eight graduate and former graduate students
administered the Verbal Scale subtests of the Wechsler
Adult Intelligence Scale to 18 paid female accomplices
who had memorized response scripts designed to yield
verbal IQs of either 130 (superior) or 96 (average). As
in the Sattler, Hillix, and Neher (1970) study, each of
these scripts contained 12 identical "ambiguous“ or
difficult to score responses. Expectancy was induced
through superior and average referral reports and through
performance on items other than the 12 ambiguous responses.
The testing sessions were tape-recorded to check for dif-
ferences in examiner behavior. Analysis of the data pro-
duced by these experiments showed that superior examinees
received more credit for ambiguous reSponses than average
examinees (p < .01), that the halo effect was operating

on all three subtests, and that only those examiners

44

who received superior referral reports gave their
superior subjects significantly more credit than their
average subjects (p < .01). Examiners did not differ

in the amount of probing questions they asked of superior
or average examinees. The results of both experiments
show that the effect of referral reports alone did not
significantly alter test scoring, but the combination

of superior referral reports and superior intellectual
performance on filler items did.

Expectancy effects in a foreign situation are
reported by Babad, Mann, and Mar-Hayim (1975). Eighteen
Hebrew University of Jerusalem graduate students scored
the Wechsler Intelligence Scale for Children record of
an Israeli fifth grader. Half of the scorers were given
a cover sheet portraying the student as underachieving
and disadvantaged while the other half were led to expect
a high-achieving, upper-middle class child. These expec-
tancies produced scoring differences in the expected
direction for the verbal subtests, but significance was
obtained only on the Comprehension subtest (p < .03).

In an effort to determine the effects of expec-
tancy in experienced examiners, Jacobs and DeGraaf (1973)
showed 32 practicing psychologists a video-taped admin-
istration of an intelligence test given to a child.
Before viewing the tape they were exposed to case his-

tories suggesting that the child was either bright or

45

dull. An analysis of their judgments concerning the

child's performance indicated significant expectancy

effect produced by the case histories. Experience in
psychological testing, apparently, does not guarantee
resistance to the effects of bias.

The experience variable was a major factor of
interest in a study by Auffrey and Robertson (1972)
which did not show significant expectancy effects. The
authors used 36 examiners at three levels of experience
as subjects. "Experts" were so classified if they had
at least two years of professional experience and at
least 50 administrations of each the Wechsler Intelligence
Scale for Children (WISC) and Wechsler Adult Intelligence
Scale (WAIS) to their credit. "Interns" were graduate
students who had completed Wechsler test training and
had previously administered between 10 and 50 of either
the WISC or WAIS. The "novice" level of experience was
represented by subjects currently enrolled in a graduate
Wechsler test class and who had scored at least one WAIS
and one WISC protocol at the time of the experiment.
After reading case histories depicting either a bleak
past with a strong tone of cultural and material impov-
erishment or a more optimistic background, the subjects
were asked to score one WAIS and one WISC protocol
previously prepared for the experiment. A control group

also scored the protocols, but was not exposed to the

46

case history information. The results showed that dif-
ferent case histories, either by themselves, or in combi-
nation with varying levels of experience, did not sig-
nificantly influence the scoring of the two intelligence
tests. Scoring reliability was significantly different
between the groups, however, as is shown by clearly higher
variability in the novice group of examiners. Although
not statistically significant, an inspection of WISC
scoring differences by treatment condition shows higher
scores were given to children with bleak pasts than to
either those with optimistic pasts or control group chil—
dren. Differences in this direction were not recorded
with the adult tests. One could almost speculate that
a slight trend toward helping the child make up for his
unfortunate past may have existed with these examiners.
In summarizing the findings from the above
literature, it appears that expectancy effects are con-
siderably easier to produce in an experimental setting
which does not provide for examiner—examinee interaction.
As was shown in the teacher expectation literature, it
is possible that the examiner in a test administration
situation may revise his expectations as a result of per-
ceiving test behavior which is in conflict with these
expectations. There is also some evidence to suggest
that positive expectancy or halo effect may be somewhat

more powerful than negative expectancy effect. This

47

possibility is shown most clearly by the Egeland (1969)
experiment and is supported in the general experimenter
expectancy literature by the original Rosenthal and Fode
(1961, 1963) person-perception experiments where positive
expectancy groups showed more of the effect than negative
expectancy groups. Expectancy effects also seem easiest
to produce in relatively ambiguous scoring situations
such as the Vocabulary, Similarities, and Comprehension
subtests from the Wechsler intelligence scales. Although
expectancy effect is difficult to show in a more straight-
forward scoring situation such as the Arithmetic and per-
formance subtests of the WISC, the Lasky, et a1. (1973)
experiment reveals that the influence of the examiner's
expectancy upon subjects' responses can be substantial,
even in highly structured scoring and testing situations
like the Peabody Picture Vocabulary Test.

Expectancy Effects in Per-
sonality Testing

 

 

The projective testing situation offers enough
ambiguity to enhance expectancy effects in the examiner's
scoring and interpretation of the data. The ambiguity of
the situation also tends to increase the examiner's
influence upon subjects who often, too, are looking for
help in structuring their responses. Even the subtlest
of cues is grasped by the examinee as is illustrated in

an expectancy study reported by Masling (1965). In this

48

experiment 14 graduate student examiners administered

the Rorschach Ink Blots to 28 subjects. Half of the
examiners were led to expect that it would reflect more
favorably upon them if they obtained more human response
percepts than animal percepts from their subjects. The
other seven examiners were given the opposite expectation.
A11 examiners were carefully warned not to coach their
subjects. The results indicated that examiners given the
high animal response expectancy actually obtained one-
third again as high an animal to human response ratio as
did the examiners led to prize human responses. Tape-
recordings of the testing sessions showed that there was
no differential verbal reinforcement of subjects'
responses. In addition, post-experiment questioning
showed that the subjects were not aware of special
interest in any particular type of response on the part
of their examiners. In examining how expectancy was
mediated in this experiment, Barber and Silver (1968,

pp. 20-21) showed that the examiners could have taken

the stimulus card away from the subjects after they gave
a desired response. If the subject did not give a
desired response, the examiner could have allowed the
subject to continue responding until the desired response
occurred. Posture, gesture, and facial expression also

could have communicated the examiner's desires.

49

Marwit and Marcia (1967) reported very strong
expectancy effects in a study with the Holtzman Ink Blots.
Thirty-six advanced undergraduate examiners administered
five of the Holtzman blots to 53 introductory psychology
students. Some of the examiners expected a high number
of responses per ink blot either because of their own
hypothesis or because of expectancy induction. The
remaining examiners expected few responses from their
subjects. The results of this experiment showed that
examiners expecting many responses because of their own
hypothesis actually obtained 59% more responses from their
subjects than examiners expecting fewer responses. Those
examiners whose expectations of many responses were exper-
imentally induced actually obtained 61% more responses
than their counterparts expecting few responses. Although
almost one-third of the examiners admitted to being aware
that expectancy effects were being tested in the experi-
ment, this awareness bore no relationship to the magnitude
of the expectancy effect. In a post-experimental analysis
of examiner-examinee interaction, no relationship was
found between the number of verbal inquiries made of
examinee responses and the actual number of responses
obtained (r = .07). In fact, contrary to the theory
that verbal conditioning may have occurred, examiners
who obtained fewer responses actually questioned their

subjects more than examiners who obtained many responses.

50

Conditioning of some kind must have occurred, however,
since greater expectancy effects were found with ink
blots shown later in the experiment. Perhaps the very
act of questioning responses was aversive to the subjects
in some way.

In a slightly more complicated study which com—
bined the dependent variables from the previous two
experiments, Marwit (1968) demonstrated strong expectancy
effects. The examiners were 20 clinical psychology grad-
uate students who were led to expect either many responses,
a high proportion of which would be animal responses, or
few responses with a high proportion of human content.
These particular combinations of response frequency and
content theoretically would not be expected from normal
subjects. The results were in the direction of expectancy
effects and were contrary to Rorschach theory. That
examiners improved in their ability to influence their
subjects is demonstrated by the fact that later contacted
subjects showed greater expectancy effects than earlier
contacted subjects.

Strauss (1968) was unable to show expectancy
effects using experienced examiners. Five female
examiners administered the Rorschach Ink Blots to six
female undergraduates under two expectancy conditions
and one control condition. Two subjects were expected

to show an introversive experience balance (comfort and

 

51

stimulation is obtained primarily from inner resources),
two were expected to show an extratensive experience
balance (comfort and stimulation is obtained from outside
sources), and two were tested without any expectation.
The results were that actual tested experience balance
was unrelated to the expectations of the examiner. In
order to determine the success of expectancy induction
in this experiment, Strauss asked the examiners to pre-
dict the obtained introversive—extratensive ratio for
their subjects. On the average, the examiners' pre-
dictions were similar to their expectancy induction.
This study suggests that examiner experience may be
negatively related to expectancy effects and is contrary
to Rosenthal's (1966, 1976) observation that more compe—
tent professional experimenters may be the ones to show
the most expectancy effects.

In projective personality testing, examiner
expectancy effects appear to depend somewhat upon the
examiner's ability and/or wish to influence his subjects.
This hypothesis is indirectly strengthened by Pfugrath's
(1962) experiment using the Taylor Manifest Anxiety
Scale (1953), a paper and pencil test of anxiety which
does not allow for as much examiner influence as pro-
jective tests. In this experiment, nine graduate student
counselors administered the Taylor to introductory

psychology students. Three examiners were led to expect

 

52

high anxiety, three expected low anxiety, and the remain-
ing three received no expectancy induction. The results
showed no significant differences among the three treat-
ment conditions.

Interview, Observational, and

Assessment of Therapy
Experiments

 

 

 

In more open-ended situations, the clinician must
exercise a great deal of judgment regarding the normalcy
of subject responses. Sometimes the same behavior can
be judged to be pathological or normal depending upon
the expectation or orientation of the clinician. Raffetto
(1967) has reported an interesting study in which eight
advanced undergraduate interviewers questioned 96 less
advanced, paid students who had spent an hour in sensory
deprivation. The interview consisted of both well-
structured yes-no questions and more open-ended questions
such as, "did you notice any particular feelings or sen-
sations?" The interviewers who were led to expect hallu-
cinatory experiences in their subjects actually reported
48% of their subjects above the grand mean for hallu-
cinatory experiences. Only 6% of the subjects were
reported above this grand mean by interviewers expecting
few hallucinatory experiences. Female interviewers
showed the greatest expectancy effects.

Expectations can haVe selective effects depending

upon their relevance to dependent measures as is

53

illustrated by Beal's (1969) doctoral dissertation. The
subjects in this experiment were 135 advanced graduate
students in clinical psychology who all viewed the same
motion picture of a purported therapy session. Prior to
seeing the movie the subjects were provided with a "staff
intake report" which varied in terms of diagnosis (neu-
rotic, psychopathic, not stated) and motivation for treat-
ment (high, low, not stated). There were, therefore, nine
possible experimental conditions. Post-Viewing ratings
of the person depicted in the film were obtained from all
subjects on dimensions such as attraction, prognosis,
willingness to treat, and perceived similarity. Regular
pauses in the film allowed for subjects to indicate what
they would say. These responses were content analyzed
for empathy and warmth. Beal's results showed that diag-
nosis suggestion created expectancy effects in the areas
of attraction, empathy and warmth, but not in prognosis,
willingness to treat or perceived similarity. Signifi-
cant expectancy effects for motivation were seen in the
subjects' ratings on prognosis, willingness to treat, and
perceived similarity, but not on attraction, empathy,
or warmth.

Kent, O'Leary, Diament, and Dietz (1974) published
a study which shows that judgments concerning data can
differ as a result of expectancy even though observations

of actual behavior do not. Kent, et al., predicted a

54

decrease or no change in the level of recorded behavior
as a function of "treatment" to two groups of observers.
Both groups viewed the same video-tapes of child behavior
in a classroom setting. The child behavior tapes were
selected to actually show no change from "baseline" to
"treatment" behavior. The results of this experiment
showed no significant differences between observer groups
on behavioral recordings, but the observers' global
evaluations of treatment were significantly affected

by the authors' predictions.

The difference between expectancy effects upon
behavioral and global ratings is also shown in an experi-
ment reported by Beasley and Manning (1973) in the field
of speech pathology. Forty master's degree students in
speech pathology were asked to evaluate a lO-minute,
tape-recorded language sample using three objective
measures and four subjective scales of language per—
formance. Four levels of biased pre-information were
presented: (a) speaker's age only, (b) speaker's age
and the statement, "this child has delayed language,"

(c) negative pre-information, and (d) positive pre-
information. The findings of this study did not show
bias effects on the objective measures of language per-
formance, but subjective ratings were significantly
affected by the pre-information. Expectancy effects in

the area of speech pathology and audiology have not

 

 

 

55

generally been found when objective measures are used as
the dependent variable (Hipskind & Rintelmann, 1969;

Meitus, Ringel, House, & Hotchkiss, 1973).

Labeling Bias in Clinical Studies

 

Szasz (1961) contended that psychiatric diagnosis
is a process of labeling social behavior rather than a
classification of diseases. The referent to which a
given person's behavior is compared, then, becomes the
social and ethical norms of society and psychiatry instead
of observable disease processes as is contended by the
professionals. This point of view is given impetus by
a series of studies authored and co-authored by Maurice
Temerlin. These studies provide some of the most blatant
and hard hitting indictments of labeling in the literature.

In his first study of labeling bias, Temerlin
(1968) trained a professional actor to portray a mentally
healthy man. He was happy in his work, warm in his
relationship with the interviewer, self-confident, secure,
and happily married. He had but the most infrequent
arguments with his wife and showed some normal concerns
over the world situation. In other words, his script was
systematically designed to be contra-indicative of mental
illness. Just prior to listening to an audio-recording of
this interview, 45 graduate students in clinical psy-
chology, 25 practicing clinical psychologists, and 25

psychiatrists heard a "prestige confederate" remark that

56

the patient on the tape was "a very interesting man
because he looks neurotic, but actually is quite psy-
chotic." Three control groups were matched and stratified
for professional identity. One group diagnosed the inter—
view without prior suggestion and the other diagnosed it
with the prior suggestion reversed (" . . . looks psy—
chotic, but actually is quite neurotic"). Another group
evaluated the interview as part of an "employment project."
A fourth control group consisted of a mock sanity hearing
with lay jurors. Subjects diagnosed the taped patient on
a list of 10 psychoses, 10 neuroses, and 10 miscellaneous
personality types, one of which was "normal or healthy
personality." The "employment evaluators" responded to

10 scales within which was embedded a mental health scale
with psychosis at one extreme and neurosis in the center.
The results of this experiment showed differences between
experimental and control groups to be significant at the
.01 level. While no control group subject ever diagnosed
psychosis, 60% of the psychiatrists, 28% of the clinical
psychologists, and 11% of the graduate students did
diagnose psychosis.

In discussing his results, Temerlin (1968) sug-
gested that psychiatrists may have displayed a poor show—
ing because organized medicine rewards conformity with
prestige figures, because psychiatrists probably encounter

more psychotics in their daily work than the other

57

professional groups, and because of the old implicit
rule in medicine, "when in doubt, diagnose illness."

An illness diagnosis, in an uncertain situation, is much
less dangerous to a physician than a health diagnosis
when illness is, in fact, present.

Temerlin and Trousdale (1969) expanded the data
from the original Temerlin (1968) experiment. They show
that this study included 156 undergraduate students and
40 law students besides the psychiatrists, clinical psy-
chologists and clinical psychology graduate students
reported by Temerlin (1968). When all of the diagnostic
possibilities (psychoses, neuroses, and character dis-
orders) are considered together, diagnoses of "mental
illness" in this experiment ranged from a low of 84%
among undergraduates to a high of 100% among the psy-
chiatrists. Diagnoses of normality ranged from 0% to
16% in the experimental groups to 57% to 100% in the
control groups. In addition to diagnosing the "patient"
on the list of diagnoses, each subject was asked to write
a brief description of the patient. They were instructed
to report a description of behaviors which indicate per-
sonality characteristics and to avoid clinical inferences.
An analysis of these reports showed a consistent failure
to describe what was heard, rather than inferred, despite

the explicit instructions to be descriptive. The subjects

 

 

58

appeared to leap to conclusions which were then defended
rather than re-evaluated in the light of subsequent
observation.

As part of Lee and Temerlin's (1970) study on the
relationship of perceived social class and psychiatric
diagnosis, the authors partially replicated the above
experiment. Here, 30 psychiatric residents were asked
to diagnose the same audio-recording as was used in the
Temerlin (1968) experiment. Diagnosis was accomplished
by rating the "patient" on a nine-point scale with psy-
chosis and normality as end points. Before diagnosing
they were told that "two board certified psychiatrists
and a psychoanalyst thought the patient 'neurotic' (N=10)
or 'psychotic' (N=10)." The control group (N=10) diagnosed
without expectancy induction. Post-experimental analysis
revealed that all subjects who expected the patient to be
neurotic diagnosed him as such; 9 of the 10 subjects who
received the suggestion that he was psychotic diagnosed
psychosis; and 9 of the 10 controls diagnosed him as
normal.

Critchley's (1970) doctoral study with 90 nursing
students also shows powerful expectancy effects resulting
from psychiatric labels. The nurses viewed a filmed
interview of three children after being informed that
each child was normal, obsessive-compulsive, or schizo-
phrenic. After viewing the film, the nurses rated each

child on an evaluation check list. The results indicated

59

that the subjects evaluated children labeled as schizo-
phrenic or obsessive-compulsive as significantly more

' disturbed than children labeled normal. Interestingly,
these subjects did not evaluate schizophrenic children
as any more disturbed than those they thought were
obsessive-compulsive.

In his attempt to explain some of the reasons for
the relatively poor showing of psychiatrists as opposed
to psychologists in his labeling bias experiments, Temerlin
(1968) noted that psychologists tend to be more critical
and skeptical than psychiatrists because of their histori-
cal origins in philosophy and current identifications as
social scientists. This is illustrated by the hetero-
geneous divisions of the American Psychological Associ-
ation and by the old joke, "wherever there are two psy-
chologists, there are three opinions." In this vein,
Langer and Abelson (1974) have published a study which
examines expectancy and its relationship to the theoreti-
cal orientation of clinical therapists. The subjects
were 40 clinicians associated with university departments
known to be either behaviorally or psychodynamically
oriented. Most of the subjects were psychologists, but
about one-fourth were psychiatric residents. In prepar-
ation for the study, the authors video-taped an interview
with a paid job applicant. Although the interview was

unstructured, it centered around the interviewee's

60

feelings and experience relating to his past work. Bias
was introduced by telling half of the subjects the inter-
viewee was a "patient" and the other half he was a "job
applicant." After viewing the video-tape, each subject
was asked to write a brief, free-response description of
the interviewee, his gestures, his attitudes, and the
factors that probably explained his outlook on life.
These descriptions were then quantified by having five
graduate students blind rate them on a scale from one
(very disturbed) to 10 (very well-adjusted). To make
sure that the clinicians from the different universities
did indeed hold different theoretical orientations, they
were asked to label their orientation and to respond to
a four-item rating scale designed to distinguish between
behaviorists and psychodynamically oriented therapists.
An analysis of variance performed on the Langer
and Abelson data revealed a significant difference
between the expectancy (job applicant vs. patient) groups
(p < .01). Analysis of the interaction between theoreti-
cal orientation and label also showed significance (p <
.05). The "patient" label produced significantly more
ascriptions of disturbance from the psychodynamically
oriented clinicians than from their behaviorally oriented
colleagues. On the other hand, no real differences
were observed between the two theoretical groups when

they thought the interviewee was a "job applicant."

61

Langer and Abelson speculated that the behaviorally
oriented therapists may have resisted bias because their
training actively encouraged a discounting of labels.

