University of Southern California Dissertations and Theses

A Critical Analysis And Evaluation Of Evidence Regarding The Reliability And Validity Of Four Selected Measures Of Self-Concept

(USC Thesis Other)

A Critical Analysis And Evaluation Of Evidence Regarding The Reliability And Validity Of Four Selected Measures Of Self-Concept

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content A CRITICAL ANALYSIS AND EVALUATION OF EVIDENCE
REGARDING- THE RELIABILITY AND VALIDITY OF
FOUR SELECTED MEASURES OF SELF-CONCEPT
by
Elbert Earl Shreve
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Education)
JUNE 1973
INFORMATION TO USERS
This material was produced from a microfilm copy of the original document. While
the most advanced technological means to photograph and reproduce this document
have been used, the quality is heavily dependent upon the quality of the original
submitted.
The following explanation of techniques is provided to help you understand
markings or patterns which may appear on this reproduction.
1. The sign or "target" for pages apparently lacking from the document
photographed is "Missing Page(s)". If it was possible to obtain the missing
page(s) or section, they are spliced into the film along with adjacent pages.
This may have necessitated cutting thru an image and duplicating adjacent
pages to insure you complete continuity.
2. When an image on the film is obliterated with a large round black mark, it
is an indication that the photographer suspected that the copy may have
moved during exposure and thus cause a blurred image. You will find a
good image of the page in the adjacent frame.
3. When a map, drawing or chart, etc., was part of the material being
photographed the photographer followed a definite method in
"sectioning" the material. It is customary to begin photoing at the upper
left hand corner of a large sheet and to continue photoing from left to
right in equal sections with a small overlap. If necessary, sectioning is
continued again — beginning below the first row and continuing on until
complete.
4. The majority of users indicate that the textual content is of greatest value,
however, a somewhat higher quality reproduction could be made from
"photographs" if essential to the understanding of the dissertation. Silver
prints of "photographs" may be ordered at additional charge by writing
the Order Department, giving the catalog number, title, author and
specific pages you wish reproduced.
5. PLEASE NOTE: Some pages may have indistinct print. Filmed as
received.
Xerox University Microfilms
300 North Zeeb Road
Ann Arbor, Michigan 48106
I '
I .
73-18,841
SHREVE, Elbert Earl, 1920-
A CRITICAL ANALYSIS AND EVALUATION OF EVIDENCE
REGARDING THE RELIABILITY.AND VALIDITY OF FOUR
SELECTED MEASURES OF SELF-CONCEPT.
University of Southern California, Ph.D., 1973
Education, psychology
University Microfilms, A XERO X Company , Ann Arbor, Michigan
THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED.
U N IV ER SITY O F S O U T H E R N C A L IF O R N IA
T H E G R A D U A TE S C H O O L
U N 1V ER SITY P A R K
L O S A N G E L E S , C A L IF O R N IA 9 0 0 0 7
This dissertation, written by
under the direction of h.% 3... Dissertation Com
mittee, and approved by all its members, has
been presented to and accepted by The Graduate
School, in partial fulfillment of requirements of
the degree of
ELBERT EARL SHREVE
D O C T O R O F P H I L O S O P H Y
Dean
D a t e . . . ™ * } 9 ! !
DISSERTATION COMMITTEE
Chairman
ACKNOWLEDGMENTS
The author wishes to thank the following persons
for their assistance and encouragement during this study.
My dissertation committee:
Dr. William B. Michael, Chairman
Dr. Thomas E. Lasswell
Dr. Earl V. Pullias
The faculty and staff at the University of
Southern California School of Education
My wife: Margaret Baker Shreve
My parents: Guy R. and Hazel M. Shreve
ii
CONTENTS
Page
ACKNOWLEDGMENTS................................ 11
LIST OP T A B L E S ....................................... V
Chapter
I. THE PROBLEM ..... ....................... 1
Introduction
Purpose of the Study
Importance of the Study
. Establishing the Need for the
Investigation
Methodological Approach
Research Problem
Scope and Limitations of the Study
Definitions of Terms
Organization of the Remainder
of the Study
II. REVIEW OP RELATED LITERATURE............... 19
Introduction
The History of Self and Ego Theories
The Development of Measures of
Self-Regard
Procedures for Reporting Test
Reliability and Validity
Research Studies Related to the
Reliability and Validity of
Pour Selected Measures
of Self-Regard
III. THE CONTENT OP POUR SELECTED MEASURES
OF SELF-CONCEPT............... '........... 59
Introduction
The Piers-Harris Children's Self-
Concept Scale (PHCSCS)
The Thomas Self-Concept Values Test
(TSCVT)
ill
Chapt er
Page
The Tennessee Self-Concept Scale
(TSCS)
The Self-Esteem Inventory (SEl)
IV. AN EVALUATION OP POUR SELECTED MEASURES
OP SELF-CONCEPT .........................