Summary and Implications
for This Study

 

Expectancy effects have been studied through two
basically different methods. One method has involved the
induction of an expectancy followed by a measurement of
the effects upon a second party such as a student in a
teaching situation, a subject in an experimental situ—
ation, or an examinee in a testing situation. The effects
of expectancy, then, were dependent upon the success of
the expectancy induction (the degree to which it was
believed or retained by the teacher, examiner, or experi-
menter) on the one hand, and the degree to which the
biased person was able to influence his subject on the
other. These factors have complicated the study of bias
effects and account for many of the inconsistencies seen
in this literature. In general, the teacher and experi-
menter expectation literature, whose studies have used
this method of measurement, have not produced over-
whelming proof of the "self-fulfilling prophecy."
Expectancy effects, when second party measurement was
used, appear to have been modified by many variables
including age, sex, social status, personality charac-

teristics, experimental task, ambiguity of task, length

62

of experiments, believability of bias induction,
experimenter-subject acquaintance, experimenter exper-
ience, and the use of outcome versus process variables

as dependent measures. The effects of these variables,
however, have not been shown to be straightforward. For
example, the sex and personality characteristics of the
experimenter have interacted with these same variables

in subjects in ways which suggest that these variables
may have more to do with the experimenter's ability to
influence subjects than expectancy effects per se. Since
this study did not provide for interaction between school
psychologists and the child they evaluated, it was not
deemed important to include measures or analyses of

these variables in the design.

A second method of measuring expectancy effects
has involved the study of the attitudes and judgments of
the biased person such as has been seen in experiments
of diagnosis, test data interpretation, behavioral obser-
vations, assessment of psychotherapy, and interviews.
Here, expectancy effects appear to have been much easier
to demonstrate than in second party measurement studies.
Greatest effects have usually been observed when persons
making a judgment did not physically interact with their
subjects, but merely observed them or their productions

on a test.

63

Since this study appears to have been the first
of its kind using school psychologists as subjects, it
was decided to maximize the probability that expectancy
effects would occur by adopting this second method of
measuring the effects. It was reasoned that if expec-
tancy effects could not be demonstrated in this study,
it would be unlikely that any other study with this same
population would show expectancy effects with psychologist-
child interaction. If expectancy effects could be demon-
strated in this study, on the other hand, an examination
of the relationships among the variables measured would
provide valuable information for the design of later
experiments. To further maximize the probability of
showing expectancy effects, this study utilized global
evaluations and inferred judgments as dependent variables
rather than structured and objective measures. Barber
and Silver (1968) have shown that structured tasks tend
to produce fewer expectancy effects than more ambiguous
ones. Their observation is supported in this literature
review most directly by the experiments of Beasley and
Manning (1973) and Kent et a1. (1974). Some of the
examiner expectancy studies reviewed here also showed
that the more ambiguous subtests of the Wechsler Intel-
ligence Scales have demonstrated greater expectancy
effects than the more structured subtests (Larabee &

Kleinsasser, 1967; Schrader & Kleinsasser, 1972).

 

 

64

The studies reviewed here have shown that the
experience variable can sometimes be important. Teacher
expectancy studies have generally indicated that less
experienced teachers tend to show greatest expectancy
effects. The experimenter expectancy literature has
been more equivocal on this variable and has sometimes
shown greatest effects for experienced experimenters
(Rosenthal, 1966, 1969a) while greatest effects have been
shown for less experienced experimenters in others
(Barber & Silver, 1968). Still others have shown the
variable to be unimportant (Ingraham & Harrington, 1966).
In testing, experience has not been found to be a critical
variable (Auffrey & Robertson, 1972; Hersh, 1971; Sattler
& Theye, 1967). Brophy and Good (1974) have suggested
that experience may be important in preventing expectancy
effects since experienced teachers may be better resisters
of expectancy induction and may also be better at per-
ceiving evidence which disconfirms their expectations.

On the other hand, Temerlin's (1968) study showed fewest
expectancy effects with inexperienced subjects. Since
this variable appears to have been important in many
studies and since its effects have been quite unpre-
dictable, this study has included experience as a
variable to be examined further.

Stuart (1970), as well as other writers (Grosz

& Grossman, 1964; Szasz, 1961) have reasoned that

 

 

65

clinicians' theoretical beliefs can influence their
diagnostic judgments. The Langer and Abelson (1974)
experiment provided the most direct test of this notion
of all the studies reviewed here. Their finding that
psychodynamically oriented clinicians were significantly
more influenced by negative expectancy than their
behaviorally oriented colleagues did much to prompt

the inclusion of the theory variable in this study.
Although several writers have speculated about the
importance of clinicians' theoretical orientations in
their practice, this author could find no studies, other
than Langer and Abelson's (1974), which have examined
this variable. Thus the inclusion of a theory variable
in this study seemed to be even more important.

Several studies have reported findings which
suggested that positive expectancy or halo effects may
be somewhat more powerful than negative expectancy
effects (Egeland, 1969; Rosenthal & Fode, 1961, 1963).
Most expectancy studies have not been designed with a
capability of evaluating possible differences in the
relative power of these two expectations since a test
of the resulting hypotheses would require both positive
and negative expectancy induction as well as a control
condition. For this reason, a control group was included

in this study.

CHAPTER III

METHODOLOGY

The Sample

 

Forty-eight lower Michigan school psychologists
from eight school district psychological services offices
participated in this study. The eight school psychologi-
cal services offices were randomly selected from those
offices in the state of Michigan which employed three or
more psychologists. The eight district offices were
located in both metropolitan and rural areas. Four of
the offices and 30 of the school psychologists were
located within intermediate school districts while the
other four offices and 18 school psychologists were part
of local city school districts. The eight offices
employed a total of 58 school psychologists, 10 of whom
(17%) did not participate in the study. Post-experiment
questioning revealed that eight of the 10 nonparticipating
school psychologists had scheduling conflicts and thus
could not attend the experimental session. Of the two
remaining, one psychologist was "too busy" to participate,

and the other was "not interested."

66

 

67

This sample of school psychologists contained a
mixture of male and female subjects with from 0 to 18
years of professional experience. Although three of the
subjects were interns, and thus had no professional
experience, the mean experience level of the group was
6.375 years. Most of the subjects (75%) indicated that
they had received their graduate training in school psy-
chology while 23% said their graduate training had been
in clinical psychology (2% had graduate training in both).
When asked to rate their theoretical orientation on a
continuum with the behavioral and psychodynamic schools
of thought as extremes (see Appendix I), 60% of the sub-
jects rated themselves more toward the behavioral side
of the median while the others identified themselves more
with the psychodynamic orientation. If subjects who used
the middle scale points (5 & 6) on the above continuum
could be described as being eclectic in their theoretical
orientation, the distribution of the group's theoretical
orientation changes to 52% behavioral, 23% eclectic, and
25% psychodynamic.

In addition to the above sample, a second study
was conducted with 12 of the 17 school psychologists
employed by Grand Rapids Public Schools in Western
Michigan. Grand Rapids is a medium-sized, diversified
industrial city whose schools educate approximately

31,000 students. The school district has extensive

68

special education programming and many of its programs
are used by other school districts within the county.
The mean experience level of this group was 5.4 years
and all but one of the subjects had received graduate
training in school psychology. They also rated their
theoretical preferences with the following results:

58% behavioral and 42% psychodynamic, or 50% behavioral,
8% eclectic, and 42% psychodynamic.

Table 3.1 shows the subject characteristics for
both of the above samples. Each school district's com-
munity type is also listed. In the case of intermediate
districts, the various community types are listed with

the number of local school districts within each type.

Experimental Stimuli

 

All school psychologists in this sample were pre-
sented with the case study of a very active, six-year-old
boy in the first grade. Each psychologist was given a
stapled packet of information concerning this child which
contained the following items in sequence: a teacher
referral report, a background information sheet, a filled
out and scored Wechsler Intelligence Scale for Children-
Revised protocol (1974), a Bender Visual Motor Gestalt
Test protocol (Bender, 1946), a Kinetic Family Drawing
(Burns & Kaufman, 1972), and a scored composite of a
Wide Range Achievement Test protocol (Jastak, Bijou, &

Jastak, 1965). Following their perusal of the above

 

69

mHOHHumHQ Hoogom ouoflpofinouch

 

 

 

H HH H.HN e.m mnoo .ouumz m a meHmmm
pcmuw
Ame Hensm
N m m.OH m.H Ame egos H m we
H N m.sH m.OH mueo .ouumz N H A
o m o.mH k.m OmaHum genus H m e
H e h.o~ N.s muoo .ouumz m m m
m a m.mH H.> Ame mmcHnm menu: m e we
HHHV Hausm
o H m.mH o.m Ame c309 o H AM
Hey Hmnsm
Ame c309
m CH m.m~ o.a Ame maaHum amass m NH am
0 H m.HN m.m mmaHum cons: m m H
HCOHCHHU Hoonom coauwpcmflho mocmflnomxm Amvomme m z uoauumflo
WEHCHMHB HMOHHOHOOSB COOS mhmww G602 \wUHgEEOU Xmm HOOfiom

 

moapmfluouomnmnu mHmEmm

H.m OHQMB

 

 

7O

printed data, the psychologists viewed a video-taped 10-
minute segment of the child in his classroom. Each of

the above stimuli are discussed in more detail below.

Teacher Referral Report

 

Three different referral reports were prepared by
the author, each of which constituted a biasing or con-
trol treatment condition for this study (see Appendix A).
The teacher referral report was stapled to the top of the
above information paCket so that it would be the first
information read by the psychologist. In addition to
stating the child's name, age, birthdate, and grade, each
referral report contained a brief description of his aca-
demic achievement, his classroom and playground behavior,
and his approach to classroom work. The reason for the
referral was also stated. Although all three referral
reports described the same child's behavior, the teacher's
choice of words and reflected attitude toward the child
differed considerably on each one.

One referral report was negatively oriented to
simulate one that could have been written by a teacher
who was both upset and angry with this particular child.
His behavior was described in negative and pathological
terms such as "hyperactive, aggressive, and impulsive."
He was referred for help with his problems and testing
was requested to determine his eligibility for special

education help.

 

 

71

A second, more positively oriented referral report
described the same behavior, but in neutral and positive
terms such as "energetic," "assertive," and "spontaneous."
This teacher enjoyed having the child in class, but
referred him for testing to " . . . see how I can help
Scott best utilize his abilities and to determine if he
has any particular needs."

The third type of referral was devoid of infer-
ential language and described the child in strict
behavioral terms. Specific examples of the behaviors
described in the other referral reports were given along
with their frequencies. The reason for this child's
referral was the same as that described in the positive

referral report above.

Background Information Sheet

 

The background information sheet contained a
parent's statement, a series of short developmental
history statements, and a few comments about the child's
kindergarten experience (see Appendix B). This sheet was
prepared by the author to present a child with a few
problems and one who was very mildly delayed in his
development. Care was taken, however, to present a

history which was not unusual or abnormal.

 

 

72

Revised Wechsler Intelligence Scale
for Children Protocol

 

 

The author filled out and scored a WISC-R record
form. The individual item responses were gleaned from
the files of actual children tested by the author, and
were selected so that the protocol would show some clini-
cal evidence of borderline developmental problems in the
areas of short—term memory (Digit Span), visual organi-
zation (Block Design and Object Assembly) and psycho—
motor speed (Coding). Sattler's (1974, p. 557) tables
for "Significant Differences Between WISC-R Scaled Soores
and Between IQ's" were consulted to make sure that none
of the differences between scaled scores on this protocol
reached statistical significance. Thus all of the scaled
score differences on this protocol could easily have
occurred by chance alone. The protocol was scored to
portray a child with low average (WISC-R full scale IQ,
92) intellectual ability (see Appendix C).

Bender Visual-Motor Gestalt
Test Protocol

 

 

An actual Bender Visual-Motor Gestalt Test proto-
col was selected from the author's files. It was
selected from the file of a child previously diagnosed
as normal and showed development at a level reached by
approximately 60-75% of children aged 6 years (Bender,
1946, p. 5). The protocol was somewhat expansive (it

covers two sheets) and one of the designs showed

 

73

development of about a year less than the chronological

age of the child (see Appendix D).

Kinetic Family Drawipg

 

This test protocol copy was also originally
obtained from the author's files of actual tested chil-
dren and it, too, was selected for its borderline quality.
The drawing depicted the child, his mother, his father,
and two of his animals. The child's younger sister, men-
tioned in the background sheet, was missing from the
drawing. The developmental level of the drawing was
approximately normal for a six-year-old (Burns & Kaufman,
1972) and all of the figures appeared to be smiling and
happy (see Appendix E).

Wide Range Achievement
Test Protocol

 

 

A composite of the three sheets for this test
was made so that the child's responses to individual
items of the Reading, Spelling, and Arithmetic subtests
could be seen on one page (see Appendix F). The test
protocol was previously scored at a level consistent with
the child's intelligence in the case of reading and
spelling (1.2 grade equivalent) and slightly above this
level in the case of arithmetic (1.6 grade equivalent).
The test protocol showed the child to have learned the

alphabet, but to be functioning at a pre-reading level.

 

74

The protocol was prepared by the author rather than

taken from the case file of an actual child.

Video-Taped Behavior Observation

 

Psychologists were presented with a lO-minute
video-tape showing the child in his classroom during a
math assignment. The child seen in the tape was selected
from 11 possibly overactive boys in the Grandville Public
Schools, Grandville, Michigan. The boy selected was
video-taped in his classroom for a full school day after
permission to do so was obtained from his parents. The
actual video-taping was done by the author from a corner
of the room while class was in progress. The child seen
on the tape was previously judged nonhyperactive by a
school social worker. Moreover, the results of a battery
of tests administered to him after the taping failed to
reveal excessive hyperactivity, learning disability, or
any other pathological condition as judged by the author
and an associate school psychologist. The boy seen on
the video-tape, then, was a normal, white, middle class
child in a traditionally structured first grade suburban
classroom. The lO-minute segment selected for this
experiment did not show the driven quality of behavior
seen in classical hyperactivity, but it did show a
left-handed boy, with little task orientation, who was
out of his seat approximately 30% of the time during

an arithmetic seatwork assignment. He was not under

 

 

75

direct teacher instruction since she was in the front of

the classroom with a reading group during this segment.

Dependent and Independent Measures

 

The dependent and independent measures used in
this study were attached to the bottom of the information
packet given to each school psychologist at the beginning
of the experiment. The measures were sequentially
arranged so that dependent measures were presented before
independent measures. Each measure is discussed below,
following the order in which it was presented during the

actual experiment.

Severity of Pathology Scale

 

This scale was prepared by the author as a
measure of the school psychologists' judgment of patho-
logical severity for the child in question. It contained
nine rating scales, seven of which related to separate
areas of psychological functioning: "social functioning,
inhibitory functioning, behavioral control, attentional
processes, perceptual-motor development, cognitive
organization, and relative academic functioning." The
other two rating scales called for overall judgments in
functioning and consistency of functioning (see Appen-
dix G). The nine rating scales were actually continua
with 10 points and upon which the psychologist could rate
the child as functioning below average, average, or

above average.

 

 

 

76

Each of the nine individual ratings for the above
measure was scored as a deviation from the center score
(5.5) on the continuum. In six of the rating scales
("Social adjustment, Inhibitory functioning, Behavioral
control, Attentional processes, Cognitive organization,
and Relative academic functioning") a deviation from the
average score in either direction (below average or above
average) was an indication of pathology and its deviation
score was expressed in positive values and in increments
of .5. For example, a rating of 3 on the Social adjust-
ment scale was expressed as 2.5 and a rating of 7 was
expressed as 1.5. In three of the individual rating
scales ("Perceptual motor development, Overall consistency
of functioning, and Overall functioning"), however, a
rating in the above average range (6-10) was considered
an indication of superior adjustment and therefore
ratings which occurred within this range were given
minus values. For example, a rating of 3 on the Per—
ceptual-motor development scale was given a value of
2.5, but a rating of 7 was given a value of -l.5.

A single "severity of pathology" score was
derived from the above nine individual ratings by a
simple summation of the nine deviation scores. A high
number was an indication that the psychologist rated the
child's pathology as severe while a low number indicated

an absence of pathology and, possibly, superior adjustment.

 

 

77

Diagnostic Classification

 

Parts of two diagnostic classification systems
were printed on the second response sheet (see Appendix H)
and each subject was asked to indicate his diagnostic
choice on both systems. One system was entitled "Man-
datory Special Education System" and included only those
diagnostic classifications from the Michigan Mandatory
Special Education Act (Public Act 198, 1971) which could
in any way be associated with the child being evaluated.
These diagnostic choices included "educable mentally
impaired, learning disabled, and emotionally impaired."
A fourth choice was placed at the top of the list entitled
"not eligible for special education." Since only two
classifications were of interest in this study, a diag-
nosis of pathology or the lack of it, a simple two-point
nominal scale score was used for analysis purposes on
this measure.

The other diagnostic classification list to
which the subjects of this study responded was entitled
"Classical System" and included several possible diagnoses
which could have been rendered by a clinical psychologist
working outside of schools. These diagnostic choices
included "functioning within normal limits, developmental
personality deviation, minimal brain dysfunction, neu-
rotic reaction, and behavior disorder." A space was

also provided for the subject to specify any other

78

diagnosis he wished if none of the above choices met
with his satisfaction (see Appendix H). The score for
this measure was derived in the same manner as that for

the "Mandatory System" described above.

Degree of Confidence Scale

 

The subjects were asked to rate the degree of
confidence they felt in making the above diagnostic
judgments (see Appendix H). They rated their confidence
on a scale from one (very unsure) to 10 (very confident).

Special Education Treatment
Recommendation

 

 

A measure of the school psychologists' treatment
recommendations was obtained by asking them to select one
of eight special education treatment possibilities. The
eight possibilities were derived from a similar list by
Dunn (1973, p. 37) and consisted of special education
interventions ranging from no special help at all to a
full day placement in a special education classroom, part
of which involved placement in a separate special school
(see Appendix H). Since this list of treatment possi-
bilities was hierarchically arranged with ever-increasing
amounts of special education intervention, scoring on
this measure was accomplished by numbering each possi-
bility from one to eight with the higher numbers indi-

cating more treatment involvement.

79

Referral for Pediatric Determination
of Medication Possibilities

 

 

A measure of the school psychologists' willing-
ness to refer this child for medication was obtained by
simply asking them to respond, in yes or no fashion, to
the question, "Would you refer this child to his pedia-
trician for determination of medication possibilities?"
The score for this measure was based on a two-point

nominal scale for analysis purposes (see Appendix I).

Theoretical Orientation

 

A measure of the subjects' theoretical preference,
behavioral versus psychodynamic, was obtained by asking
them to rate, on a six-point scale, their relative agree-
ment with four statements adapted from those used by
Langer and Abelson (1974, p. 7) for the same purpose.

The statements presented areas of disagreement between
behavioral and psychodynamic schools of thought. In
addition, the subjects were asked to rate their theoreti-
cal preference on a continuum from one (strongly behavior—
istic) to 10 (strongly psychodynamic) (see Appendix I).

A single measure for this independent variable was
obtained by summing the scale points (with statement
number one reversed) on all five rating continua. Thus
scores of between 5 and 34 were possible with scores

above the median (19.5) indicating more agreement with

the psychodynamic theoretical orientation.

80

Experience and Training

 

The measure of the school psychologists' pro-
fessional experience in this study was obtained by asking
them to respond to the request, "Please write the number
of years you have been practicing in school psychology
since your graduation from training" (see Appendix I).

A measure of the type of training experienced by the sub-
jects was also obtained. Subjects simply checked whether
their graduate training was primarily in school or clini-

cal psychology (see Appendix I).

Procedures

SamplinggProcedure

 

A list of school psychologists in the state of
Michigan, arranged by school districts, was obtained from
the state consultant for school psychology (Department of
Education, State of Michigan). After each school dis-
trict office had been assigned a number, a table of
random numbers was entered and offices (school districts)
were selected to be used in this study as their number
appeared in the table. In April, 1976, key school psy-
chologists within each selected school district office
were then telephoned and asked if they would poll their
colleagues to determine their willingness to participate
in "a case study examining the diagnostic process." A

follow-up call revealed answers in the affirmative for

81

the first nine offices contacted and tentative dates for
the experimental sessions were set for May and June, 1976.
A potential sample of 63 school psychologists was thus
obtained in the nine Michigan school districts. One of
the school districts, employing five school psychologists,
was later forced to drop out of the study because of an
especially heavy work load that developed at the time

the experiment was scheduled to take place. The eight
remaining school districts appeared to be fairly repre—
sentative of the types of communities within the state.
The Grand Rapids study sample was not selected randomly
as above, but was arbitrarily chosen as an example of a

metropolitan school district.

Experimental Procedure

 

Prior to each experimental session, the author
compiled the above mentioned information packets and
response sheets so that each packet represented a treat-
ment condition (negative, positive, and behavioral) as
indicated by its top sheet, the teacher referral report.
These packets were then piled in random sequence by a
throw of a die, with enough of them in the pile to assure
one for each potential subject.

A room large enough to accommodate all of the
psychologists within a given school district was secured,
beforehand, through the efforts of the contact person in

that district. The room was usually a conference room

 

82

located within the office building that housed the school
psychologists' offices. Before each experimental session
the author placed a Panasonic, seven-inch, reel-to-reel
video-tape deck and 19-inch, black and white television
monitor in the front of the room. The equipment was
tested to make sure that it was in good working order

and the video-taped behavior segment was threaded and
pre-set to begin at the appropriate place.

After all of the participating school psycholo-
gists within that given school district had assembled,
the author thanked them for their cooperation and
explained that this experiment was designed to study
the relative importance of several variables involved
in the diagnostic process. They were told the experiment
would last for approximately 45 minutes and were given
the following instructions:

You will each be given a packet of information which
contains the case study material of a six-year-old,
first grade boy referred for psychological testing.
The top sheet of each packet shows a teacher's
referral which I would like you to read before look-
ing over the other material. Besides the referral,
you will find a short background information sheet
which gives the parent's impression of the boy at
home, some developmental history, and some comments
made by his kindergarten teacher last year. Follow-
ing that is a WISC-R protocol which I administered.
The test is already scored and I have included all
of the WISC-R record form pages so that you can look
over the quality of the boy's individual responses
if you wish. Following the WISC-R is a copy of the
boy's Bender Gestalt Test, a Kinetic Family Drawing,
and a composite of his Wide Range Achievement Test.
On this composite, I tried to show all of his actual
test responses on one sheet. If you have any
trouble understanding that sheet, just let me know

83

and I'll try to show you how to interpret it. I
tried to include tests from a basic_battery that
might be used by most school psychologists, but
just in case, is there anyone who is not familiar
with one of the tests I mentioned? Following
the test protocols, your information packet has
three response sheets on which I'd like you to show
your evaluation of the boy, but before you do that
you will want to see a piece of video-tape I have
of him in his classroom. For now, just take a look
at his referral and the other data and let me know
when you are ready to see the tape. I'd appreciate
it too, if you would not discuss the case with each
other until after we are finished and your packets
have been handed in. I'm interested in your inde-
pendent judgments and your packets contain slightly
different information so discussion with each other
would be irrelevant to your case anyway.

The information packets, each constituting one of the
three treatment conditions, were then handed to the sub-
jects in the randomized order that had previously been
arranged. The subjects were observed to make sure that
the teacher referral report was read before anything
else. When the group had finished reading the historical
and test data, the author gave the following explanation
of the video-tape:

What you are about to see is about 10 continuous
minutes of this boy in his classroom. I did the
video-taping from a front corner of the room with
a zoom lensed camera so although it may appear at
times that the camera is very close to the boy,
it's only the zoom lens that makes it look that
way. Although he is obviously aware that I am
filming, he does not know that he is the object of
interest. This video—taping was done before he
was seen for testing and as far as the boy is con-
cerned, I am just taking pictures of his whole
class. The portion of his behavior you are about
to see occurred just after the morning recess at
about 10:30. We had already been taping for about
an hour and a half so the kids were pretty accus-
tomed to the camera and me by this time. The
setting is a traditionally structured, first grade,
suburban classroom in my school district, Grandville.

84

This school is located in a middle-class neighbor-
hood and the kids are mostly from middle-class homes.
The kids are all working on different assignments
and the empty seats you will see belong to students
who are in the front of the room with their teacher
in a reading group. Our boy is working on a math
seatwork assignment and his teacher and I both agree
that the behavior you will see is fairly typical

of his classroom behavior in general.

Following the above explanation, the video-tape
was turned on and the subjects viewed it without comment.
After the tape was shown the subjects were asked to fill
in their evaluations of the child on the three response
sheets located at the bottom of their information packet.
Questions were handled by whispering the consultation to
the subjects who asked them, except when the question had
to do with the intent of a response item. In this case
the explanation was given to the entire group. The
response phase of the experiment usually lasted for
approximately 15 minutes, after which time the packets
were collected and the school district name plus a sub-
ject number were recorded on each one. In most cases,
the group then discussed the case and the experiment.

Questioning of each subject, at this time, revealed that

no one guessed the actual intent of the experiment.

Experts' Judgments

 

To gain an unbiased and independent opinion
regarding the functioning of the child to be evaluated
by the experimental subjects, the author conducted an

experimental session with four highly regarded school

 

 

85

psychologists in the same way as explained above, but
without exposing them to the teacher referral reports.
These school psychologists were not aware of the purpose
of the study and were only told that their expert opinions
were necessary to "standardize the experimental instru-
ments." The average experience level for this group was
12 years. Three of the four were past officers in the
Michigan Association of School Psychologists and two of
them had been directors in the National Association of
School Psychologists.

The responses of these experts showed that three
of the four diagnosed the child as "not eligible" within

the "Mandatory Special Education" classification system

(i = 1.25, s.d. = .433). Nor did three of them recommend
any Special education intervention (x = 1.5, s.d. = .866)
or medication (i = 1.75, s.d. = .433). The fourth expert

diagnosed emotional impairment under the "Mandatory"
classification system and recommended that the child
receive special education instructional materials and
consultative services to the teacher (see Appendix H).
Three of the four wrote in a mildly pathological diagno-
sis within the "classical" diagnostic system (E = 1.75,
s.d. = .433). Their "severity of pathology" scores
ranged from 15.5 to 20.5 with a mean of 17.5 and standard

deviation of 2.121.

86

gypotheses

 

The hypotheses tested in this experiment were
concerned with the effects of biased teacher referral

reports across four dependent measures.

gypothesis 1:

 

School psychologists will attribute significantly
greater pathology to a child when the teacher's
referral of him describes his behavior in negative
and pathological terms than when that referral
describes his behavior in positive or behavioral
terms.

Hypothesis 2:

 

School psychologists will diagnose a child as
having a handicapping condition within the Michigan
Mandatory Special Education diagnostic classifi-
cation system with significantly greater frequency
when the teacher's referral of the child describes
his behavior in negative and pathological terms
than when the referral describes his behavior in
positive or behavioral terms.

Hypothesis 3:

 

School psychologists will place a child within a
clinically pathological diagnostic category with
significantly greater frequency when the teacher's
referral of the child uses negative and pathological
descriptive terms than when the descriptive terms
are positive or behavioral.

Hypothesis 4:

 

School psychologists will rate a child as needing
significantly more special education intervention
when the teacher's referral of the child uses nega-
tive and pathological terms than when the terms are
positive or behavioral.

87

The relationships and associations among the
variables measured in this study (theoretical orien-
tation, length of experience, confidence, severity of
pathology estimate, "Classical" diagnosis, "Mandatory"
diagnosis, Special education treatment recommendation,
and referral for medication) were examined for agreement
with relationships that would be expected on the basis
of previous studies. The data for these samples of
school psychologists and the significance tests based
upon these data cannot be regarded as a test of expec-
tations for any real population since these samples have
been treated experimentally and are therefore not repre-
sentative of a real population. Thus relationships
found in these samples can only be considered as sug-
gestive of relationships for which further evidence
might be sought in future studies.

The relationships of special interest in this
study were those between school psychologists' length
of experience and theoretical orientation, on the one
hand, and their clinical judgments, such as severity of
pathology estimates, "Classical" diagnoses, "Mandatory"
diagnoses, and special education treatment recommen-
dations, on the other. In particular, a positive cor-
relation or association was expected between experience
and theoretical orientation, on the one hand, and the

various examples of clinical judgment, on the other,

88

eSpecially in the negatively biased teacher referral con-
dition. Thus, as was shown by Langer and Abelson (1974),
school psychologists with increasing degrees of psycho-
dynamic theoretical preference were expected to show
increasingly pathological clinical judgments when they
were exposed to negatively biased teacher referral
reports. This relationship, however, was not to be
expected when they were exposed to positive or behavior-
ally oriented teacher referrals.

Although past studies have not shown clearly what
to expect with regard to the experience variable, this
study was sufficiently similar to the Temerlin (1968)
experiment to anticipate similar results. That is,
increasing experience was expected to be associated with
increasing pathology in the subjects' clinical judgments
when they were exposed to the negatively biased teacher
referral condition. It was reasoned that inexperienced
school psychologists could be expected to be influenced
more by actual test data than by teacher referral reports
since their more recent experience with tests would pre-
dispose them to feel more comfortable with this type of
data than with the teacher reports. It was also reasoned
that inexperienced school psychologists might be somewhat
more reticent to diagnose pathology than their more
experienced colleagues since such a diagnosis may be

more fearsome to them.

 

 

 

 

89

All of the dependent variables ("severity of
pathology," "Classical" diagnosis, "Mandatory" diagnosis,
special education treatment recommendation, and referral
for medication) were expected to be related or associated
with each other. Since all of these variables were, in
one form or another, measures of pathology; logic would
dictate their relationship. However, due to the unrelia-
bility of traditional diagnostic systems, the "Classical"
diagnostic measure in this study was not expected to show
the same degree of inter-relatedness as the other depen-
dent variables. The "Mandatory" diagnostic measure and
the special education treatment recommendation were
expected to be closely associated since their agreement
is required by law (Michigan Mandatory Special Education

Act, 1971).

Design and Analyses
The basic design for this experiment was a three
level single factor, fixed effects model with four depen—
dent measures (Figure 3.l). The design called for random
assignment of subjects to one of three fixed expectancy

or treatment conditions (T -T3) within which they were

1
nested.

A test of the four hypotheses listed in the above
section was accomplished by multivariate analysis of

variance and subsequent one-way univariate analyses of

variance on each of the four dependent measures. These

90

Dependent Measures

Treatments Subjects M1 M2 M3 M4

 

S1

T . n=l7

S17

 

818

Design over ' _
Subjects T2 ' n-lS
S32

 

S33

T3 . n=16

S48

 

 

 

 

 

 

 

 

Tl: Negative teacher referral condition
T2: Positive teacher referral condition
T3: Behaviorally oriented teacher

referral condition

Ml: "Severity of pathology" measure

M2: "Mandatory Special Education"
diagnosis

M3: "Classical System" diagnosis

M4: Special education treatment recom-

mendation

Figure 3.1. Experimental Design

91

analyses were performed with the data from both the
larger random sample of school psychologists (48 subjects)
and the Grand Rapids (12 subjects) replication group.

In order to examine relationships among the inde-
pendent and dependent variables in this experiment,
several additional analyses were carried out. School
psychologists' theoretical orientation and length of
professional experience (both continuous variables) were
correlated, by the Pearson product moment method, with
all other variables. A correlation matrix was generated
separately for each treatment group, the Grand Rapids
group, and for the entire randomly selected sample of 48
school psychologists with the treatment groups combined.
In situations where a continuous variable was being
associated with a dichotomous variable, such as was the
case with the two diagnostic labeling measures, the
values generated by the product moment method were
equivalent to point-biserial correlation values. It
should be noted that the point-biserial correlation
method assumes a true dichotomy in the variables pre—
sented as such and not a dichotomy that is artificially
imposed. In this case, the dichotomy between a diagnosis
of normalcy versus pathology was assumed to be true since
a child in actual practice will receive one of these
diagnoses. The practice of diagnostic labeling is nomi-
nally scaled and is not based upon a normally distributed

continuous variable.

 

92

In addition to the four dependent and two inde-
pendent variables discussed above, several other variables
were measured and analyzed in this study. The school
psychologists' willingness to refer a child for medi-
cation and their degree of confidence in their judgments
were also included in the above correlation matrices.
When two dichotomous variables were associated, Fisher's
Exact Tests and Chi Square Tests were performed depending
upon sample sizes. Finally, the effects of training type
(clinical versus school psychology) and school district
upon two dependent variables (severity of pathology and
special education treatment recommendation) were analyzed

through separate univariate analyses of variance.

 

 

CHAPTER IV

ANALYSIS AND DISCUSSION OF RESULTS

Chapter IV will begin with a presentation of the
statistical analyses for the hypotheses of this study.
This section will be followed by an analysis of the
relationships and associations shown among all of the
variables measured. The results of the Grand Rapids
study will then be reported and, finally, a summary and

general discussion of the results will be presented.

Analyses for Hypotheses

 

The basic and preliminary statistical test for
all four of the hypotheses stated in Chapter III was a
multivariate analysis of variance. This analysis allowed
for a test of the effects of the teacher's referral
language upon four dependent measures taken together.
The results showed an F-ratio of 1.816 which, with 8
and 84 degrees of freedom, could have occurred by chance
with a probability of .085. The data, then, failed to
support all four of the hypotheses, each associated with
one of the dependent measures, at the more acceptable

.05 level.

93

94

Table 4.1 shows the results of the univariate
analyses of variance carried out on the same data. As
would be expected from the nonsignificant multivariate
analysis above, none of the univariate analyses achieved
significance; however, it appears that the "severity of
pathology" scale may have been the most affected of the

four dependent measures by the treatments.

Table 4.1

Random Group Univariate Analysis of Variance

 

 

Variable giggtggfiiie F-ratio P

Hl "Severity of Pathology" 42.108 2.460 .096

H2 "Mandatory" Diagnosis .027 .114 .892

H3 "Classical" Diagnosis .475 1.992 .148
H Special Education Treat—

ment Recommendation 3.630 1.186 .315

 

Degrees of freedom for hypotheses = 2

Degrees of freedom for error = 45
Table 4.2 bears out this speculation, and also shows that
the amount of variance on this measure precludes any sig-
nificance that could be attached to differences between
the treatment condition means.

An interesting trend also appears to have occurred.
Although not statistically significant, the behaviorally
oriented teacher referral treatment condition has drawn

higher estimates of pathology from the subjects on all

95

four dependent measures than the negative teacher referral.
This trend was not expected since the behaviorally oriented
teacher referrals were originally intended to be a control

condition.

Table 4.2

Random Group Cell Means and Standard Deviations

 

Treatment Conditions

 

Variable

 

 

 

 

Negative Positive Behavioral
i s.d. i s.d. Q s.d.
"Severity of
Pathology" 12.03 (3.52) 11.57 (4.37) 14.59 (4.51)
"Mandatory"
Diagnosis 1.29 ( .47) 1.33 ( .49) 1.38 ( .50)
"Classical"
Diagnosis 1.59 ( .51) 1.40 ( .51) 1.75 ( .45)
Special Educa-
tion Treat-
ment Recom-
mendation 2.53 (1.77) 3.47 (1.77) 3.13 (1.71)

 

Analyses of Relationships and
Associations

 

 

Table 4.3 shows the correlations and associations
between the variables of this study for the combined data
of all three treatment groups (48 subjects). These cor-
relations and associations are also shown within the nega-
tive teacher referral condition (17 subjects) on Table 4.4,
within the positive teacher referral treatment condition
(15 subjects) on Table 4.5, and within the behaviorally

oriented teacher referral treatment condition (16 subjects)

96

 

 

«a
mo. v m
k.
HOOonunm HOOoHQ MOQoHQ
««m>m. «*mmv. mmo. Goflumofipmz How Hmuummmm
Hoo.ud Hoo.ud oee.ud
.«mmm. mesme. emo.- mamddddea =suoummcmz=
mao.ud ome.ud mm~.nd
«mam. «esm. mme. memocmmeo =Hmommmmau=
¢H0.um mvv.um COAHMGsGEEoomm ucmE
«Hem. mao.| upmmue cowumospm HMflommm
Hoo.um vmm.nm
«mane. oeo. gmmoeonumm mo suenm>mm=
Hee.ud ema.nd
NNo. Hwa. mocmpflmaoo
moo.um
eeoem. doeumudmeuo HmomumMOdae
cod 8 cwfifioom
wmmflummue m mmoaonumm mocmtamdoo coaumucmeuo dogmaummxm
coflumOSCM . HMOflumuomnB .
amaommm wpfluo>mm ,

 

Amen: “m9 paw .NB .HBV Umswnfioo mmsouw udeummHE

Ham nufl3 mam>mq oOCmOHwHGmHm cam .mcoHHMHoommm .mdowuwamhnoo msouw Eoccmm

m.v mHQMB

 

97

Ho.vm

««

 

 

«Ne.ud omo.ud mam.ud emo.ud Hee.ud
omm. .oem. mNH.I «Gee. mme. aoeumoecmz mom Hmuummmm
Hee.ud mmo.ud eoe.ud emm.ud ee~.ud
«ewe». .eee. emo.u wee. mme.n mHmOdmmma =>Houmcemz=
mao.nd moo.ud see.ud mme.ud dam.nd
momm. mum. mmo.| «one. mma. memOdmmeo =Hmoemmmao=
avo.um mmH.um cam.nm 5mm.um coaumpcoEEoomm puma
«mme. mew. oma. meo.u spawns domumoscm Hmeomdm
mo~.ue emo.nd em~.ud
mew. mam. «se.- gsmoeoeumd mo mummm>om=
mH~.ud «ma.ue
mo~.u mmm. mudmemmdoo
mme.ud
Gem. doeumuammmo Hmoeumuomne
£04.. 0 QGEEOUM
wwwmummms m wmoaonumm coaumusmano
mo mocmpawcoo . . mocmflummxm
COmeUSCW wuHHmkrmm . HMOHflmHOGSB
Homommm .

 

ASH": .HBV cofluflccou ucmfiummua m>wummmz
may cflcpfls mam>mq wocmofimwcmflm cum .meHuMHOOmmfl .mGOHumeHHoo msouw Eoccdm

v.v MHQMB

98

 

 

*«
mo. v d
em
eee.ud eee.ud eme.nd ewm.ne wmw.ud
eeeee. eeewe. mee. ewe. ewe. cowumoeedz Mom emmummdm
eee.nd eee.ud ewe.ud Hee.ud wmw.ud
wmm. ewe. mem. men. eee. memoeemee =wmoumecmz=
« *e e ««
eme.ne eme.ud eee.ud ewe.ud eee.nd
eeee. «flee. eme. eemm. eee. memoeemwe =emowmmmeo=
Hoo.nm mvo.um mma.um mam.nm coaumpqmfieoomm pcmfi
eeeme. eeme. eew. eee. nummua cowumosem emwomdm
ewe.nd wee.ud mme.ud
wwm. eeem. eem. gmmoeonume mo mpeuo>mm=
mew.nd eem.ud
eee. wee.) mocoeemeoo
Hme.ue
wee. dowumpdmwmo emowpmuomee
.1
COHUNUCGEEOOGM
pawEummHB mmoamwumm mocmpemcoo coaumucwfluo mocmahmmxm
coHumospm muaum>mm . HMUHumuomce .
Hmeommm .

 

Amaucv :owuepcoo unofiummne m>Hvaom
map CHSUH3 mam>mq mudmoHMHsmem can .mcoquHUOmmfl .mcowumHmHHou macho Eopcmm

m.v manme

99

on Table 4.6. Each variable will be presented below
together with a discussion of its related or associated
variables when an uncorrected significance level of .01
or less was reached. The raw data frequency distribu-

tions for these variables are presented in Appendix J.

Length of Experience

 

The length of experience variable in the first
column on Table 4.3 (combined data group) shows a sig-
nificant relationship with only one other variable,
theoretical orientation (r = .37, p = .005). Thus it
appears that the more experienced school psychologists
in this sample tended to be somewhat more psychodynami-
cally oriented than their less experienced colleagues.
This relationship was strongest in the positive teacher
referral situation (Table 4.5, r = .493, p = .031) and
weakest in the behaviorally oriented teacher referral
situation (Table 4.6, r = .304, p = .126).

The expectation that school psychologists'
length of professional experience would show greatest
levels of relationship and association with the depen-
dent variables in the negative teacher referral treatment
condition was not realized in this study. An inspection
of Table 4.4 shows that the highest association reached
in this treatment condition was a very unimpressive
point-biserial correlation of .123 between the experience

variable and the "Classical" diagnosis measure.

-.«

100

 

 

««
mo.vm
.<.
Hee.ud wee.nd wee.ue eee.ud wem.ud
«meme. mmm. eee. eeeee. eee.- cOHumowemz mow ammummdm
eee.nd mww.ud www.ud eee.nd eew.nd
eewwe. mew. mew.: mew. ewe.u memocemwe =emoemecmz=
mme.nd mem.ud mme.ud Hee.ud mww.nd
eee. eee.u eeme.u . emw.- mew.- memoaemee =emowmmmeo=
emo.um mam.um moo.um omm.um somumpcmEEoomm Dame
eeem. eew.u eeemm. wee.- uummue coeumosem Hmwomdm
mew.ud mwe.ud emm.ud
eee.n emem. mee. =meoHoeumd mo muemm>mm=
ewe.ud mee.nd
mew. emee. deemeemcoo
ewe.nd
eem. cowumudmeuo emowemuooee
somumpcmfifioomm
ucmfiammne moodmmumm mocwpa coo soaumucwmuo mocmeummx
cowumospm .m Hmoapmuomne . m
mueum>mm .
Hamommm .