Introduction
Evaluation of the Pour Self-Concept
Tests with Respect to the
Standards of Validity
Evaluation of the Pour Self-Concept
Tests with Respect to Standards
for Reliability
Summary
V. SUMMARY,, CONCLUSIONS, AND RECOMMENDATIONS .
Summary
Conclusions
Recommendat ions
89
159
REFERENCES 173
LIST OP TABLES
Table Page
1. Summary of Findings Showing Judgments
Rendered Regarding the Reliability
and Validity of Pour Selected -
Measures of Self-Concept .................... 158
v
CHAPTER I
THE PROBLEM
Introduction
There are few concepts so important to a child as
hqw he feels about himself. If a pupil feels that he has a
good personality and desirable traits., his behavior and his
performance will probably reflect this positive self-image.
He is likely to be a happy child at home and a successful
child at school. Self-confidence and self-esteem are often
more important to school success than ability or IQ. School
failure, on the other hand, produces school dropouts. Dig-
gory (1 9 6 6) presented evidence that lowered self-evaluation
can become so devastating that it is a prelude to suicide.
Since high self-esteem can often produce happiness and suc
cess while it eliminates tragedy and failure, most research
designed to improve the self-esteem of school children
should have a relatively high priority.
The concept of the self and the ego has an impor
tant place in personality theory. A partial list of con
tributors to self-theory is most impressive. The list in
cludes Adler (1 9 2 7)* Allport (1955)* Angyal (1941), Bertoc-
ci (1945), Cattell (1 9 6 5), Chein (1944), Combs and Snygg
1
2
(1959)* Freud (1940), Hilgard (1949), James (1 8 9 0), Koffka
(1935), Leeky (1945), Lundholm (1940), Mead (1934), Murphy
( G r . Murphy, L. Murphy, & Newcomb, 1937), Murray (Murray,
Barrett, Langer, Morgan, Diven, Prank, Jones, Homburger,
Mekeel, White, Mackinnon, Rosenzweig, Sanford, Wheeler,
Beck, Christenson, Cobb, Inglis, Kunze, Moore, Rickers-
Ovsinkina, Peterson, Sears, Shevach, Smith, Trowbridge,
Whitman, & Wolf, 1 9 3 8), Rogers (1 9 5 1), Sarbln (1952),
Sherlf and Cantrll (1947), and Wilhelm and Jung (1931).
This long list of theorists, while incomplete, shows the
importance of the self in the history of psychology. In
Chapter II, Review of the Literature, background informa
tion on the development of self and ego theories will be
presented In enough detail to facilitate an understanding
of the theoretical background on which the research of this
study is based.
Relationship of Self-Esteem
to School Learning
Is a child's self-esteem a matter that should be of
concern to the schools? An affirmative answer would appear
to be justified, as many experimenters have found a posi
tive relationship between self-esteem and academic achieve
ment (Brookover, Thomas, & Patterson, 1964; Coopersmith,
1959; Pink, 1 9 6 2; Rosenberg, 19^5; Williams & Cole, 1 9 6 8).
In the following paragraphs several compelling reasons for
answering this question in the affirmative are set forth.
The idea that schools are responsible only for cognitive
development but not for affective development in children
seems to be an untenable position.
The effects of affective or emotional components on
learning have a sound basis in psychological research and
theory. Olds (1956) discovered pleasure centers in rats
that could be stimulated with electrodes. He demonstrated
that repetition of an act, bar pressing, was greatly in
creased when reinforcement for the behavior was connected
to the stimulation of the pleasure centers in the rats;
Wilson, Robeck, and Michael (1 9 6 9) postulated the theory
that "observation of the effects of pleasure and punishment
on children suggests they also have connections to both the
pleasure and punishment centers . . . in the brain [p. 39]."