 

Amancv dowampcoo Hmnummwm Honomma Umucmfluo WHHMHOH>mnmm
one cflcuflz mam>mq mocMUAMHGmHm can .maoquHoommm .mcowuwamunoo msouw Eopcmm

m.v mHQMB

lOl

Theoretical Orientation

 

The combined data table (Table 4.3) shows that
the theoretical orientation variable, in addition to its
relationship with experience, was significantly related
to the subjects' estimates of pathological severity for
the child they evaluated (r = .471, p = .001). This
relationship was strongest in the positive teacher
referral group (Table 4.5, r = .579, p = .012) and
weakest in the negative teacher referral treatment con-
dition group (Table 4.4, r = .383, p = .064). Contrary
to expectation, the negative and pathological language
used to describe the child in the negatively biased
treatment condition did not produce the strongest evi—
dence of a relationship between these two variables.

Table 4.3 also reveals a significant association
between theoretical orientation and "Mandatory" diagnosis

(r .427, p = .001). Here again it was the positive

pbis =

treatment condition that showed the strongest association

between these two variables (r .785, p = .001),

pbis =
and the negative treatment condition (rpbis = .168,
p = .260), the weakest association. The referral for

medication variable was significantly associated with
theoretical orientation as well, with a point-biserial
correlation of .453 (p = .001). This association was
most strongly supported in the behaviorally oriented

referral situation (Table 4.6, r .711, p = .001)

pbis =

102

and was least supported in the positive referral situ-

ation (Table 4.5, .127, p = .327). The

rpbis =
behaviorally oriented teacher referral group (Table 4.6)
showed a significant relationship between theoretical
orientation and special education treatment recommen-
dation (r = .65, p = .003).

For the school psychologists in this study, then,
increasing agreement with the psychodynamic theoretical
orientation was related to higher estimates of pathology
and was associated with more frequent pathological label-
ing within the Michigan "Mandatory Special Education"
classification system than increasing agreement with a
behavioral theoretical orientation. This was especially
true when the teacher's referral of a child described him
in positive terms. It appears that the psychodynamically
oriented school psychologists were the ones most likely
to refer an overactive child for medication, especially
when the teacher's referral was expressed in behavioral
terms. In addition to the psychodynamically oriented
school psychologists' tendency to refer for medication,
they showed an increasing tendency to use special edu-
cation as a treatment when the referral of a possible

learning disabled boy was expressed in behavioral terms.

Confidence

 

Table 4.3 does not show the confidence variable

to be significantly related to any of the other variables

103

in this study. However, Table 4.5 shows a rather inter-
esting trend. In the positive referral situation, a
mildly positive relationship was shown between confidence
and special education treatment recommendation (r = .45,
p = .046). A positive association was also shown with

"Mandatory" Diagnosis (r .503, p = .028). Thus

pbis =
the psychologists in this treatment condition tended to
feel somewhat confident with their "Mandatory" diagnosis
of pathology and recommendations for increased amounts of
special education. In the behaviorally oriented situation
(Table 4.6), however, they tended not to feel confident
with their "Classical" diagnosis of pathology (rpbis =
.46, p = .036).

"Severity of Pathology"

 

The combined data correlation and association
matrix on Table 4.3 shows a significant relationship
between the "severity of pathology" measure and the special
education treatment recommendation measure (r = .525,

p = .001) besides the relationship with theoretical
orientation already presented. This relationship was
highest in the positive (Table 4.5) treatment condition
(r = .764, p = .001), but was well supported by all
treatment groups. The evidence is fairly strong, then,
that when estimates of pathology were high, subsequent

treatment was also recommended at higher levels.

104

The "severity of pathology" measure was also
significantly associated with "Mandatory" Diagnosis

(Table 4.3, r .416, p = .002). This association

pbis =

was also most strongly supported in the positive

(Table 4.5) referral group (r .626, p = .006),

pbis =
but did not reach significance in the behaviorally

oriented referral group (Table 4.6, r .205, p =

pbis =
.223). In the general sample the referral for medication
variable was significantly associated with "severity of

pathology" (Table 4.3, .498, p = .001). Again,

r . =
pblS
it was the positive referral group (Table 4.5) that most

strongly showed this association with a point-biserial
correlation of .626 (p = .006). Increasing estimates

of pathology, then, have shown association with increas-
ing frequency in the referral for medication of a pos-
sible learning disabled boy.

Special Education Treatment
Recommendation

 

 

Table 4.3 shows that in addition to its relation-
ship with "severity of pathology" already discussed, the
special education treatment recommendation measure was
significantly associated with "Mandatory" Diagnosis
(rpbis = .653, p = .001). This association was strongly
shown in all of the referral situations, but most sig-
nificantly in the behaviorally oriented referral situ-

ation (Table 4.6, = .722, p = .001) and the

rpbis

105

negatively biased referral situation (Table 4.4,
rpbis = .702, p = .001).
The referral for medication variable, true to
its previous behavior in this study, has also shown a
significant association with the special education
treatment recommendation variable in the combined data

sample (Table 4.3, r .573, p = .001). Although

pbis =
the positively biased referral situation (Table 4.5,
rpbis = .801, p = .001) and the behaviorally oriented

referral situation (Table 4.6, r .756, p = .001)

pbis =
showed relatively high degrees of association between
these two variables, the association was very weak in
the negatively biased referral situation (Table 4.4,
rpbis = .23). Increasing amounts of special education
intervention were associated with increasing willingness
to refer for medication except when the referral to the
school psychologist was negatively and pathologically
biased. In addition, increasing amounts of special edu-
cation intervention recommended tended to relate to
higher estimates of pathology and to be associated with
diagnoses of pathology in Michigan's "Mandatory Special
Education" diagnostic classification system.

"Classical" Diagnosis, "Mandatory" Diagnosis
and Referralifor Medication

 

 

Since the associations of the two diagnostic

measures and the referral for medication measure with

106

the continuous variables have already been reported in
previous sections, only the associations among these
dichotomous variables themselves are presented here.
Tests of association were used in these analyses because
significance levels were desired. Fisher's Exact Tests
were used in analyzing associations within the individual
treatment conditions because the small sample size
allowed for computations of exact probabilities within
the sample. These results are presented in Appendix K-2
for the negatively biased referral group, Appendix K-3
for the positively biased referral group, and Appendix K-4
for the behaviorally oriented referral group. The data
for all groups combined was analyzed by the chi square
and are presented in Appendix K-l. The 95% confidence
level was arbitrarily chosen as acceptable for signifi-
cance.

The association between the "Classical" diagnosis
and referral for medication measures was not significant.
However, the referral for medication and "Mandatory"
diagnosis variable were significantly associated with a
corrected chi square of 5.668 (p = .017). Almost 48%
of the psychologists who diagnosed no pathology in the
"Mandatory" classification system were also unwilling
to refer the child for medication. This trend was

generally seen in each of the individual treatment

107

conditions, but only approached significance (p = .059)
in the behaviorally oriented referral group.

The combined data sample also showed a significant
association between the two diagnostic measures. The cor-
rected chi square of association for these variables was
6.696, which, with one degree of freedom, was significant
at the .01 level. Significance in the association between
these two variables was also reached in the negative
referral (p = .041) and the positive referral (p = .047)
situations. However, the association was far from sig-
nificant in the behaviorally oriented referral situation
where agreement on the two diagnoses was reached by only
half of the sample.

In summary, the "Classical" diagnosis and "Man-
datory" diagnosis variables were in close agreement,
except in the behaviorally oriented referral situation.
Only the "Mandatory" diagnosis variable was significantly
associated with referral for medication, especially in
the behaviorally oriented referral situation.

Graduate Training in School
Versus Clinical Psyphblogy

 

 

The effects of graduate training were separately
analyzed on two dependent variables, "severity of path-
ology" and special education treatment recommendation.
The technique was a one-way fixed effects model analysis

of variance with the type of graduate training as the

108

"treatment" variable. Since one of the subjects had
training in both clinical and school psychology, his

data were deleted from this analysis. Most of the sub-
jects in this study received their graduate training in
school psychology (n = 36). The mean "severity of path-
ology" score for these subjects was 12.986, as opposed
to a mean score of 12.682 for the clinically trained
subjects (n = 11). Table 4.7 shows this difference to
be far from significant (p = .833). On the special edu-
cation treatment recommendation measure, school psychology
trained subjects achieved a mean of 3.167 as compared
with the mean of 3.0 for the clinically trained subjects.
Table 4.7 shows that the difference between these two
groups on the treatment recommendation measure was also

nonsignificant (p = .892).

Table 4.7

Analyses of Variance for Graduate Training

 

Degrees Mean F P

Source of Freedom Square

 

"Severity of Pathology" by Training

Main effects 1 .780 .045 .833
Error 45 17.431
Total 46 17.069

Special Education Treatment
RecommendationIby Training

Main effects 1 .059 .019 .892
Error 45 3.128
Total 46 3.061

 

109

School District

 

The "severity of pathology" and special education
treatment variables were also analyzed by school districts
to determine whether the school psychologists differed in
their responses to these two variables depending upon
where they worked. Table 4.8 shows the one-way fixed
effects analyses of variance for these two variables
with school district as the "treatment" or "main" var—
iable. As can be observed, both the "severity of path-
ology" and special education treatment recommendation
dependent measures were significantly different among
the various school districts with F-ratios of 3.33 and
6.17 respectively (p < .01). Thus it appears that the
school district within which the psychologists worked
made a significant difference upon their judgments of
pathological severity and especially upon their recom-
mendations for special education intervention. Table 4.8
also shows the mean scores on both dependent variables
by school district. As can be observed, the average
"severity of pathology" score ranged from 8.25 to 15.67
depending upon the school district. The average special
education treatment recommendation score ranged from
1.67 to 5.2. While it is difficult to translate the
"severity of pathology" mean scores into qualitatively
understandable entities, an examination of the hier-

archically arranged list of special education

110

Table 4.8

School District Groups Analyses

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Degrees Mean
Source of Freedom Square F P
Analysispgf Variance for the "Severity of
Pathology" Dependent Measure by School
DistrICt Groups
Main effects 7 44.969 3.333 .007
Error 40 13.493
Total 47 18.181
Analysis of Variance for the Special Education
Treatment Recommendation Dependent Measure
by School District Groups
Main effects 7 10.755 6.172 .001
Error 40 1.742
Total 47 3.085
Dependent Measures
Digfigiit n Special Education
"Severity of Pathology" Treatment
Recommendation
i s.d. E s.d.
School District Group Mean Scores
and Standard Deviations
1 4 12.75 1.479 3.25 .433
2 5 15.67 2.618 4.07 1.691
3 4 9.75 4.437 1.75 .829
4 7 11.21 4.199 1.57 .903
5 5 14.90 4.841 5.20 .979
6 3 12.50 2.828 2.67 1.247
7 6 10.50 3.316 1.67 .471
8 4 8.25 2.165 2.25 1.299

 

lll

interventions upon which the subjects responded (Appen-
dix H) does allow for this kind of translation. It was
possible, then, for the child being evaluated to have
received a range of services from special education
materials only to placement within a special education
resource room, depending upon his (and the psychologist's)

school district.

Grand Rapids Study

 

The multivariate analysis of variance, testing
for differences on the four dependent measures across
the three fixed types of referral situations, resulted in
an F-ratio of 4.540. With 8 and 12 degrees of freedom,
this ratio was significant at the .0097 level. Table 4.9
shows the results of the univariate analyses carried out
for each of the dependent variables. An examination of
this table reveals that only one of the F-ratios for the
individual dependent measures reached the .05 level of
significance necessary to reject its null hypothesis.
The "severity of pathology" measure reached this level
(p = .014) and the hypothesis for this variable as

stated in Chapter III was:

Hypothesis 1:

 

School psychologists will attribute significantly
greater pathology to a child when the teacher's
referral of him describes his behavior in negative
and pathological terms than when that referral
describes his behavior in positive or behavioral
terms.

112

Table 4.9

Grand Rapids Group Univariate Analyses of Variance

 

Hypothesis Univariate

 

Variable Mean Squares F-ratio P
Hl "Severity of Pathology" 68.723 7.099 .014
H2 "Mandatory" Diagnosis .025 .086 .918
H3 "Classical" Diagnosis .125 .794 .481
H Special Education Treat—
ment Recommendation 2.025 1.097 .375

 

Degrees of freedom for hypotheses = 2

Degrees of freedom for error = 45
The tests for this hypothesis were post-hoc comparisons
between the negative referral treatment condition and the
other two referral treatment condition means. Two tailed
t-tests were used for these comparisons because, with
the small sample size (n = 12), they were slightly more
powerful than techniques using the F statistic. Table
4.10 shows the means and standard deviations for these
data.

The 98% confidence interval for the difference
between the negative referral and behaviorally oriented
referral treatment condition means was 1.08 i 7.80 and
nonsignificant. The 98% confidence interval for the
difference between the negative referral and positive
referral treatment condition means was 7.43 t 7.14 which

was significant. Hypothesis 1, then, was only partially

ll3

supported by the data in this experiment. Only the dif-
ference between the negatively and positively biased
teacher referral conditions reached significance on the

"severity of pathology" measure.

Table 4.10

Grand Rapids Group Cell Means and Standard Deviations

 

Treatment Conditions

 

 

 

 

 

Variable Negative Positive Behavioral
i s.d. i s.d. :2 s.d.

"Severity of

Pathology" 17.33 (5.80) 9.90 (1.67) 16.25 (1.71)
"Mandatory"

Diagnosis 1.67 ( .58) 1.60 ( .55) 1.75 ( .50)
"Classical"

Diagnosis 1.67 ( .58) 2.00 ( .00) 1.75 ( .50)
Special Educa-

tion Treat—

ment Recom-

mendation 3.33 (2.08) 4.60 ( .89) 4.75 (1.25)

 

Tables 4.11 and 4.12 show the correlations and
associations between pairs of variables for this sample.
Raw data frequency distributions for these variables are
also presented in Appendix J. As can be observed, the
"Classical" diagnosis measure was significantly associ-

ated with the confidence measure (r .600, p = .020),

pbis =
and the special education treatment recommendation

measure (rpbis = .800, p = .001). The negative assoc1-

ation between "Classical" diagnosis and the "severity of

 

 

114

 

 

 

ee
mo. v m
a»
wem.nd eee.ud mwe.nd eee.nd mee.nd
wwe. wee. mew. wee.n ewm. dowpmomemz nee emummemm
eee.ne eee.ud wem.ud eee.ud wwe.ud
eem. eee.: eee.: emm. wmm. memoeemwe =wuoumeemz=
Hee.nd eee.nd ewe.nd Hew.ud mwm.ud
eee. eee.: eee. eee. wee.) memoeemwe eemowmmmeo=
¥¥ “5.1 .1
eme.ud eee.ud mew.nd wew.nd downmecmasoomm paws
emem.| ewe. eee. eee.: nummue dowumosem emwomdm
eee.ud eme.ue eee.ue
wem.u mmw.- eem. =we0eoeumm eo wuwum>mm=
eee.ud mew.nd
hemom. mma.l mocopwmcoo
eee.ne
wee. cowumuememo Hmuwnmuomee
GOHHMOCTEEOOGM >00
= Hoeumd
usmfiummua GOHUMpcmHHO
coapmospm muauwmmm mocwpwwcoo HMOHumHomse ecumenmmxm
ammommm . =
Awe": “we use .we .eee emdwnaoo mesome udmsnmmme Hem

npflz mHm>wA wocmommwsmwm paw .mGOH#MHUOmm¢ .mempmamnhoo muono mwflmmm pcmuw

HH.v magma

Table 4.12

115

Grand Rapids Group Fisher's Exact Tests with all
Treatment Groups Combined (n=12)

 

Referral for Medication

 

 

 

 

 

 

 

Row Total
yes no
"Classical" Diagnosis
n=1 n=1 n=2
Normal row % 50.0 row % 50.0 % = 16.7
col % 16.7 col % 16.7
tot % 8.3 tot % 8.3
=5 n=5 n=10
Pathology row % 50.0 row % 50.0 % = 83.3
col % 83.3 col % 83.3
tot % 41.7 tot % 41.7
Column Total =6 n=6
% = 50.0 % = 50.0 p = .773
"Mandatory" Diagnosis
n=2 n=2 n=4
"Not Eligible" row % 50.0 row % 50.0 % = 33.3
col % 33.3 col % 33.3
tot % 16.7 tot % 16.7
n=4 n=4 n=8
Pathology row % 50.0 row % 50.0 % = 66.7
col % 66.7 col % 66.7
tot % 33.3 tot % 33.3
Column Total n=6 n=6
% = 50.0 % = 50.0 p = .727
"Classical" Diagnosis
"Mandatory? Diagnosis Normal Pathology Row Total.
n=1 n=3 n=4
"Not Eligible" row % 25.0 row % 75.0 % = 33.3
col % 50.0 col % 30.0
tot % 8.3 tot % 25.0
n=1 n=7 n=8
Pathology row % 12.5 row % 87.5 % = 66.7
col % 50.0 col % 70.0
tot % 8.3 tot % 58.3
Column Total n=2 n=10
% = 16.7 % = 83.3 p = .909

 

 

 

 

116

pathology" measure (r = -.660) in this group is par-

pbis
ticularly interesting since it means that increasing
estimates of pathological severity were associated with
a diagnosis of normality in the "Classical" diagnostic
classification system. The "severity of pathology"
measure was also negatively correlated with the special
education treatment recommendation measure (r = -.543,

p = .034). Inspection of the data revealed that only
two subjects in this group diagnosed normality within
the "Classical" system. These two subjects averaged a
"severity of pathology" score of 20.25 as compared with
a mean score of 13.88 for the entire group. These two
subjects also recommended less special education inter-
vention than any of the other members of their school
psychology staff. These associations, then, rather than
being a reflection of the general trend in Grand Rapids,
were based primarily upon the behavior of two psycholo-
gists who were somewhat atypical of their group.

In summary, the results of this replication study
showed that negatively, positively, and behaviorally
oriented referrals produced significant (p < .05) dif-
ferences among the subjects' clinical judgments concern-
ing the child they evaluated. Only the "severity of
pathology" measure appears to have been significantly
affected by the referral conditions and post-hoc analyses

showed significant (p < .02) differences between the

ll7

negative and positive treatment condition means for this
measure. However, the differences between the negative
and behaviorally oriented treatment means failed to reach
this level (p > .02). Most subjects diagnosed the boy
they evaluated as pathological within the "Classical"
(83.3%) and "Mandatory" (66.7%) diagnostic systems and
almost all of the subjects (92%) believed he ought to
have some special education help. A discussion with
several of the subjects, after théyexperiment, revealed
that many of the psychologists in the school district
feel pressure from their administration to help place
children in special education programs. The behavior of
these school psychologists, then, may have been more a
reflection of the special education philosophy within

their school district than anything else.

Summary and Discussion of Results

 

The four hypotheses for these experiments pre-
dicted that school psychologists' evaluations of a pos-
sible learning disabled boy would be significantly
affected by a negatively biased teacher referral report.
Each hypothesis made this prediction in relation to one
of four dependent variables measuring the subjects'
estimations of pathological severity for this youngster
(H1), their choice of diagnostic label within the

Michigan "Mandatory Special Education" (H2) and

 

118

"Classical" (H3) diagnostic classification systems, and
their estimation of the amount of special education the
boy would require (H4).

The results of the study conducted with 48 prac-
ticing school psychologists from eight randomly selected
school districts in lower Michigan failed to support
any of the four hypotheses. The results of the same
study conducted with 12 practicing school psychologists
in the Grand Rapids Public School District did show sig-
nificant differences among the three types of teacher
referral treatment conditions. Subsequent analyses
showed these differences to be significant only on the
subjects' estimates of pathological severity for the
child they evaluated. This finding only partially sup-
ported the hypothesis that the subjects would estimate
more pathology when exposed to a negatively biased teacher
referral than when exposed to either positively biased or
behaviorally oriented referrals. Although the differ-
ence between the negative and positive treatment con-
dition means was significant (p < .02) on this measure,
the difference between the negative and behaviorally
oriented treatment condition means was not (p > .02).
Several possible reasons for the negative findings of
the major experiment are advanced below.