Thus it is likely that connections to the pleasure centers
lead to learning while connections to the punishment centers
lead to avoidance of learning or activity.
Combs indicated the importance of self-concept to
school people when he summed up the discussion of self-
concept at a symposium in Los Angeles. He said, in part:
I think that Dr. Coopersmith and Dr. Aspy and
Dr. Brookover are . . . telling us that the self-
concept is tremendously important in an individual's
experience, and they are also telling us that the
self-concept is brought to class by the student and
what the teacher does affects the self-concept whether
the teacher knows it or not. The child does not park
his self-concept at the door; he brings it right on in
4
with him and he is affected by his teacher and by
other students and by his experience and his self-
concept is being affected by this whether the teacher
is aware of the fact that she is making this kind of
change or not [1 9 6 9K
Jersild also emphasized the importance of school
experiences on the developing attitudes of children. He
stressed that schools rightfully ought to be concerned
with the psychological development of children:
It might be maintained that it is dangerous for
teachers, or parents, to dabble with psychology. They
might damage a child's mind just as an amateur surgeon
might damage a child's body. This point certainly de
serves attention. But let us be realistic. Every
hour, every day, millions of parents and thousands of
teachers practice psychology and, in effect, teach
psychology in their dealings with children whether
they know it or not. They are involved in situations
in which children meet success or failure, acceptance
or rejection, and countless other circumstances in
which children are discovering themselves and develop
ing attitudes regarding themselves for better or worse
[1951, P. 126].
The schools need to pay increasing attention to the
development of self-attitudes in children. Positive self
esteem is undoubtedly an important factor in school achieve
ment. In addition, the school plays an inescapable role In
the development of a child's self-image whether or not
school people are ready to accept this fact. Without in
truding on the responsibility of parents, schools can do a
great deal to Insure the development of a well-adjusted and
self-assured child.
Purpose of the Study
The major purpose of this investigation was to pro
vide a critical analysis and evaluation of different kinds
of evidence about the reliability and validity of each of
four instruments purporting to measure a self-concept con
struct. The criteria employed in the evaluation were those
set forth in the Standards for Educational and Psychologi
cal Tests and Manuals [referred to as the Standards] pre
pared by Joint Committee of the American Psychological
Association (APA), American Educational Research Associa
tion (AERA), and the National Council on Measurement in
Education (NOME) (American Psychological Association, 1966)
Importance of the Study
Evaluations of psychological instruments, theories
and practices contribute to the advancement of science.
Perdew (1950) stated that research has the purpose of ex
tending, correcting, or verifying knowledge. It Is hoped
that this study not only can examine what exists against
meaningful standards, but also can serve as a guide to
those embarking on any type of research and developmental
efforts that involve, respectively, the improved under
standing of the construct of self-concept and the construc
tion of instruments to measure self-concept that are theo
retically and statistically valid.
6
Instruments of this type may then gain that general
level of acceptance by professional personnel in public
schools that is now accorded to measures of academic
achievement and scholastic aptitude. Widespread use of
reliable and valid measures of self-concept in the public
schools could provide teachers with important information
regarding entering behaviors of students. Such informa
tion could be used in planning instructional strategies
and experiences that would allow students to achieve suc
cess and thus to receive the necessary reinforcement for
enhancing their feelings of self-esteem and self-confidence.
Establishing the Meed for the Investigation
The need for an analysis and evaluation of the re-
i
liability and validity of selected measures of self-concept
may be established upon several grounds. First,, despite
widespread interest in the construct of self-concept no
instrument has yet gained enough acceptance to be in gen
eral use in the public schools. No instruments have the
same popularity as similar tests of intelligence and
achievement. This situation indicates that the available
measures of self-concept have not demonstrated the adequacy
of their reliability and validity to public school people.
Second, in the seventh edition of The Mental Meas
urements Yearbook by Buros (1 9 7 2), the reviews of currently
available, commercial measures of self-concept have tended
to be somewhat critical. The evaluations of this investi
gation are expected to provide prospective users of self-
concept measures with additional reliable and well docu
mented information that may help to clarify the advantages
and disadvantages of available tests. Such information
can assist those who select self-concept measures for use
in public schools, research, or clinics.