Brophy and Good (1974) have suggested that many

failures to demonstrate expectancy effects may be due

 

 

119

to a failure in the induction of the expectancy itself.
Since expectancy, in these two experiments, was induced
through a biased, written referral report, it may be that
personal contact with the referring teacher would have
more forcibly induced the expectancies. Or, it may be
that school psychologists are influenced only by certain
teachers with whom they are acquainted. Since the refer-
ring teacher was unknown to the subjects in these experi-
ments, perhaps any chance of expectancy induction was,
therefore, removed. It is also possible that school
psychologists simply do not give much credence to a
teacher's judgments concerning a child; and, therefore,
although the referral reports were noted by the psycho-
logist subjects, they were not influenced by them. It
should be remembered, too, that all three referral
reports used as treatments in these studies described

the same behavior of a child and differed only in the
language used to describe the behavior. No false pre-
evaluation information was used to induce expectancy,
only biased information was used. Most expectancy studies
have used false information to induce expectancy, and
therefore the induction method in these two experiments,
although more realistic, may be considered to be somewhat
weaker than other experiments. Further, it is possible
that the positive teacher referral condition (Appendix A-2)

may not have induced truly positive expectations. The

120

author may have failed to write a plausible "reason for
referral" in this case, since such glowingly described
children are rarely referred for testing in actual prac-
tice. Some past experiments have surmounted this problem
by suggesting that the reason for referral was "to evalu-
ate the child for possible placement in an accelerated
class, or class for gifted children." Although this
method may produce greater expectancy effects, it was

not used here because it is not only false, but perhaps
is too unfair to the subjects as well.

These studies may have fallen victim to the very
thing they were designed to measure. All of the experi—
mental sessions were conducted personally by the author,
who is himself a school psychologist. Although consid—
erable care was taken to avoid it, the author may have
inadvertantly communicated an unconscious desire for his
colleagues to resist the effects of expectancy and bias.
Certainly the author would like to believe that he could
resist such effects himself and perhaps this desire was
projected to his colleagues as well.

The experience level of the subjects may also
have been responsible for the lack of positive findings
in these experiments. Although the subjects ranged from
interns to 18-year veterans, even the interns were con-
siderably more experienced in evaluation than the subjects

of most other studies finding expectancy effects.

121

Ingraham and Harrington (1966) and Brophy and Good (1974)
have each suggested that well-trained and experienced
subjects may not allow their expectancies to influence
their behavior as much as their less experienced counter—
parts.

Expectancy effects, in these studies, were measured
on what BrOphy and Good (1974) have termed "product" or
outcome measures. Since they have shown that expectancy
effects are much more likely to be observed by "process"
measures or, in this case, school psychologists' inter-
active behavior, the fact that "process" measures were
not used in these experiments may have lessened the pro-
bability of positive findings.

The sex variable may also have contributed to the
failure to find expectancy effects in the main study.
First, it is possible that the findings would have been
different if the child to be evaluated had been a girl
instead of a boy. Certainly the referred child's behavior
would be considered more unusual for girls, and this
novelty might have interacted with the different types
of referral to produce significant expectancy effects.
Secondly, the video-tape showed the referring teacher
to be a woman. Some studies have shown that females may
be less able to exert expectancy effects upon their sub-

jects (Rosenthal, 1966, 1976) and thus if the referring

 

 

122

teacher had been male, greater expectancy effects may
have been observed in these experiments.

There is some anecdotal evidence to suggest,
however, that significant expectancy effects did not
occur because the subjects revised their expectations
in the face of disconfirming evidence. Several subjects
commented after the experimental session that the refer—
ring teacher was incorrect in her description of the
child, and many suggested that the teacher should be
counseled regarding the proper management of active
children. Several of the subjects wrote comments on
the back of their response sheets such as " . . . discuss
with the teacher to remove moral tone of some words—-
clarify what hyperactive really means," and "She needs
help in aiding him to channel excess energy," and also
"Inform the teacher that this is ip.pg_way a boy who
could be considered for special education placement."

Despite the suggestion of several authors that
less experienced experimenters and teachers show the
greatest expectancy effects, it was anticipated that
the most experienced subjects in these two experiments
would show the greatest expectancy effects. The reasons
for this departure were, first, that their suggestion
was based on studies that have measured expectancy by
observing its effects in a second party, a procedure

not employed in these two experiments. Secondly,

 

123

Temerlin's (1968) study, which was somewhat similar to
these experiments, found highly significant bias effects
in the best trained and experienced clinicians and few
of these effects in less experienced and trained sub—
jects. Although a direct test of the effects of exper—
ience was not possible in these experiments, the data
showed very low correlations and associations between
experience and the dependent variables. These results
coincide with other researchers who have not found
experience to be a critical variable (Sattler & Theye,
1967; Hersh, 1971; Auffrey & Robertson, 1972).

School psychologists' theoretical orientation
(behavioristic versus psychodynamic) was also expected
to be significantly related to their evaluation behavior
on the four dependent measures when they were exposed to
a negatively biased teacher referral report. The results
of the main experiment failed to support this expectation.
The correlations and associations between theoretical
orientation and the four dependent measures failed to
reach an acceptable significance level (p > .01) in the
negatively biased referral treatment condition. Thus
the rather powerful expectation effects found with psy-
chodynamically oriented clinicians by Langer and Abelson
(1974) were not duplicated in this experiment. Perhaps
the subjects in this experiment were not as uniformly
extreme in their theoretical preferences as Langer and

Abelson's subjects.

 

 

 

124

This study did, however, find a broad, general
relationship between theoretical preference and several
dependent measures, but these relationships were most
significant in treatment conditions other than the nega-
tive condition. It appears that increased agreement with
the psychodynamic perspective was related to an increased
tendency to diagnose pathology (especially within the
"Mandatory" classification system), increased estimates
of pathological severity, and an increased tendency to
refer a child for medication. The relationship of theo-
retical orientation with the dependent variables appears
to have been somewhat different in each type of referral
situation. For example, the behaviorally oriented
referral situation seems to have produced the lowest
associations between theoretical orientation and diagnos—
tic measures, but the highest with the referral for med—
ication and special education treatment recommendation
measures. It is difficult to make sense out of these
differences, except that it is possible that behaviorally
oriented referrals produced a need, on the part of more
psychodynamically oriented psychologists, to take some
action and a low need for the behaviorists to do so.
Positive referral situations, on the other hand, may
have produced little need for action, but more of a

proclivity for pathological diagnoses on the part of

 

 

125

the psychodynamically oriented psychologists (and more
inclination toward normal diagnoses by the behaviorists).

Despite Tolor and Brannigan's (1976) finding
that their sample of school and clinical psycholOgy
trainees differed on such variables as cognitive style,
perception of clients, conception of professional role,
personality, sex, age, and kind and amount of previous
experience; this study found no difference between the
two groups, both practicing as school psychologists.
Whether the school psychologists in this experiment had
their graduate training in school psychology or clinical
psychology did not appear to have made any difference
in their estimates of pathological severity or their
estimates of the amount of special education needed
for the child they evaluated. It is possible that dif-
ferences might have occurred if the clinically trained
subjects had been working in a clinical setting. Since
all of the subjects worked in schools, however, any dif-
ferences that might have existed were likely attenuated
by common experience.

Generally, all of the dependent measures showed
significant relationships with each other. The pattern
of the relationships changed somewhat in each treatment
condition with the positive treatment condition showing
the closest correspondence to the pattern shown when all

data were combined. The "Classical" diagnosis measure

 

 

126

generally showed the lowest levels of association with
the other dependent measures. The unreliability of the
traditional diagnostic classification systems is well
documented (Goldfarb, 1959; Grosz & Grossman, 1964;
Szasz, 1961) and supported in this study as well. The
"Mandatory" diagnostic classification system, on the
other hand, was remarkably stable in its relationships
with other dependent variables throughout all treatment
conditions. Perhaps the reason for this stability is
the fact that this system's criteria for the various
diagnostic categories are somewhat more distinct than
the more traditional system.

The referral for medication variable was included
in these studies to gain some estimate of Michigan school
psychologists' referral practices in this controversial
area. The random sample of 48 school psychologists was
rather evenly split in their opinion as to whether or
not the boy they evaluated should be referred to his
physician for determination of medication possibilities.
Forty-one and seven-tenths percent indicated they would
refer while 58.3% indicated they would not. They were
also evenly divided on this issue within each treatment
condition, except in the positive condition, where, two
to one, they indicated they would not refer for medi-
cation. Thus, if the teacher's feelings about the child

had any impact on the psychologist's decision whether

127

to refer for medication, it would only have been the
teacher's positive feelings which helped him decide
not to refer.

The results of the behaviorally oriented teacher
referral report were a great surprise in these studies.
Although originally intended to be a control condition,
this type of referral actually showed higher estimates
of pathology in the main experiment and estimates nearly
as high as the negative referral condition in the Grand
Rapids experiment. Although these results were not sta-
tistically significant, they were, nevertheless, in an
unanticipated direction and may be indicative of a trend.
If these results do reveal trends, they are difficult
to understand, since the experiments did not include a
control group which was not exposed to a teacher referral
report. On the one hand, it is possible that the behav-
iorally oriented referral reports did not produce a con—
trol situation and were viewed fully as negatively by
the subjects as were the negatively biased referral
reports. There could have been several reasons for a
negative reaction to the behaviorally oriented teacher
referral. Examination of the referral report (Appendix
A—3) shows that the report tends to focus primarily upon
negative behavior. Perhaps the subjects were only
responding to this focus. It is also possible that the

frequencies of behavior reported in this referral really

 

128

do indicate pathology. The absence of published
behavioral frequency norms dictated that the frequen-
cies had to be invented entirely by the author and they
may have suggested more pathology than was intended.
The behavioral referral, since it did not contain evalu-
ative or inferential statements, could also have functioned
more as a screen for the subjects' projections than the
other referrals. Since it is probable that most psycho-
logists have a general bias toward pathology, this refer-
ral may have allowed some projective fruition of this
general bias. On the other hand, perhaps the behaviorally
oriented referral reports truly functioned as a neutral
control situation. In this case, the negative referral
reports were somewhat less powerful in producing expec-
tancy effects than the positive referral reports. This
interpretation would be in agreement with the results
of Rosenthal and Fode (1961, 1963) and Egeland (1969)
which showed relatively more powerful halo effects than
negative expectancy effects. The case for a halo effect
trend also gains partial support by a comparison of the
"severity of pathology" mean scores for the two experi-
mental groups with the group of experts' mean score of
17.5 on this measure. The experts, it will be remem—
bered, were not exposed to the teacher referral reports.
Perhaps one of the major findings here is that

the school psychologists in this study functioned more

 

129

similarly within each school system than between school
systems. This was shown by the significant results of
the analyses of variance for the eight school districts
in the major experiment and by the fact that the experi-
ment performed with a single school district (Grand
Rapids) could show significant treatment effects when

the same experiment performed across several school
districts did not. There are several possible reasons
for these significant differences between school districts.
The differences could have been caused by an unequal dis-
tribution of the treatment conditions within the school
districts. An examination of the data revealed that

six of the eight districts received relatively equal
frequencies of each treatment, but two of the districts
received only two of the treatments. School district
number one on Table 4.8 did not receive the positive
teacher referral treatment, and district number eight
did not receive the behaviorally oriented referral treat-
ment. The absence of one of the treatments may have
reduced the variance of scores within these school dis-
trict groups which, in turn, could have added to the
F-ratio in the analyses of variance. School district
number one had particularly low standard deviations on
both dependent measures. Considering the insignificant
F-ratios.for the univariate analyses of variance across

treatment conditions (Table 4.1), this possibility seems

 

130

somewhat remote. There is no reason to assume that the
treatment conditions made any real difference in sub-
jects' responses, especially on the special education
treatment recommendation variable. This variable showed
the most significance on the analysis of variance across
school districts (Table 4.8).

Another possible reason for the differences
among the school districts is that the experimental
session or the events surrounding the experiment were
somehow different for the school districts. Since each
experimental session was administered separately at a
different time and place, over a six-week period, any
number of external events could have caused psychologists
within those sessions to react differently to the experi-
ment. If differences in external events existed, how-
ever, the author was unaware of them. The possibility
that some of the sessions were conducted differently also
seems unlikely because each session was administered by
the same experimenter (the author) and the administration
was rather carefully standardized.

Three explanations for the differences among
school district groups seem most plausible. One is that
people who work together for a period of time may
mutually influence each other to think and behave in
similar ways. The literature on group processes would

certainly support this notion. It is also reasonable

 

131

to suppose that the influences and pressures upon school
psychologists may be somewhat different in each school
district. The comments of several of the replication
study subjects suggested some pressure from their special
education administration to diagnose and place large
numbers of children in special education programs. If
this pressure truly existed, it may have accounted for
the relatively high estimation of pathology and greater
amount of special education treatment recommended within
that group of psychologists, and perhaps some of the
other groups as well. Finally, it is possible that
certain of the school districts sampled contain such
large numbers of deviant children that the subjects
within those districts viewed the study's experimental
child as needing less help than subjects who worked in
districts with fewer deviant children. Thus, it is
possible that local norms were influential in producing

differences between the school districts.

 

CHAPTER V
SUMMARY, CONCLUSIONS, AND IMPLICATIONS

Summary

Two experiments were conducted to test the effects
of biased teacher referrals upon the diagnostic judgments
and treatment recommendations of Michigan school psycho-
logists. In the main experiment, 48 school psychologists
from eight randomly selected school districts were ran-
domly exposed, in school district groups, to one of three
types of teacher referral reports. Each referral report
described the same child, but differed in the language
used for this description. The first referral report
described the child in negative and pathological terms
such as "hyperactive, aggressive, and impulsive"; while
the second positively oriented referral report described
him in terms like "energetic, assertive, and spontaneous."
A third behaviorally oriented referral reported only the
child's observable behavior. After reading one of the
three referral reports, all of the subjects studied an
identical information packet which presented the history

and test protocols of an active six-year—old boy in the

132

 

 

133

first grade. All subjects also viewed the same video-
taped recording, about 10 minutes in length, of the boy
in his classroom. Following the viewing of the video-
tape, the subjects were asked to rate the child's func-
tioning on nine continua designed to measure pathologi—
cal severity. They also selected diagnostic labels from
two lists taken from the Michigan "Mandatory Special
Education Act" and clinical or classical diagnostic
classification systems, each of which included a normal
or psychologically healthy category. To gain a measure
of the amount of change that they would recommend in the
child's school program, the subjects were asked to select
a treatment procedure from a hierarchically arranged
list of special education interventions ranging from no
special help at all to full-time placement in special
education classrooms. Finally, the subjects were asked
to indicate whether they would refer the child to his
physician for determination of medication possibilities
and to rate the degree of confidence they felt in making
their judgments. Several independent variables were
measured including the subjects' theoretical orientation
(psychodynamic versus behavioral), length of profes-
sional experience, and type of graduate training
(school psychology versus clinical psychology).

Four hypotheses were advanced, each of them

predicting that the negatively biased teacher referral

134

report would influence the school psychologist subjects
to overestimate the child's pathology on one of four
dependent measures. The four dependent measures were:
the "Severity of Pathology" scale, "Mandatory Special
Education" diagnosis, clinical or "Classical" diagnosis,
and the special education treatment the school psycholo-
gists decided to recommend.

The results of the main experiment failed to show
significant differences among the three types of teacher
referral treatment conditions on any of the four dependent
measures. The same experiment, administered to a group of
12 school psychologists all employed by a medium-sized
city school district, however, did result in significant
(p < .05) differences in the subjects' estimates of
pathological severity for the child they evaluated. The
negative referral condition produced significantly
(p < .02) higher estimates of pathology than the posi—
tive referral condition, but not higher than the
behaviorally oriented referral condition.

An analysis of the correlations and associations
among the variables measured in the main experiment
revealed that as the subjects increased in experience,
they tended to be somewhat more in agreement with the
psychodynamic theoretical orientation (r = .37, p = .005).
Contrary to expectation, the experience variable was not

appreciably related to any of the dependent variables.

 

 

 

 

 

 

 

 

 

 

135

The "Severity of Pathology,‘ "Mandatory Special Education"
diagnosis, "Classical" diagnosis, special education treat-
ment recommendation, and referral for medication variables
all appeared to be in general agreement with each other.
However, the "Classical" diagnosis measure agreed with
the other variables at a somewhat lower level, perhaps
reflecting the well-documented unreliability of tradi—
tional and clinical diagnostic classification systems.
The confidence measure was not even moderately related
to any of the variables measured by this experiment.
Interestingly, although the theoretical orienta-
tion variable showed moderate overall correlations and
associations with the dependent measures, the pattern of
these correlations and associations changed rather
markedly within each type of teacher referral situation.
The patterns of changes were quite complex and an under-
standing of why they occurred could not be determined
with confidence. The positive teacher referral, rather
than the negative one, showed the highest correlations
and associations between theoretical orientation and
the dependent measures. The direction of these relation-
ships was that increasing levels of psychodynamic orien-
tation were associated with increasing estimates of
pathology and recommendations for special education

intervention.

 

 

136

Whether the subjects in the main experiment re—
ceived their graduate training in school or clinical psy-
chology made no difference in their estimates of patho—
logical severity or the amount of special education needed
by the child they evaluated. Any real differences between
these two groups were likely attenuated by their working
together or by working outside of a clinical setting.

Surprisingly, the behaviorally oriented teacher
referral treatment condition produced higher estimates
of pathology on three of the four dependent measures than
the negatively biased teacher referral condition in the
main experiment, and nearly as high in the second experi—
ment. Though these differences were not statistically
significant, they could have been indicative of a trend
in both experiments. The behaviorally oriented referral
could have failed in its intended function as a control
situation and could have produced as much negative expec—
tancy as the negatively biased teacher referral. An
equally, and even more plausible hypothesis, is that
neither the negatively biased nor the behaviorally
oriented referral reports influenced the subjects as
much as did the positively biased referral.

School psychologists working together within a
school district showed significantly greater agreement
with each other on two dependent measures than between

the various school districts. The analyses of variance

137

for the "severity of pathology" and special education
treatment recommendation measures by school district
groups in the main experiment were significant at the
.007 and .001 levels respectively. The supplementary
study data also showed areas of homogeneity within that
group which were not seen in the combined data of the

main experiment.

Conclusions

 

In the strictest sense, these experiments were
not tests of whether expectancy effects could be demon-
strated in Michigan school psychologists; for this could
probably be done with enough experimental manipulation.
Rather, these experiments tested whether it is reasonable
to expect that Michigan school psychologists are likely
to be substantially and inappropriately influenced by
their primary referral agent, the teacher, when they
make judgments about children. The type of referral
situation of special interest was the active, somewhat
developmentally immature, early elementary boy who finds
school something he would rather avoid. The line between
problems such as this in a normal child and those in a
hyperactive child with a pathological learning problem
is, many times, difficult to see in psychological evalu-
ation. Both the symptom pattern and the diagnostic situ—
ation are extremely common in schools today. After

participating in the experiment, one of the subjects

 

 

138

was heard to say, "I know what this boy's real name is,
it's Legion." This remark is well substantiated by
several incidence studies which have found these same
symptoms in from one-third to almost one-half of normal
boys up to nine years old (Grieger & Richards, 1976;
McFarlane, Allen, & Hanzik, 1954; Werry & Quay, 1976).
The results of the experiments indicate that,
at least in situations similar to those used in the
experiment, Michigan school psychologists are unlikely
to be adversely affected by the bias of the referring
teacher. To be sure, the replication experiment demon-
strated that some effects are possible in the case of a
particular school district, but it is doubtful that
these effects would extend to the point of materially
changing the child's educational program. The support
for this conclusion comes from the fact that differences
between the experimental conditions were far from sig—
nificant on variables that would affect the child's life
such as the "Mandatory Special Education" diagnosis,
and special education treatment recommendation. Although
school psychologists in certain settings may be influenced
by the teacher's referral report when estimating pathology
in a child, their recommendations for treatment may be
influenced by other factors such as the style of remedi-
ation within a school district or the availability of

special programs. In addition, there is some evidence

 

 

139

to suggest that placement committees like those in Mich-
igan may make decisions for treatment quite independently
of the psychologist's recommendations (Morrow, Powell,

& Ely, 1976).