Finally, unbiased and critical evaluations of
existing measures of self-concept may serve to encourage
test authors and test publishers to strive for even higher
standards.
Methodological Approach
Prior to the explanation of the research problem
in terms of eight questions to be answered, it is both
helpful and probably essential to describe the methodologi
cal approach employed, as the understanding of the ques
tions depends upon knowledge of (a) the procedures employed
in the selection of four instruments or measures of self-
concept studied, (b) the sources of information and evi
dence regarding both the reliability and validity of the
four tests chosen, (c) a description of the criteria and
of a system for their symbolization as set forth in the
Standards, and (d) a brief consideration of the model of
congruence being employed in the investigation.
8
Procedures Employed in Selection of
the Pour Measures of Self-Concept
Instruments available for measuring self-image
include commercially published tests and noncommercial
tests. The writer pursued three courses to select the best
possible measures of self-perception for critical evalua
tion. First., through DATRIX (Direct Access to Reference
Information: A Xerox Service) a search was made of all
prior dissertations on file with University Microfilms at
Ann Arbor, Michigan. The search (1 9 6 9) involved all dis
sertations dealing with measurement of self-concept, self
esteem, self-image, self-perception, and self-regard.
Second, a personal library search for measures of self-
image was conducted by the writer. This effort included a
careful review of such publications as the American Educa
tional Research Journal, the Journal of Educational Meas
urement , and the Educational and Psychological Measurement
journal. Third, the writer reviewed a chapter on self-
concept In the publication, Tests and Measurements in Child
Development by Johnson and Bommarlto (1971). In preparing
this book, the authors stated that they searched every
English-language publication, world-wide— every journal,
foreign and domestic, plus books, research reports, and
other publications over a 10-year period. These authors
selected 14 instruments as the best of the available un
published, noncommercial measures of self-concept.
Pour self-concept and self-esteem measures were
selected for research and for critical analysis. Three of
the tests were chosen because they were the only three com
mercially published measures of self-concept listed in The
Seventh Mental Measurements Yearbook edited by Buros (1 9 7 2).
The three tests found in this comprehensive two volume edi
tion are (a) the Piers-Harris Children's Self-Concept
Scale (The Way I Peel About Myself) (Piers & Harris, 1 9 6 9);
(b) the Tennessee Self-Concept Scale (Pitts, 1 9 6 5); and
(c) the Thomas Self-Concept Values Test (Thomas, 1 9 6 9).
The fourth test selected for evaluation is not published
commercially. It was chosen after considerable research
because it has been widely used in scientific studies and
because it is based upon experimental evidence and theoret
ical underpinnings which have been published in a book,
The Antecedents of Self-Esteem by Coopersmith (1 9 6 7). This
measure is called the Self-Esteem Inventory,
Three of the tests to be evaluated have test man
uals; the fourth test has a book published that explains
the instrument. The three test manuals and the book pro
vide much of the significant material for the research of
this study. A second source of information on the adequacy
of the reliability and validity of these four tests is a
review of the literature and the research articles that
involve the tests.
Sources of Information and Evidence
10
Evidence regarding the reliability and validity of
the four selected measures of the self-concept was obtained
from several sources including test manuals, books, and
monographs dealing primarily with the tests selected for
analysis. A review of related literature provided addi
tional data for making judgments regarding the reliability
and validity of these four tests. Additional consideration
of sources of data is to be made at other appropriate points
within this dissertation.
The Standards as the Basis for Establishing
Criteria for Evaluation of Reliability and
Validity of the Four Self-Concept Measures
History. The Standards has been approved by the
governing bodies of the American Psychological Association
(APA), the American Educational Research Association (AERA),
and the National Council on Measurement in Education (NCME).
The Standards is a revision of two former documents pre
pared by these organizations. A joint committee of the
three organizations worked for over five years to produce
the current Standards. During this time questionnaires and
letters of inquiry were sent to professional and research
workers in the field of educational and psychological meas
urement. Several open meetings were held at the annual
conventions of the three cooperating professional organiza
tions. The Standards was published in 1966 by the Ameri
can Psychological Association.
The Standards as a basis for generation of criteria.
The booklet referred to as the Standards provides a scien
tific, objective, and relatively unbiased set of criteria
for the analysis of the four measures of self-image. The
appropriateness of the Standards for measures of self-
concept is clarified by the following two quotations from
the Introduction:
These recommended standards cover not only tests
as narrowly defined, but also most published devices
for diagnosis, prognosis, and evaluation. The stand
ards apply to interest inventories, personality in
ventories, . . . [p. 3].
The present standards apply to devices which are
distributed for use as a basis for practical judgments
rather than solely for research. Most tests which are
made available for use in schools, clinics, and indus
try are of this practical nature. . . . [p. 3].
The l6l separate recommendations found in the body
of the Standards are grouped in three levels: ESSENTIAL,
VERY DESIRABLE, AND DESIRABLE. ESSENTIAL standards are
those which
are intended to represent that concensus of present-
day thinking concerning what is normally required for
operational use of a test. . . . The ESSENTIAL stand
ards indicate what information will be needed for most
tests in their usual applications. When a test pro
ducer falls to satisfy this need, he should do so only
as a considered judgment [p. 5].
VERY DESIRABLE Is used for standards that "draw attention
to types of information which contribute greatly to the
user's understanding of the test [p. 5]." This level of
desirability might include such information as that gen
erated from long-term follow-up studies. DESIRABLE "in-
_cludes Information which would be helpful, but less so
than the ESSENTIAL and VERY DESIRABLE information [p. 5I."
Organization of the Standards. The l6l separate
recommendations are divided in the Standards into six major
divisions: (l) A.— Dissemination of Information; (2) B.—
Interpretation; (3) C.— Validity; (4) D.— Reliability;
(5) E.— Administration and Scoring; and (6) F.— Scales and
Norms. Notation for individual standards is as follows:
Twenty-seven standards having a more general application
to most psychological tests are printed in bold face let
ters. These standards are designated by the capital let
ters referring to their major division. For example, all
standards for "validity" are designated with the letter
7-C." It should be noted that the letter for each division
is indicated above. For each of the 27 major standards the
divisional letter is followed by a number. That is, the
major standards for "Validity" are Cl, C2, C3* C4, C5> C6,
and C7. Corollary or more detailed standards are indicated
by a decimal system of notation. For example, under the
7th major standard for "Validity" which is C7, the corol
lary standards are indicated as follows: C7.1.> C7.H.»
C7.12, C7.13> C7.2, etc. This same notation has been em
13
ployed throughout the present study wherever references
are made to specific standards reported by the Committee on
Test Standards.
Criteria pertaining to validity and reliability.
In the publication, Standards,seven major standards
(criteria) and 54 corollary standards (criteria) in "Divi
sion C.— Validity" are present. For "Division C.— Relia
bility" six major standards (criteria) and 28 corollary
standards (criteria) exist. In the present study the
corollary standards were used as a guide for selecting the
kind of data that would support an evaluative Judgment re
garding each major standard recommended by the Committee.
In this study all the major standards for "Reliability”
and "Validity" have been applied as criteria to each of the
four selected measures of self-concept.
After examining all the evidence for a particular
standard and for a particular test, the writer attempted to
make an unbiased Judgment to indicate how satisfactorily
the test had met that standard. The evaluations were
classified in four categories: "satisfactory," "unsatisfac
tory," and "questionable." Major standards that did not
apply to a particular test were rated as "not applicable."
The Congruence Model
As the previous paragraph indicated, an effort was
made to determine how closely available evidence regarding
certain reliability or validity characteristics of a test
or measure of the self-concept corresponded to or was con
gruent with a given standard or criterion set forth in the
Standards. Thus a model of congruence evaluation consti
tuted the research methodology of this investigation.
Congruence models essentially demonstrate discrepancies
between what has been obtained or realized and what has
been judged to be adequate or appropriate. The smaller
the discrepancies the more satisfactory or effective is the
instrument; the greater the discrepancies the less satis
factory or effective is the instrument. As revealed in the
previous paragraph., discrepancy judgments on the part of
the writer were classified in four categories.
Research Problem
The design of this research did not lend itself
to the testing of hypotheses through the use of signifi
cance tests. Therefore, in the explication of the research
problem, the writer found it necessary to raise questions
to be answered rather than to test the tenability of hy
potheses. Thus, in light of available evidence found in
the professional literature that could be related to the
criteria of test evaluation set forth in the Standards,
15
the following questions were posed:
1. To what extent does the Piers-Harris Children’s
Self-Concept Scale meet the major standards for
validity (CI-C7)?