Why did the experiments fail to demonstrate
expectancy effects? Of the many explanations possible,
it appears that many of the subjects were either never
influenced by the teacher referral reports, or revised
their expectations when they studied the child in ques-
tion. The literature of expectancy effects is so vast
that experimental support could be enlisted for either
of these explanations. Although the experimental task
was ambiguous in these studies, the subjects were not
unfamiliar or untrained with respect to what had to be
done. As Ingraham and Harrington (1966, p. 460) have
pointed out,

. . . Ambiguity adequately accounts-for the
results obtained in typical E-bias studies.

Es are clearly placed in an ambiguous situation
for they are naive about handling Ss and the
experimental method. When they encounter $5 for
the first time they have little experience to
draw upon, so they must respond on cues available
to them, namely, what they have been told to
expect and their own varied attitudes.

Despite the realistic and commonplace nature of
the evaluative problem which confronted the subjects in
these studies, the fact remains that all subjects were

aware that they were involved in an experimental task

and not a real one. It is possible, then, that the

140

results would have been different in an actual practice
setting. In an actual practice setting the psychologist
would have actually interacted with the child which
would have introduced a host of possible influencing
variables. Although such interaction could have
increased the power of expectancy through the psycho-
logists' influence upon the child's behavior, some
literature (Brophy & Good, 1974) suggests that the
longer such interaction occurs, the less expectancy
effects tend to be shown. The author's own review of
the psychometric expectancy literature, too, indicates
that greatest expectancy effects tend to be shown in
experiments which have not provided for examiner-examinee
interaction. It is submitted, therefore, that since
this study did not result in significant expectancy
effects when the probability for such effects was maxi-
mized through noninteraction between psychologist and
client; it is unlikely that another experiment, in an
actual practice setting, would produce significant
effects.

A definite conclusion regarding school district
effects would not be proper here because the studies
were not specifically designed to test for these effects.
Hence, the statistical analyses performed were upon the
same data that had shown nonsignificance in other analyses;

also, the studies did not control for confounding

141

variables such as the effects of time and place. How-
ever, the high degree of significance reached in the
main experiment (p = .007 and .001) suggests that the
school district variable may be extremely influential,
particularly upon the treatment recommendation activities
of school psychologists. Homogeneity within a school
district was also shown in the second experiment where,
despite significant differences in estimates of patho-
logical severity, the subjects tended to recommend
approximately the same amount of special education
intervention.

Finally, though again not part of the main exper-
imental design and hypotheses, it is probably safe to
conclude that whether school psychologists experienced
their graduate training in a clinical or school psychology
university program makes no difference in the pathological
severity they ascribe to a child or the amount of special
education intervention they recommend, in so far as the
situation is like that presented in these experiments.

Three special limitations of the main study are
worth mentioning as a word of caution in generalizing
this study's results. First, although the sample for
this study was selected randomly, the population from
which it was selected included only school districts
which employed three or more school psychologists. The

results, therefore, cannot be generalized to school

142

psychologists who work alone or with a partner in inde—
pendent local school districts, or to independent psy-
chologists who contract their services to local dis-
tricts. Secondly, the three types of teacher referral
reports used in the experiment were treated as fixed
variables and were not randomly selected from all
possible or known referral reports. Thus, despite their
apparent representativeness, generalization to other
types of referral reports should only be done with care-
ful attention paid to their content. Lastly, generali—
zation should be confined to situations very similar

to the type of evaluation situation presented in these
experiments, that is, the evaluation of a possible learn-
ing disabled lower elementary boy.

Implications for Practice and
Future Research

 

 

The results of these experiments suggest that
extensive efforts on the part of school systems or
school psychological service departments to reduce
possible negative expectancy effects in their school
psychologists are probably not necessary, at least in
the context of teacher referrals. If it is felt that
precautions are necessary, however, it is doubtful that
a behaviorally oriented referral format, of the type
used in this experiment, would be useful for this pur-

pose. Although, as the Grand Rapids experiment showed,

 

 

143

it is possible for expectancy effects to occur in an
individual school district; it is doubtful that these
effects would substantially modify the child's education.
There is some evidence to suggest that school
psychologists may be influenced by the special education
philosophy within their school district or by local
deviance norms within that district. At any rate, they
appear to estimate pathological severity and to recommend
special education intervention somewhat differently among
the various school districts. Since this influence
appears to make itself felt most significantly in the
area of treatment recommendations, it can substantially
affect the daily life and education of the child. This
finding suggests that something should be done about
possible misdiagnosis and unnecessary treatment of
learning disability and emotional disturbance in certain
school districts. Perhaps State Department of Education
inservices with special education administrators and/or
school psychologists would be an appropriate means of
correcting this situation and standardizing the place—
ment of children in programs. Some attention should
also be given to the development of clearer, less
ambiguous criteria for the diagnosis of learning
disability and emotional disturbance in school chil—
dren. Recognition should be given to efforts already

expended in this development but further improvement

 

 

144

appears to be necessary. Finally, university training
for school psychologists can also be addressed to this
problem. For example, Bloom and Tesser's (1971) experi-
ment found that forewarning subjects about the possibility
of not being neutral reduced expectancy effects in their
most influenceable subjects. Perhaps forewarning school
psychologists in training about the possible influence

of their school districts would be equally as beneficial.
Universities would also do well to provide experiences
with normal children and children with only mild problems
so that trainees would be able to observe the differences
between normalcy and moderate deviance.

Practicing school psychologists should be aware
of the relationship that may exist between their theo-
retical orientation and diagnostic and recommendation
practices. The psychodynamically oriented school psy-
chologists in these experiments showed a moderate ten-
dency to overdiagnose pathology and to overrecommend
treatment. It is also possible that behaviorally
oriented school psychologists may be overly influenced
by positive teacher reports of a child. Further, more
than a few psychologists in these studies tended to rely
somewhat too heavily upon special education services for
the child, even when they did not diagnose a handicapping
condition under the "Mandatory" classification system.

It appears that many school systems should have other

 

145

types of services available to normal children with
problems, or perhaps school psychologists need more
time or inclination to help classroom teachers work
with these problems themselves.

One of the implications of these studies for
further research is that more should be discovered about
the effects of behaviorally oriented referrals and their
relationship to expectancy. A replication of the main
experiment reported here with the addition of more sub-
jects and a control group receiving no referral could
help to answer some of the questions about this type of
referral. Does this type of referral convey neutral
expectancies or do these expectancies tend to be neg—
ative? Various combinations of behavioral data together
with biased teacher reports could also help to answer
the question of the relative impact of behavioral data
versus impressionistic or inferential teacher reports.

Future studies should also provide some method
of determining whether or not the intended expectancy
was actually induced in the experiment. This type of
check could be accomplished through an additional con-
trol group of subjects who are asked to rate the posi—
tive, negative, or neutral qualities of the induction
technique. Another method of checking the success of
the induction would be to ask subjects to predict the

child's adjustment on a dependent variable prior to

 

 

146

their seeing the observational or test data. Different
post-experiment dependent variable measurements could
then be compared with these predictions to determine
whether earlier expectations were revised in the light
of later data.

Further studies of expectancy with school psy—
chologists would probably do well to avoid using exper—
ience as an independent variable. Its influence in
these studies appears to be limited and not worth
valuable Space in a multivariate research design.

The studies reported here found school psy-
chologists' theoretical orientation to be an important
variable which seems to interact with dependent var-
iables and the type of teacher referral in complex and
unpredictable ways.) There was some evidence that school
psychologists of Similar theoretical orientation may find
themselves somewhat more in agreement with each other
when making treatment recommendations than when diagnos-
ing, when the referral is behaviorally oriented. Posi-
tive referrals, on the other hand, may find them more in
agreement on diagnostic than treatment variables. The
interaction of these variables could be studied experi-
mentally in a multivariate design with enough subjects
to include theoretical orientation as an independent
variable in the analysis. A series of univariate two

factor experiments could also study this interaction.

 

147

Because of these possible interactions caution should be
exercised when making generalizations regarding the
effects of theoretical orientation upon a single type

of dependent variable.

Most importantly, the influence of the employing
school district upon judgments and recommendations of
school psychologists Should be explored further. A cor—
relational study which attempts to identify variables
within a school district which might be related to school
psychologists' judgments would be an excellent beginning.
For example, the number and type of Special education
programs prOportionate to the student population in a
school district could be considered an indexing variable
for special education philosophy or the district's
investment in special education. Various indexes of
local deviance norms such as the number of Title I stu-
dents or drop out rate could also be explored.

In the event that correlational studies would
identify promising variables, experimental studies Should
be designed to test their effects. Such studies, of
course, would need to control for the possibility of
confounding the experimental sessions with differences
in time and place. Perhaps a single large experimental
session, such as during a convention or large meeting,
would provide enough subjects within several school

districts to adequately test hypotheses concerning

 

_
_.

 

148

school district effects. The study could simply test
for school district effects or, in a two factor design,
could combine these effects with expectancy to observe

possible interaction effects.

 

APPENDICES

 

APPENDIX A

TEACHER REFERRAL REPORTS

 

 

APPENDIX A

TEACHER REFERRAL REPORTS

APPENDIX A1

NEGATIVELY BIASED TEACHER REFERRAL

Name: Scott Jones

.Age: 6 years, 8 months
Birthdate: August 8, 1968
Grade: 1

Brief Description of the Child:

Scott is doing poorly at the low average academic level. He
is a hyperactive child in the classroom as well as on the playground.
His written work is hastily done. He is aggressive and impulsive in
his relationships with the other children. Scott has a concentration
problem and usually can't stick with one thing for very long. His
meddlesomeness and overactivity make him impossible to have in class.

Reason for Referral:
Psychological testing is requested to help Scott with his

problems and to determine if he is eligible for special education
help.

149

 

 

 

150

APPENDIX A2

POSITIVELY BIASED TEACHER REFERRAL

Name: Scott Jones

.Age: 6 years, 8 months
Birthdate: August 8, 1968
Grade: 1

Brief Description of the Child:

Scott is doing acceptably well at the low average academic
level. He is a very energetic youngster in the classroom as well as
on the playground. His written work is quickly done. He is assertive
and spontaneous in his relationships with the other children. Scott
has a constant flow of ideas and usually likes a lot of variety in his
work. His curiosity and spirit make him a pleasure to have in class.

Reason for Referral:
Psychological testing is requested to see how I can help

Scott best utilize his abilities and to determine if he has any
particular needs.

 

151

APPENDIX A3

BEHAVIORALLY ORIENTED TEACHER REFERRAL

Name: Scott Jones

Age: 6 years, 8 months
Birthdate: August 8, 1968
Grade: 1

Behavioral Description of the Child (please confine statements to
objective data and/or observable behavior):

Scott is working at the 35th to 40th percentile in reading,
math and Spelling. During the typical hour class period Scott is out
of his seat ten to fifteen times. On the playground he can be observed
in at least one argument over game rules and to initiate a new game or
activity (about half the time while another activity is in progress)
at least once per recess. His seatwork is accurately done when assign—
ments are short, but his accuracy decreases to 50—60% when assignments
last for more than 20 minutes.

Reason for Referral:

Psychological testing is requested to help deteriine how best
to work with Scott and to discover if he has any particular needs.

APPENDIX B

BACKGROUND INFORMATION SHEET

 

 

APPENDIX B

BACKGROUND INFOMATION

Parent ' s statement:

Scott's mother states that he has always been a very energetic
child, constantly "getting into things" and interested in
activities of his own choosing. He prefers rougher outdoor-type
play to indoor activities and has never enjoyed things like
coloring, painting or stories very much. He has been a fairly
difficult child to discipline and is sometimes subject to temper
tantnzms. She is concerned that he is in the lower reading group
at school and may not be doing as well as he could be. She would
like as much help for him as possible.

Developnental data:

- normal pregnancy and delivery

7 lb. 10 oz. birth weight

- no serious childhood illnesses

- sat up without support at 7 mos.

- walked unaided at 13 mos.

- began using short sentences in speech at approximately 28 mos.

- toilet trained at 3 years, but still infrequently vets the bed.

- sleeps well, but often reluctant to go to bed

- one younger female sibling, showed some initial difficulty in
adjusting to her

- high fevers, 2 days duration, at 1: mos. (infection)

School history:

- enjoyed kindergarten

- below average coloring, writing and cutting
- higher than average activity level

- preferred action-type play

- fair listening skills

152

APPENDIX C

WECHSLER INTELLIGENCE SCALE FOR

CHILDREN-REVISED PROTOCOL

APPENDIX C

WECHSLER INTELLIGENCE SCALE FOR CHILDREN-REVISED PROTOCOL

WMAGE _6__ssx_£_'1_
WISC-R :22

 

 

 

 

 

 

 

 

 

 

 

 

 

PARENTS NAME
scoot cups ' /
WM! Idem Sale
MW ‘ PIACEOF‘I'ESVII‘KL TESTED IYg
mesa-v '
WISC-l PROFILE Year Month Day
Clinicians with d rofileeheuldflumlhechld'eceledmbfile all»):
w.mmm:x:u:5uemunnnuue$mmm,m?mon: whim" li_3_2é_
cennedlng the X'e.‘ Woofliflh a _Z_ L
wrest: momma TESTS Age _. (6.. L LLB
_ g “ § 3 law Scaled
3 g. I; i g ‘ j 3 : Score Score
32:; , fight—5 mm'rssrs .
E a 2 > a ‘ g 3 'an ‘ q
’.....""" [:1 I: [:1 [3 [:1 “se..“ [:1 C] D [:1 C1 Cl “3....“ Similarities ___6_ .__2__
‘9 e e e e e e ', e e e e e e ', ka Z t‘ 2
'. e e e ' '. e e e e e e ‘. vuawa" I 2 to
'7 l7 . O . . . . i: Comprehension __1L _L
z: z: , , , , , , ,5 (Dion span) (4) t__L)
“ '4 e e e e e e “ VOPbOI $60!.
13 I: ' ' ' ° . ° ° 13 PERFORMANCE TESTS
:2 I: ' ' ' ' ° ° '3 Picture CompOefion A __£_0
II II . ' . .. . ' .‘2‘ .. Picture AnangemeM_/Z. A
{:0 - '7' : .2 :3 ; -. ° .;.?'..:..'*7..'.". -j. " r... .1: . m, 0“,... z 7
z . . . onion Assembly __‘7_ _£_
7 7 7 Coding 11 __l_
6 6 6 (Mazes) L__) (__.)
5 5 5 Performance Score LL
4 4 4
° 3 3 fl“? :0
2 2 2
1 I I Verbal Score ‘18.
‘See Chemo! 4 In Doe peeve! let a dlueufion e! the significance el dMeveecee We were: on Ibo but. 'OWMOM. 5C0" 1&3—2Z
NOTES Full Scale Score .21.. 11'.

Printed in ILSA.

 

 

 

0”, . , / ; ...........................

Pie W M. M Yeti. NJ. lfllP 74-103“

153

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

154

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

'1"; gmfgmllnogwn bug". 120:; 2. PICTURE COMPLETION Discontinue aherlconsecetive failures.
Sco ‘ S: r
1. Finger / 1 or'; 1 33
1. Comb I 14. Playing Card /
2. Ears I 2, Woman /‘ II o 15. Girl Running L
3, l... / 3. Fox 7 / Io. Cour M2. (9
‘ ”I / 4. Hand I 17. Day 4L o
5. Co: I Is. Scissors 44 O
. Nickel q .——— 6. Mirror I 19. GM a“ O
. Cow I 7. Clock 0 20. 5".MEE 0
w.“ 5 Q .1 I s. Elephant / 21. Cow ’
. ' , 9. ladder / 22. Thermometer
a. March M ’ Io. Dresser / 23. House
9. Bacon 4 I I = ’ 11. Bell% W O 24. Telephone
12. Man I 25. Profile
10. Do ‘
30" d4 ’ 13. Door I 26. Umbrella
1:. Seasons 4“ ‘_ Me,
12. Amrkbd‘.‘ M " h” ‘ ’2—
13. Stomach 3. suwmmss autumn». afhrScansecafive failures. m
14.5.». 1.Whee|—ball 4M ”4‘ka .7 C)
15. leap Year _ .
2- a... ...... M at. 421 r
‘6. M“, 3' Shh-‘8‘" ‘A‘ ('00: W Cl...“ I
17. 1776 4 Pia m_. uflar
‘ 1M 4 (M O
18. Oil 5 A ppk ha no Scors_mgr_‘
e — ‘ - I
19. 30 d - ' 5%:
"' 6.M—VIMM“$ Q M ’
20. Ton 4“ c ' L ‘ pg“:
7.cd_muhz(mmim '
21. Chile .
, 4‘7 O
22. Glass _
thiamine—rad MMMA-D‘fl.“ I
23. Greece W
' 10. Pound—yard
24. Tall «- . O
11. A . I 6 I I d:
25. Barometer ”fl—ky M e 7 O
12. Scissors—copper o " “‘2‘ MM o
26. Bust w
27 LosAngelss 13. Mountain—lake ,
23. Hieroglyphics "‘ u'”"’"“""“
29. Darwin 15. First-last
30. Turpentine ‘16. Thenembers 49 and 121
“IF :7. son—m

YoOal

 

 

 

'Iltbecnndplvese Imrmonelro- M,sey."l0ewelseererbe
numbers 49 .4 I21 elite?” Total

«xi

 

 

 

 

155

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

’Give Sample he. first.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4. PICTURE ARRANGEMENT Discontinue oflerSconsecuOive failures.
Score
Arrangement Time Order (Circle the appropriate score for each hem.)
Scale (SAMPLE)
u I / 5 0‘47
1. Figh' 45 2 D ‘
L
2. Picnic 45" ; 4‘ °°‘ o ‘ 9
"Rm——
3. Fire 45" :30 it o (D #2
I3 ("I "H
' 2
4.P1anl 45" ' M (D ‘ was
31,0 M w"!
I "as 140
5. Burglar 45" Z 6 o 4
Y TM M43 n I! I so
6.Sleeper 45“ 35 MKS ® l3 4 5|
was
r "as no
7. Artist 45" 2,0 [/41 o 4 5
VA»
u-u In: mo
0. lasso 45" 30 CW RD 3 4 5
2 :95 I to
9. none 60" fl 0 2 4 5
ss-eo was ms
10. Gardener 60" q? M0 2 4 s
'0!”
3m “.33 ms
1 1. Bench 60" {/3 Ecelfli Q 2 3 4 5
”2! 22‘5”
30.40 was ms
12. Rain 60" ((0 (00+ 0 2 3 4 s
COtW CtOUD
’Masfil
total [1]
5. ARITHMETIC
Discontinue afterSconsecuOive failures. 6 BLOCK DESIGN Discontinue aher2consecu9ive failures.
Sc
Probiem Response 1:; ' .Score ‘
1 30., ( L | Design 1m Pgstail (Circle the appropnare score for each design.)
0 . M M ‘ ’1 F 2 I .
.2. 30' q I 1. 45 2 ’5 F ® 1 [’0 R
3. 30. T l ‘ u ' to p
4. 30" '? o 2' ‘5 2 o I
5. 30"
6 30" 7‘ l “ ”H" 3. 45" ' I! P
. '1 I 2 0 l
7. 30“ c I 4. ‘5" 3 3' F 9 22:5 “.5” "25 L710
8' 30 [4 J 5 75“ 21.75 16~20 11-15 L")
9.30" ’0 o - ‘40 F (D 4 5 6 7
1 . 30" u -' . i . s . o
I? 30" (at. O 6. 75 .73 F ® 21:: ms» I61 I71
‘2: 30,, o 7. 75" o 2115 16520 11515 L710
'3' 30“ or 26-75 21v2S 16-20 1.15
14. 45" 8. 75 0 4 5 6 7
‘5. 45a 9, '20” 0 .5.-“20 35555 2.8” L725
'6' 75" u "430 56-73 41.3: Ho
17. 75" '°‘ "o o 4 s a 7
13. 75" ".120“ 0 euro “5” 0.33 2.740
'ProNe 2 d J Max.=1
given 1;.”13. each"; ” Max.= 62
child meleserrerbur aor- 1b”. 7 Total

redo it within lime limit.
the“ bell-scores upward.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

156

. VOCABULARY Discontinue after 5 consecutive failures.

. Knife ‘1

un—s—d-e-e-‘d-e—s
OOVOM‘Un-‘o

parlour?»

. Umbrella i, 4441 W 142’

Clock -
"°' '41“ a
Bicycle ‘

Nail

Alphabet ,4 . .

. Donkey 4

Thief 5;

. Join I
. Brave

. Diamond I'é

.o..... a". 4......

. Nonsense m

. Prevent fl

. Contagious ’2’

. Nuisance 4‘

. Fable fi 441--

. Hazardous

. Migrate
. Stanza
. Soclude
. Mantis

. Espionage

. Belfry
. Rivalry

. Amendment

. Compol

. Affliction

. Obliterate '

. imminent

. Dilatory

2.53::
z
i
2.
2.
z
2.
i
Z
Z
O
O
Q
l
o
c:
Q
Q
o

 

157

 

8. OBJECT ASSEMBLY Give entire test to all children.

 

 

iEnter Number '
of Correctly Multiply Score
Obiect Time Joined Cuts by (Circle the appropriate score for each item.)
Apple (SAMPLE)

 

(0-6) suso use no
. , o i 2 4 5 | 6 7 I |
1.6M 120' 8'2. 1 Q rune! Assault
«use was was

[5‘ (M) 1 o I 2 (D 415 s 7 by]

 

 

 

 

 

 

 

 

 

 

 

 

 

2. Horse 150"
_ rssrscr assusu
(0.9) smso as.” as.” us
. I SO . o I Q s 4 1 5 a 7 g
3. Car iSO’ ll; . ' ma mm“
(012) wuss 51.7.1 ss-so ms
4.Foce180" I” v.0 0 O 2 3 4 5 l 6 7 8 9]
rssrscr assign
'Ieund hail-scores opt-ad. Max.=33
Total q

 

9. COMPREHENSION Discontinue after 4 consecutive failures.
1. Cut finger « - a.“ “a. X
2. Find wallet

'3. Smoke

‘4. Policemen
5. lose boil
6. Fight

'7. Build house

‘8. License plates

'9. Criminals

10. Stamps

11. inspect meat

‘12. Charity

)3. Secret ballot

‘14. Paperbacks

 

 

 

 

 

 

 

 

15. Promise
‘ 16. Cotton
'17. Senators
on the child repiies with anty one idea. ask him for e second response. Iephrese the test item ”'wmly, ”7‘09. "1." me =
enetherthingtedetreason “heir-Megs 00-...“ I //
otal
[ i0. CODING n" s...
I A (for children under a) 120" 23 ‘m’ .‘
I s (for children s s. older) :20" I i I “55’

 

 

 

 

 

 

158

 

i i . DiGlT SPAN (Optionali Discontinue after failure on both trials of any item.

Administer both trials of each item, even if child passes first trial.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DiGlTSFOliWAltD Score

Triaii Pass—Fail Trial 2 Pass-Fail 1,1.or0
I. 3-8-6 P" 6-1-2 7 2..
2. 3.4.1.7 3412. F 6-i-5-I ‘15s: P i
3. 0.4-2-3-984511 if 5444-6371.“ F o
4. 3.13-9.1-7.4 7-9-6-4-8-3
5. 54-7-4234 '94-5-24-6-3
6. 1-6-4-5-9-7-6-3 2-9-7-6-3-i-5-4
7. 5-3-8-7-i-2-4-6-9 4.2-6-94.7444

Max.=i4

Administer DiGiTS IACKWAID even if Total Forward
child rm 0 on mmswm.
DiGiTS IACKWARD Score

Triali PassnFail“ Trial2 Pass-Fail 2,1,orD
i. 2.5 pflsa P 7.
2. 5-7-4 451 F 2.5-9 {zq F O
3. 7-2-9-6 s-4-9-3
4. 4-I-3-5-7 9.74-5-2
5. i-6-5-2-9-8 3-6-7-1-9-4
6. 8-5-9-2-3-4-2 4-5-7-9-244
7. 6-9-i-6-3-2-5-8 3-1.7.9-s.4-s.2

Max.=i4
Totailockvard

 

 

 

Max.-
3 + z = 5
Forward Sachard Total

 

 

i 2. MAZES (Optional)

Discontinue after 2 consecutive failures.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Maze M2332" Errors (Circle the appropriictyzore for each maze.)
SAMPLE
1. 30" 1 0 ' 51"“ ° 5"”
2. 30" 1 0 1 51"" o 5'2""
3. 30" 1 O ' '1'" ° 5'2""
4. 30" 2 0 5 '7’” ' 5"" ° 9""
5. 45" 2 0 ’ E700 1 Inst 0 Esters
6. 60” 3 0 s Ir‘rers 2 lrzrors I s... o 5.4,...
7. 1 20" 3 0 3 Iirers 2 Irarors I Ines o :2...
8. 120" 4 o “f" "5"" 2.5.... It?! elm
9. 150" 4 O ‘ F?“ 3 ‘5'” 3 'm i I‘m 0 Emu
Max.=30

 

 

 

 

 

APPENDIX D

BENDER VISUAL MOTOR GESTALT TEST PROTOCOL

APPENDIX D

BENDER VISUAL MOTOR GESTALT TEST PROTOCOL

00090©
239900000
0000000

0
0c
9

 

9.0.01'00000 6 0 DO

159

160

 

 

 

 

 

 

 

APPENDIX E

KINETIC FAMILY DRAWING

02953 5.2: 03sz

 

\ VA... vow. Q...

VWV \umn ®

«.0

\Jsv \

IH\\ J. n

 

m XHQmemd

 

161

APPENDIX F

WIDE RANGE ACHIEVEMENT TEST PROTOCOL

APPENDIX F

WIDE RANGE ACHIEVEMENT TEST PROTOCOL

WIDE RANGE ACHIEVEMENT TEST

 

 

 

 

”M'flf‘flg’fl "' Reading. Spelling. Arithmetic from Pre-Schooi to College 7,127,333;
$322.3“... , By J. F. Junk. s. w. Bijou. s. R. Jsstsk am...- “5......
Nune$<°TT ...... .QI’K.§.§. ..................... Birthdate ? ' g' C I M. F. Chron. Ageé’?
.c . - c..a.---l.....R..dsn. 5.......Z.Scnd..-é.25nnd.sc.ZZ.7.51.3.9.
Referred by Spelling Seore.-.§..9....0rsde.m.SIand-Sc.zz..%ile....3.9
Dete3'7'G‘z7s ........ Ellflllmtmm/ai ............. Arithmetic Scannézfsndenéué..StendoSc..ZZ‘..%ile.5../..Z
rmwndmmsmmuwmmmmhmumw.
Iasell-Gpellmg—Gsodefleasae. Levellt—lplwh ”dun” '
lImeGredsl ace-Green lemeGrsds huh-sheet.“ IlomGredrIsmeGsedslsmeOseb Item‘s“: IsmeGreds WI Md"

I . N: Is [as I: I.s ss s.s ss s.r so In e la II s.s II at u u sI In an“ cw"
s N.s Is II" II Ls ss s.s ss s.s s1 I... I [s.s Is s.s es s.s ss s.s ss In T" m 7'" 5""
s PM Is In II M ss s.s s1 s.s as II: I Gnu Is s.s ss s.s ss s.s ss Is.s 9”"! “‘7'“

s as Is Kn s u st s.1 as as re it! s s.s Is s.s s4 s.s ss s.s ss ”.0 3"" ,3, :3," g
s It: Is In It s.s ss s.s se s.s ss Iss 4 u Is s.s II 1.4 as us a Is.e Inert Is Is .I
s m II [.3 rs s.s ss s.s ss m s! Isa s I: Is s.s ss s.s ss ".5 ss I“ Home Name

I n: Is an: m s.s ss u u 1.7 es Is.s s s.s IT s.s er r.s s1 us 41 I... I km! 9 i law f
s m ss s.s « s.r ss s.s ss us 1 s.s Is sI ss u m III as ”.1 3 “m” '° 1 '"i'" ’
t at m) u s.s ss s.s Is at so In s s.s Is s.s ss as re II.s so In ' he ,, 'ffll‘,‘ .
Is Kn s I; ss s.s s.s s.s ss s.s es Is: I s.s ss s.s ss s.1 as In ss Ir.s mm‘ In ... n I.
II I... II I.s ss s.s ss s.s ss s.s Is s.s II In ss sI

 

 

 

 

 

 

 

 

 

 

 

 

 

C'i—i

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-|/\OX.JV7+/\l—A:1LJ
-~I/\O}Jv1+AVq3u CH
‘
l d l6 32
\l
2 Cf I7- 33
3..- A ii! i 34 _ -
LEVEL l
91' 85V resin to’ hir wpI-lr boolr eat: way his-n .hosr ,.
' tbefl' open letter jar deep even spell awake block size ..
weather should lip finger tray felt stalls cliff lame stru‘cic ss
rge “
P... 2. Arithmetic ify n
LEVEL I. Oral Pm e e e o O o O O o e 0 e e e e .ne
1.
1r 3’ 9 5 t 3 Fingers. 8 fingers. 9 or 6? W U
3 pennies. spend l ?__25._ I 3 + 4 apples?___1__: 9 marbles. lose 3?.__.é...- 1° "t ”
Written pm. . 'IY "
l ' 3 2 -
l s
144—(g; s 5 24 Ixz-i. 23 29 15 "U
" .3m + z...— — :I + s o x 3 ...-gt s , + 3 ...--......es ...,
‘ _ i - __ ”~4— . ———.
{ fix’ Q) ..c " Z H ‘ I Q S E B O "t
452" '3 IBOSERTHPIUZQnoe
l 3 7 C + 3 II .___.
+sss

162

APPENDIX G

SEVERITY OF PATHOLOGY SCALE

APPENDIX C

SEVERITY OF PATHOLOGY SCALE

Please estimate this child's functioning on the following continua.

Social Adjustments AVERAGE RANGE

1 2 3 h l 5 6 i 7 8 9 10

anti— overly

social . ’ conforming
Inhibitory Functioning: ' I

1 2 3 h 5 6 I 7 8 ‘ 9 10

very ' very

impulsive | . inhibited
Behavioral Control:

1 2 3 a ' 5 6 ' 7 8 9 10

hyper- , ' overb

active controlled
Attentional Processes: ' ,

1 2 3 Li ' 5 6 7 8 9 10

distractible . ' highly a.

narrowly
' ' focused

PerceptualéMotor Development I

1 2 3 Li ' 5 6 | 7 8 9 10

retarded . accelerated
Cognitive Organization: ' i

1 2 3 ii 5 6 , 7 8 9 10

Disorganized ' over-

i organized,
' stereotypic

Relative Academic Functioning: '

1 2 3 1i ' 5 6 | 7 8 9 10

under— overb

achieving | ' achieving
Overall Consistency of'Functioning:

1 2 3 LI ' 5 6 ' 7 8 9 10

very _ very

SPOttY I. I even
Overall Functioning: I I

1 2 3 a I 5 6 I 7 8 9 10

mal- superior

adjusted adjustment

163

APPENDIX H

RATING SHEET:

DIAGNOSES

CONFIDENCE

SPECIAL EDUCATION TREATMENT RECOMMENDATION

APPENDIX H
RATING SHEET

Please indicate your choice of diagnosis using both classification
systems.

MANDATORY SPECIAL EDUCATION SYSTEM CLASSICAL SYSTEM
not eligible for special functioning within normal
education limits
educable mentally developmental personality
impaired deviation
learning disabled minimal brain dysfunction
emotionally impaired neurotic reaction

behavior disorder
other (specify)

 

 

Please rate the degree of confidence you feel in the diagnostic
judgments you have made.

1 2 3 h 5 6 7 8 9 10
very very
unsure confident

 

Keeping in mind the diagnostic judgments you have made, try to choose
the best of the following administrative educational programs for
this particular child.

enrollment in regular day class; 29 special education instruction,
materials, or equipment.

special education instructional materials and equipment only;
enrollment in-regular day class.

special education instructional materials and equipment plus
special education consultative services to regular teacher only;
enrollment in regular day class.

itinerant special education tutors; enrollment in regular day
class.

special education resource room and teacher; enrollment in regular
day class.

part-time special day class where enrolled; receive some academic
instruction in regular day class.

self—contained special day class where enrolled; receive no
academic instruction in a regular day class.

combination regular and special day school; receive no academic
instruction in a regular day class.

164

APPENDIX I

RATING SHEET:

REFERRAL FOR MEDICATION

THEORETICAL ORIENTATION

TRAINING

EXPERIENCE

APPENDIX I
RATING SHEET

Would you refer this child to his pediatrician for deterziration of
medication possibilities?

yes no

_m— —-—a—I—

Please indicate the degree to which you would agree or disagree with
the following statements.

"If you have eliminated the behavior, you have usually solved the
problem." (as opposed to the idea that underlying motivation will,
very likely, cause symptom substitution if only surface behavior is
treated)

1 2 3 u 5 6
strongly strongly
disagree agree

"The examination of early childhood experience is essential to the
effective treatment of behavior/emotional problems in children." (as
opposed to the idea that these problems can usually be treated effect-
ively with a knowledge of current reinforcement patterns only)

1 2 3 h 5 6
strongly strongly
disagree agree

"Projective personality instruments are very useful in the study and
treatment of childhood behavior/emotional problems." (as opposed to
the idea that these instruments provide little information of practical
use in treatment)

1 2 3 4 5 6
strongly strongly
disagree agree

"Most people need some kind of psychotherapeutic help at some time in
their life." (as opposed to the idea that only a very small percentage
of people need this kind of help)

1 2 3 4 5 6
strongly strongly
disagree agree

Please rate your theoretical preference along the following continuum.

 

1 2 3 h 5 6 7 8 9 10
strongly strongly
behavioristic psychodynamic
Please write the number of years you have been practicing in school
psychology since your graduation from training: years.
Was your graduate training primarily in school psycho105y or in

clinical psychology?
165

APPENDIX J

RAW DATA FREQUENCY DISTRIBUTIONS

 

APPENDIX J

Table J-l
Raw Data Frequency Distributions

Length of Experience

Years 0-1 2-3 4
Frequency

Negative 2

Positive 2

Behavioral 1

5

2

5 6 7 8—9 10-12 13-18

Total
Grand Rapids

hs0\ou»-Q
knChknt9"
c>uanac>+e
toknk*hoh)
AJO\&)AJA)
c>xaea£rn3

Theoretical Orientation

Score 5-10 11—14 15-18 19-22 23-26 27-30
Frequency
Negative 1
Positive 1
Behavioral 1
Total 3
Grand Rapids 0

HWKQKJ‘H
zv\0\nnona
\u\0+4ro<m
\nkn{?{?\l
HWNHO

Confidence

Rating 1-2
Frequency
Negative 0
Positive 0
Behavioral O
O
O

 

b.)
:-
U!
0\
\1

9-10

Total
Grand Rapids

WHOOH
Hit-H04“)
HwI-‘HH
HMKAON
N\OHO\(\)
NO\\O\»-P' OO
NOH-F‘Kn

Severity of Pathology

Score 4.5-7.5 8.5-11.5 12.5-15.5 16.5-19.5 20.5-24.0
Frequency

Negative 3

Positive 3

1

7

 

Behavioral
Total
Grand Rapids 1

{rtmmm
-F“Q(Dk0<h
n>cnxukoro
F‘RDAJCDCD

Special Education Treatment Recommendation

 

Score 1 2 3 4 5 6 7 8
Frequency
Negative 7 1 7 O O 1 1 0
Positive 2 3 3 3 2 1 1 O
Behavioral 5 1 2 3 5 O O 0
Total 14 5 12 6 7 2 2 0
Grand Rapids 1 O 2 1 7 1 O O

166

 

167

Table J-l (Continued)
_glassical Diagnosis

Diagnosis Normal Pathology
”Frequency
Negative 7 10
Positive 9 6
Behavioral 4 12
Total 20 28
Grand Rapids 2 10

Mandatory Diagnosis

Diagnosis "Not Eligible" Pathology
LFrequency
Negative 12 5
Positive 10 5
Behavioral 10 6
Total 32 16
Grand Rapids 4 8

Referral for Medication

 

Referral Yes No
Frequency
Negative 7 10
Positive 5 10
Behavioral 8 8
Total 20 28
Grand Rapids 6 6
Graduate Training
Program School Clinical
Frequency
Negative 15 2
Positive 8 6
Behavioral 13 3
Total 36 11
Grand Rapids 11 1

 

APPENDIX K

RANDOM GROUP TESTS OF ASSOCIATION

 

APPENDIX K

Table K-l
Random Group Chi Square Tests of Association with
all Treatment Conditions Combined (T1,T2, and T3; 11:48)

Referral for Medication

"Classical" Diagnosis yes no row total
n=6 n=14 11:20
row 76 30.0 row % 70.0 % = 41.7
normal col 76 30.0 col % 50.0
tot 76 12.5 tot % 29.2
11:11.} n=16} n=28
row 76 50.0 row % 50.0 % = 58.3
pathology col 7% 70.0 col 76 50.0
tot % 29.2 tot % 29.2
column total n=20 n=28
%=41.7 %=5-3 p=.276
Referral for Medication
"Mandatory" Diagnosis yes no row total
' n=9 n=23 n=32
row 75 28.1 row 76 71.9 % = 66.7
"not eligible" col % 45.0 001 % 82.1
tot % 18.8 tot % 47.9
n=11 n=5 n=16
row % 68.8 row % 31.3 % = 33.3
pathology col % 55.0 col %»17.9
tot % 22.9 tot % 10.4
column total n=20 n=28
% = 41.7 % = 58.3 p = .017
"Classical" Diagnosis
"Mandatory" Diagnosis normal pathology row total
n=18 n=14 n=32
row % 56.3 row % 43.8 % = 66.7
"not eligible" col % 90.0 col 9‘ 50.0
tot % 37.5 tot % 29.2
n=2 n=14 n=16
row % 12.5 row % 87.5 76 = 33.3
pathology col % 10.0 col % 50.0
tot 76 4.2 tot % 29.2
column total =20 n=28
% = 41.7 %'= 58.3 p = .010

169
Table K-2

Random Group Fisher's Enact Tests Within the
Negative Treatment Condition (n=1?)

Referral for Medication

"Classical" Diagnosis yes no row total
n=2 n=5 n=7
row 76 28.6 row % 71.4 % = 41.2
normal col 76 28.6 col % 50.0
tot % 11.8 tot % 29.4
n=5 n=5 nzlo
row % 50.0 row 76 50.0 % = 58.8
pathology col % 71.4 col % 50.0
tot % 29.4 tot % 29.4
column total n=7 n=10
7% = 41.2 75 - 58.8 p = .354
Referral for Medication
"Mandatory" Diagnosis yes no row total
n=4 n=8 n=12
row % 33.3 row 96 66.7 76 = 70.5
"not eligible" col % 57.1 col % 80.0
tot % 23.5 tot % 47.1
n=3 A n=2 n=5
row % 60.0 row % 40.0 % = 29.4
pathology col % 42.9 col % 20.0
tot % 17.6 tot 76 11.8
column total n=7 n=10
%=41.2 %=58.8 p= .314
"Classical" Diagnosis
"Mandatory" Diagnosis normal patholoar row total
n=7 n=5 n=12
row 76 58.3 row % 41.7 % = 70.6
"not eligible" col 9% 100 col 76 50.0
tot % 41.2 tot % 29.4
n=0 =5 n=5
0 row % 100 % = 29.4
pathology 0 col 96 50.0
0 tot 75 29.4
column total n=7 n=10
9% = 1.1.2 % = 58.8 p = .041

170

Table K-3

Random Group Fisher's Exact Tests Within the
Positive Treatment Condition (n=15)

Referral for MediCation

"Classical" Diagnosis yes no row total
n=2 n=7 n=9
row % 22.2 row 76 77.8 % = 60.0
normal col 76 40.0 col 7% 70.0
tot % 13.3 tot % 46.7
n=3 n=3 n=6
row % 50.0 row % 50.0 % = 40.0
pathology col 7% 60.0 col 76 30.0
tot % 20.0 tot 76 20.0
column total n=5 n=10
% = 33.3 % = 66.7 p = .287
Referral for Medication
"Mandatory" Diagnosis yes no row total
n=2 n=8 n=10
row % 20.0 row 96 80.0 % = 66.?
"not eligible" col % 40.0 col % 80.0
tot 96 13.3 tot % 53.3
n=3 n=2 n=5
row % 60.0 row 75 40.0 % = 33.3
pafliology col % 60.0 col 76 20.0
tot % 20.0 tot 76 13.3
column total n=5 n=10
% = 33.3 9% = 66.7 p = .167
"Classical" Diagnosis
"Mandatory" Diagnosis normal pathology row total
n=8 n=2 n=10
row 76 80.0 row % 20.0 % = 66.7
"not eligible" col 9% 88.9 col % 33.3
tot % 53.3 tot 76 13.3
n=1 n=4 n=5
row 0 20.0 row % 80.0 36 = 33.3
pathology col % 11.1 col % 66.7
tot 76 6.? tot 96 26.?
column total n=9 n=6
% = 60.0 %!= 40.0 p = .047

171

Table K-4

Random Group Fisher's Exact Tests Within the
Behaviorally Oriented Teacher Referral Condition (n=16)

Referral for Medication

"Classical" Diagnosis yes no row total
n=2 n=2 n=7
row % 50.0 row 75 50.0 % = 25.0
normal col % 25.0 col % 25.0
tot % 12.5 tot 76 12.5
n=6 n=6 n=12
row % 50.0 row % 50.0 % = 75.0
pathology col 76 75.0 col 75 75.0
tot % 37.5 tot % 37.5
column total n=8 n=
% = 50.0 75 = 50-0 P = '715
Referral for Medication
"Mandatory" Diagnosis yes no row total
n=3 n=7 n=10
_ row % 30.0 row % 70.0 % = 62.5
"not eligible" col % 37.5 col % 87.5
tot % 18.8 tot % 43.8
n=5 n=1 n=6
row 93 83.3 row % 16.7 % = 37.5
patholog col 7% 62.5 col % 12.5
tot 76 31.3 tot % 6.3
column total n=8 n=8
% = 50.0 % = 50.0 p = .059
"Classical" Diagnosis
"Mandatory" Diagnosis normal pathology row total
n=3 n=7 n=10'
row % 30.0 row 96 70.0 % = 62.5
"not eligible" col % 75.0 col % 58.3
tot 96 18.8 tot % 43.8
n=1 n=5 n=6
row % 16.? now % 83.3 % = 37.5
pathology col 9‘ 25.0 col % 41.7
tot % 6.3 tot % 31.3
column total n=4 n=12
% = 25.0 % = 75.0 p = .511

REFERENCES

 

REFERENCES

Ash, P. The reliability of psychiatric diagnoses.
Journal of Abnormal and Social Psychology, 1949,
53” 272-277.