2. To what extent does the Piers-Harris Children’s
Self-Concept Scale meet the major standards for
reliability (D1-D6)?
3. To what extent does the Tennessee Self-Concept
Scale meet the major standards for validity
(C1-C7)?
4. To what extent does the Tennessee Self-Concept
Scale meet the major standards for reliability
(D1-D6)?
5. To what extent does the Thomas Self-Concept
Values Test meet the major standards for
validity (C1-C7)?
6 . To what extent does the Thomas Self-Concept
Values Test meet the major standards for reli
ability (D1-D6)?
7. To what extent does the Self-Esteem Inventory
meet the major standards for validity (C1-C7)?
8. To what extent does the Self-Esteem Inventory
meet the major standards for reliability (DI
PS)?
16
Scope and Limitations of the Study
The present study has been limited to a critical
review of four selected measures of self-regard: the Piers-
Harris Children's Self-Concept Scale (PHCSCS), the Tennes
see Self-Concept Scale (TSCS), the Thomas Self-Concept
Values Test (TSCVT), and the Self-Esteem Inventory (SEl).
A specific judgment on the adequacy of the reliability and
validity of these four measures has been rendered only on
the major standards found in Division C.— "Validity1 1 and in
Division D.— "Reliability" of the Standards for Educational
and Psychological Tests and Manuals. Corollary standards
in the same divisions of Standards have been investigated
to support the judgments reported for the major standards.
The findings of this study were based upon the ap
plication of sound criteria to information documented by
research. The writer attempted to make objective evalua
tions. Since the evaluations rendered do, however, involve
some subjectivity, they must be interpreted in this light.
Definitions of Terms
Standards. The word Standards refers to the publi
cation, Standards for Educational and Psychological Tests
and Manuals (American Psychological Association, 1 9 6 6).
The words "standard" or "standards" without a capital let
ter refer to one or more of the l6l separate recommenda
tions made by the Committee on Test Standards.
17
Scale. For purposes of literary style and variety,
the terms scale, test, measuring Instrument, evaluation In
strument, and Inventory are used Interchangeably In this
study.
Self-coneept. The terms self-concept, self
esteem, self-image, self-regard, and self-perception have
been used Interchangeably in many Instances. However, if
a question of exact meaning was involved, or if a particu
lar term was being evaluated, the writer attempted care
fully to choose the appropriate words.
Organization of the Remainder
of the Study
In Chapter II of this Investigation is a review of
literature pertaining to (a) the history of self-theories,
(b) the development of measures for self-regard, (c) the
procedures for reporting evidence about test reliability
and validity, and (d) studies relating to the reliability
and the validity of the four selected measures of self-
regard.
In Chapter III the content of four selected meas
ures of self-concept is described: the Piers-Harris Chil
dren's Self-Concept Scale (PHCSCS), the Tennessee Self-
Concept Scale (TSCS), the Thomas Self-Concept Values Test
(TSCVT), and the Self-Esteem Inventory (SEl). The descrip
tion of the tests includes information about the test items
18 ,
and the test scales, since these factors must he considered ;
in an evaluation of test reliability and validity.
Chapter IV reports the evaluation of the four se
lected measures of self-regard. The evaluation is based on
application of the criteria abstracted from the Standards
to the findings based upon the research reported in Chapter
II.
In Chapter V a summary of the study and conclusions
are offered based upon the evaluative information reported
in Chapter IV. Recommendations are made for further re
search to improve the reliability and the validity of self-
concept measures.
CHAPTER II
REVIEW OP RELATED LITERATURE
Introduction
It Is the purpose of this chapter to review the
literature that relates to the reliability and validity of
self-concept measures. Particular attention is given to
the research literature that bears directly upon the reli
ability and validity of the four selected measures of self-
image: (l) the Piers-Harris Children's Self-Concept Scale
(PHCSCS), (2) the Tennessee Self-Concept Scale (TSCS), (3)
the Thomas Self-Concept Values Test (TSCVT), and (4) the
Self-Esteem Inventory (SEI). This chapter which reviews
the literature is divided into four major sections. The
following paragraphs indicate how the research reviewed in
each section is related to the purposes of this study.