 

Auffrey, J., & Robertson, M. Case history information
and examiner experience as determinants of scor—
ing variability on Wechsler intelligence tests.
Proceedings of the 80th Annual Convention of the
American Psychological Association, 1972, 553-554.

 

 

Babad, E. Y., Mann, M., & Mar—Hayim, M. Bias in scoring
the WISC sub-tests. Journal of Consulting and
Clinical Psychology, 1975, ii, 268.

 

 

Baker, J. P., & Crist, J. L. Teacher expectancies: A
review of the literature. In J. D. Elashoff &
R. E. Snow (Eds.), Pygmalion reconsidered.
Worthington, Ohio: CharIes A. Jones, 1971.

 

Barber, T. X., & Silver, M. J. Fact, fiction and the
experimenter bias effect. Psychological Bulletin,
1968, 10, 1-29.

 

Beal, A. Biased therapists: The effects of prior
exposure to case history material on the thera—
pisth attitudes and behaVior toward patients.
Unpublished“60ctoral diSsertation, Syracuse
University, 1969.

 

 

 

Beasley, D. S., & Manning, J. I. Experimenter bias and
speech pathologists' evaluation of children's
language skills. Journal of Communication Dis-
orders, 1973, 5, 93-101.

 

Beez, W. V. Influence of biased psychological reports
on teacher behavior and pupil performance. 3597
ceedings of the 76th Annual Convention of the
American Psychological Association, 1968, 3,
60§¥606.

 

 

172

173

Bender, L. Instructions for the use of visual motor
Gestalt test. New York: American Orthopsychia-
tric Association, 1946.

Bloom, R., & Tesser, A. On reducing experimenter bias:
The effects of forewarning. Canadian Journal of
Behavioral Science, 1971, 2, 198-208.

 

 

Boekel, N. The influence of teacher expectations on the
performance of the educable mentally retarded.
Focus on Exceptional Children, 1969, 1, 6-10.

Brophy, J. E., & Good, T. L. Teachers' communication of
differential expectations for children's class-
room performance: Some behavioral data. Journal
of Educational Psychology, 1970, 6;, 365—374.

Brophy, J. E., & Good, T. L. Teacher-student relation—
ships: Causes and consequences. New York:
Holt, Rinehart, & Winston, 1974.

 

Brown, W. The influence of student information on the
formuIation of teacher expectancy. Unpublished
doctoral dissertation, Indiana University, 1969.

Burns, R. C., & Kaufman, S. H. Actions, styles, and
symbols in Kinetic family drawings (EFF-D): An
integpretive manual. New York: Brunner/Mazel,
1972.

 

 

Chaikin, A., Sigler, E., & Derlega, V. Non-verbal medi-
ators of teacher expectancy effects. Unpublished
manuscript, cited in J. E. Brophy & T. L. Good,
Teacher-student relationships: Causes and con-
sequences. New York: HoIt,_Rinehart, & Winston,
1974.

 

Claiborn, W. L. Expectancy effects in the classroom: A
failure to replicate. Journal of Educational
Psychology, 1969, 99, 377-383.

 

Conn, L., Edwards, C., Rosenthal, R., & Crowne, D. Per-
ception of emotion and response to teacher's
expectancy by elementary school children.
Psychological Reports, 1968,‘22, 27—34.

 

Critchley, D. L. The relationship between the induced
set of psychiatric diagnostic labels and close
mindedness on the perception of child behavior
among baccaiaureate students in nursing. Unpub-
liShed doctoral dissertation, New York Uni-
versity, 1970.

174

Dana, J. M., & Dana, R. H. Experimenter bias or task
bias? Perceptual and Motor Skills, 1969, 29, 8.

Dangel, H. L. Biasing effect of pretest referral infor—
mation on WISC scores of mentally retarded chil-
dren. American Journal of Mental Deficiency,
1972, 11, 354-359.

Dunn, L. M. Exceptional children in the schools: Special
education in tranSition T2nd edition). New York:
Holt, Rinéhart, & Winston, 1973.

Dusek, J. B. Do teachers bias children's learning?
Review of Educational Research, 1975, 3:, 661-
681.

Education for all Handicapped Children Act. Public Law
94-142, November 29, 1975.

Egeland, B. Examiner expectancy: Effects on the scoring
of the WISC. Psychology in the Schools, 1969,
6, 313-316.

Elashoff, J. D., & Snow, R. E. (Eds.). Pygmalion recon-
sidered. Worthington, Ohio: Charles A. Jones,
1971.

 

Fielder, W., Cohen, R., & Feeney, S. An attempt to
replicate the teacher expectancy effect.
Psychological Reports, 1971, 39, 1223-1228.

Flanagan, J. C. Test of general ability: Technical

report. Chicago: Science Research Associates,
1960.

Fleming, E. S., & Anttonen, R. G. Teacher expectancy or
my fair lady. American Educational Research
Journal, 1971a, E, 241-252.

Fleming, E. S., & Anttonen, R. G. Teacher expectancy as
related to the academic and personal growth of
primary-age children. Monographs of the Society
for Research in Child Development, 1971b, 36.

 

Foster, G. G., Ysseldyke, J. E., & Reese, J. H. I
wouldn't have-seen it if I hadn't believed it.
Exceptiona1_Children, 1975,‘il, 469-473.

 

Gillingham, W. H. An investigation of examiner influence
on Wechsler-Inteiligence Scale for Children
scores. Unpublished doctoral dissertation,
Micfiigan State University, 1970.

 

175

Goldfarb, A. Reliability of diagnostic judgments by
psychologists. Journal of Clinical Psychology,
1959, 15, 392-396.

 

Grieger, R. M., & Richards, H. C. Prevalence and struc-
ture of behavior symptoms among children in
special education and regular classroom settings.
Journal of School Psychology, 1976, £3, 27-38.

 

Grosz, H. J., & Grossman, K. G. The sources of observer
variation and bias in clinical judgments: The
item of psychiatric history. Journal of Nervous
and Mental Disease, 1964, £22! 105-113.

 

Grosz, H. J., & Grossman, K. G. Clinician's response
style: A source of variation and bias in clini-
cal judgments. Journal of Abnormal Psychology,
1968, 1;, 207-214.

Guskin, S. L. Dimensions of judged similarity among
deviant types. American Journal of Mental
Deficiency, 1963a,‘§§, 2189224.

 

Guskin, S. L. Measuring the strength of the stereotype
of the mental defective. American Journal of
Mental Deficiency, 1963b, 61, 569-575.

 

Hersh, J. B. Effects of referral information on testers.
Journal of Consulting and Clinical Psychology,
1971, 21, 116-122.

Hipskind, N., & Rintelmann, W. Effects of experimenter
bias upon pure tone and speech audiometry.
Journal of Audiological Research, 1969, 2,
298-305.

 

Hobbs, N. The futures of children. San Francisco:
Jossey-Bass,’l975.

 

Hollinger, C. S., & Jones, R. L. Community attitudes
toward slow learners and mental retardates:
What's in a name? Mental Retardation, 1970,
g, 19-23.

Ingraham, L. H., & Harrington, G. M. Psychology of the
scientist: SVI. Experience of E as a variable
in reducing experimenter bias. Psychological
Reports, 1966, 19, 455-461.

 

 

I76

Jacobs, J., & DeGraaf, C. Expectancy and race: Their
influences upon the scoring of individual intel—
ligence tests. Paper presented at the Annual
Meeting of the American Educational Research
Association, 1973.

Jastak, J. F., Bijou, S. W., & Jastak, S. R. Wide range
achievement test: Reading, spelling, arithmetic
from pre—sdhool to college. Wilmington, Delaware:
Guidance Associates, 1965.

 

 

 

Jensen, A. R. How much can we boost IQ and scholastic
achievement? Harvard Educational Review, 1969,
39, 1-123.

 

Jones, R. L. Labels and stigma in special education.
Exceptional Children, 1972, 38, 553-564.

 

Jose, J., & Cody, J. J. Teacher-pupil interaction as it
relates to attempted changes in teacher expec-
tancy of academic ability and achievement.
American Educational Research Journal, 1971, g,
39-49.

 

Katz, M. M., Cole, J. 0., & Lowery, H. A. Studies of
the diagnostic process: The influence of symptom
perception, past experience and ethnic background
on diagnostic decisions. American Journal of
Psychiatry, 1969, lgéj 9374947.

 

 

Kent, R. N., O'Leary, K. D., Diament, C., & Dietz, A.
Expectation biases in observational evaluation
of therapeutic change. Journal of Consulting
and Clinical Psychology, 1974, 42, 774-780.

 

 

Kessel, N., & Shepherd, M. Neurosis in hospital and
general practice. Journal of Mental Science,
1962, 108, 159-166.

 

Kester, 8., & Letchworth, G. Communication of teacher
expectations and their effects on achievement
and attitudes of secondary school students.
The Journal of Educational Research, 1972,

66, 51-55.

 

Langer, E. J., & Abelson, R. P. A patient by any other
name . . .: Clinician group difference in
labeling bias. Journal of Consulting and
Clinical Psychology, 1974, 42, 4-9.

 

 

177

Larrabee, L., & Kleinsasser, L. The effect of experi-
menter bias on WISC performance. Unpublished
manuscript, St. Louis: Psychological Associates,
1967.

Lasky, D. I., Felice, A., Moyer, R. C., Buddington, J. F.,
& Elliot, E. S. Examiner effects with the Pea—
body Picture Vocabulary Test. Journal of Clini-
cal Psychology, 1973, 22, 456-457.

 

 

Lee, S. D., & Temerlin, M. K. Social class, diagnosis,
and prognosis for psychotherapy. Psychotherapy:
Theory, Research and Practice, 1970, 1, 181-185.

 

 

MacFarlane, J., Allen, L., & Honzik, M. A developmental
study of the behavior problems of normal children
between twenty-one months and fourteen years.
Berkeley: University of California Press, 1954.

Mandatory Special Education Act. Public Acts of Michigan,
1971, No. 198.

Marwit, S. J. An investigation of the communication of
tester bias—by means ofImOHeIing. Unpublished
doctoral diesertation, State University of New
York at Buffalo, 1968.

 

 

Marwit, S. J., & Marcia, J. E. Tester bias and response
to projective instruments. Journal of Consult-
ing Psychology, 1967, 31, 253-258.

 

 

Masling, J. M. The effects of warm and cold interaction
on the administration and scoring of an intelli-
gence test. Journal of Consulting Psychology,
1959, 33, 336-311.

 

Masling, J. M. Differential indoctrination of examiners
and Rorschach responses. Journal of Consulting
Psychology, 1965, 39, 198-201.

 

 

Medinnus, G., & Unruh, R. Teacher expectations and verbal
communication. Paper presented at the Annual
Meeting of the Western Psychological Association,
1971.

Mehlman, B. The reliability of psychiatric diagnoses.
Journal of Abnormal and Social Psychology, 1952,
$1, 557-578.

Meichenbaum, D., Bowers, K., & Ross, R. A behavioral
analysis of teacher expectancy effect. Journal
of Personality and Social Psychology, 1969, $3,
306-316}

178

Meitus, I. J., Ringel, R. L., House, A. S., & Hotchkiss,
J. C. Clinical bias in evaluating speech pro-
ficiency. British Journal of Disorders of Com-
munication, 1973, 8, 146-151.

 

 

Meyers, C. E., Sitkei, E. G., & Watts, C. A. Attitudes
toward special education and the handicapped in
two community groups. American Journal of Mental
Deficiengy, 1966, 11, 78-84.

 

 

Morrow, H. W., Powell, G. D., & Ely, D. D. Placement or
placeba: Does additional information change
special education placement decisions? Journal
of School Psychology, 1976, ii, 186-191.

 

Pflugrath, J. Examiner influence in a group testipg
situation Withyparticular reference to examiner
bias. Unpublished masterTs thesis, UniverSity
of North Dakota, 1962.

 

 

Raffetto, A. M. Experimenter effects on subjects'
reported halIucinatory_experience under visual
and auditory deprivation. UnpubliShed master‘s
thesis, San Francisco State College, 1967.

 

 

 

Rist, R. Student social class and teacher expectations:
The self-fulfilling prophesy in ghetto education.
Harvard Educational Review, 1970, fig, 411-451.

 

Rosenthal, R. Experimenter effects in behaviorgl
research. New York: Appleton-Century-Crofts,
1966.

 

 

Rosenthal, R. Expggimenter effects in behavioral
research: EnIarged edition. New York:
Halsted Press, 1976.

 

 

Rosenthal, R. Interpersonal expectations: Effects of
the experimenter's hypothesis. In R. Rosenthal
& R. L. Rosnow (Eds.), Artifact in behavioral
research. New York: Academic Press, 1969a,

181-277.

 

Rosenthal, R. Task variations in studies of experimenter
expectancy effects. Perceptual and Motor Skills,
1969b, g9, 9-10.

 

Rosenthal, R., & Fode, K. L. The problem of experimenter
outcome bias. In D. P. Ray (Ed.), Series research
in social psychology, Symposia Studies Series,

No. 8. WaSHington, D.C.: National Institute
of Social and Behavioral Science, 1961.

 

 

179

Rosenthal, R., & Fode, K. L. Psychology of the scientist:
V. Three experiments in experimenter bias.
Psychological Reports, 1963, 11, 491-511.

 

Rosenthal, R., & Jacobson, L. Teachers' expectancies:
Determinants of pupils' IQ gains. Psychological
Reports, 1966, 19, 115-118.

 

Rosenthal, R., & Jacobson, L. Pygmalion in the classroom:
Teacher expectations and_pupils' intellectual
development. New York: Holt, Rinehart, &
Winston, 1968.

 

 

Rosenthal, R., & Rosnow, R. L. (Eds.). Artifact in
behavioral research. New York: Academic Press,
1969.

 

 

Rosenthal, R., & Rubin, D. B. Pygmalion reaffirmed. In
J. D. Elashoff & R. E. Snow (Eds.), Pygmalion
reconsidered. Worthington, Ohio: Charles A.
Jones, 1971.

 

 

Rothbart, M., Dalfen, S., & Barrett, R. Effects of
teacher's expectancy on student-teacher inter-
action. Journal of Educational Psychology,
1971, 63, 49-542

Rubovits, P. C., & Maehr, M. L. Pygmalion analyzed:
Toward an explanation of the Rosenthal-Jacobson
findings. Journal of Personality and Social
Psychology, 1971, 11, 197-203.

 

Sarbin, T. R. On the futility of the proposition that
some people be labeled "mentally ill." Journal
of Consulting Psychology, 1967, 11, 447-453.

 

Sattler, J. M. Assessment of children's intelligence:
Revised reprint. Philadelphia: W. B. Saunders,
1971.

 

Sattler, J. M., Hillix, W. A., & Neher, L. A. Halo
effect in examiner scoring of intelligence test
responses. Journal of Consulting and Clinical
Psychology, 1970, 16, 172-176.

 

Sattler, J. M., & Theye, F. Procedural, situational,
and interpersonal variables in individual intel-
ligence testing. Psychological Bulletin, 1967,
66, 347-360.

 

180

Sattler, J. M., & Winget, B. M. Intelligence testing
procedures as affected by expectancy and IQ.
Journal of Clinical Psychology, 1970, 16, 446-448.

Saunders, B. T., & Vitro, F. T. Examiner expectancy and
bias as a function of the referral process in cog-
nitive assessment. Psychology in the Schools,
1971, 1, 168-171.

Schroeder, H. E., & Kleinsasser, L. D. Examiner bias:
A determinant of children's verbal behavior on
the WISC. Journal of Consulting and Clinical
Psychology, 1972, 11, 451-454.

 

Shames, M. L., & Adair, J. G. Experimenter-bias as a
function of the type and structure of the task.
Paper presented at the Meeting of the Canadian
Psychological Association, Ottawa, May, 1967.

Shotel, J. R., Iano, R. P., & McGettigan, J. F. Teacher
attitudes associated with the integration of
handicapped children. Exceptional Children,
1972, 11, 677-683.

 

Simon, W. E. Expectancy effects in the scoring of
vocabulary items: A study of scorer bias.
Journal of Educational Measurement, 1969, 6,
159-164.

Snow, R. L. Unfinished Pygmalion. Contemporary Psy-
chology, 1969, 16, 197-199.

 

Strauss, M. E. Examiner expectancy: Effects on
Rorschach experience balance. Journal of
Consulting and Clinical Psychology, 1968, 11,
125-129.

 

Stuart, R. Trick or treatment--How angjwhen psycho-
therapy fails. Urbana, I11.: Research Press,
1970.

Szasz, T. The myth of mental illness. New York:
Hoeber-Harper,gl96I.

Taylor, J. A. A personality scale of manifest anxiety.
Journal of Abnormal and Social Psychology,

Temerlin, M. K. Suggestion effects in psychiatric diagno-
sis. Journal of Nervous and Mental Disease,
1968, 147,8349-359.

181

Temerlin, M. K., & Trousdale, W. W. The social psy-
chology of clinical diagnosis. Psychotherapy:
Theory, Research and Practice, 1969, 6, 24-29.

 

 

Thorndike, R. L. Review of Robert Rosenthal and Lenore
Jacobson, Pygmalion in the classroom. American
Educational Reseagch Journal, 1968, 6, 708-711.

 

 

Tolor, A., & Brannigan, G. G. How different are students
of school psychology and clinical psychology?
Psychology in the Schools, 1976, 11, 279-283.

 

Towbin, A. When are cookbooks useful? American Psy-
chologist, 1960, 11, 119-123.

 

 

Wartenberg-Ekren, U. The effect of experimenter knowledge

 

of a subject's séholastic standing on the per-
formance of’a reasoning task. UnpubliShed
master‘s theSis, Marquette University, 1962.

 

 

Wechsler, D. Manual fggthe Wechsler Intelligence Scale
for Children--Revised. New York: Psychological
Corporation, 1974.

 

WechsleglIntelligence Scale for Children--Revised:
Record form. New York: Psychological Cor-
poration, 1974. .

 

 

Werry, J. S., & Quay, H. C. The prevalence of behavior
symptoms in younger elementary school children.
American Journal of Orthopsychiatry, 1971, 61,
136-143.

 

 

 

 

11111111111111111111111111111111111111111111111111111