The first section of this chapter, The History of
Self and Ego Theories, is a review of the theoretical under
pinnings that are essential for the development of self-
concept tests. Particular emphasis is given to theories of
the self that appear to be a fundamental part of the con
struct validity of the four measures of self-image selected
for critical analysis in this research.
19
20
The second section of this chapter., The Develop
ment of Measures of Self-Regard, reviews difficulties and
sources of invalidity in the measurement of the self-
concept and outlines alternative procedures for the de
velopment of a test to measure a construct such as self-
regard.
The third section of this chapter is entitled Pro
cedures for Reporting Test Reliability and Test Validity.
Although literature about reliability and validity theory
is not reviewed, procedures for reporting evidence about
test reliability and test validity as it relates to an
analysis and evaluation of the four self-concept scales
within the framework of the Standards are discussed.
The fourth section of this chapter is entitled
Research Studies Related to the Reliability and Validity
of Four Selected Measures of Self-Regard. In this section
an attempt is made to summarize the findings of those re
search studies that throw light on the reliability or
validity of any one of the instruments subjected to crit
ical analysis in this dissertation.
An evaluation of four measures of self-concept is
presented in Chapter IV. This evaluation is supported in
part by the review of the literature discussed in each of
the four sections of this chapter.
The History of Self and Ego Theories
Every measure of self-regard is based upon a
theoretical definition of the self that incorporates be
liefs concerning how self-regard develops and how it is
manifested. In the paragraphs that follow* selected theo
retical views concerning self-concept are discussed. The
purpose of the discussion is to establish a number of
reference points against which the adequacy of the theo
retical foundations of self-regard measures may be judged.
Particular attention is given to those theories of the
self that appear to be related to the four measures of
self-concept that have been selected for analysis and
evaluation in this research.
The Usefulness of the Term
"Self" as a ConstruclT
No attempt to present a comprehensive history of
the development of self-concept theories is made in this
section. Excellent summaries of the history and develop
ment of self-theory are already available. Hall and Lind-
zey (1957) and Lindzey and Hall (1 9 6 5) reviewed theories
of the self and the ego in their books* Theories of Per
sonality and Theories of Personality: Primary Sources and
Research. Wylie (1 9 6 1) wrote a comprehensive work en
titled The Self-Concept. She discussed both theories of
self-concept and the methodologies used in self-concept
research. Diggory (1 9 6 6) reviewed the history of self and
22
ego theories. Wiggins, Renner, Clore, and Rose (1971)
summarized the current status of self-theory in their hook,
The Psychology of Personality.
Even though the behavioral psychologists in general
do not consider the term "self" to be a helpful construct,
there are, nevertheless, many indications that the self
plays an important part in personality development. Some
of the evidences for the existence of the self and for the
usefulness of such constructs as self-concept and self
esteem follow; f '
In his Principles of Philosophy, Rene Descartes set
forth his views. He was willing to doubt the existence of
God, the heavens, and his own body. He was willing to
doubt his own existence, but in order to doubt, one must
think. In order to think, he reasoned, one must exist.
"I think, therefore I am," was the most certain conclusion
of a careful philosopher according to Descartes.
William James (1 8 9 0) devoted an entire chapter to
the self in his classic work, the Principles of Psychology.
James divided the self, for purposes of exposition, into
two parts. The self as known he called the empirical self
or me. The self as knower, James felt was not necessarily
an entity, but might be simply a series of separate passing
thoughts with each successive thought incorporating all the
appropriate characteristics that belonged to the thought
that preceded it. For the purposes of psychology, James
23
postulated a self as knower that changed slightly with
every passing thought.
The self as known or the me, according to James,
was closely and intimately related to everything a man can
call his. This self as known included:
Not only his body and psychic powers, but his clothes
and his house, his wife and children, his ancestors
and friends, his reputation and works, his lands and
horses, and yacht and bank-account [p. 2 9 1].
James paid particular attention to self-esteem.
He believed aspirations to be a fundamental part of self
esteem. He said:
Our self-feeling in this world depends entirely on
what we back ourselves to be and do. It is deter
mined -by the ratio of our actualities to our supposed
potentialities; a fraction of which our pretensions
are the denominator, and the numerator our success;
thus self-esteem = ^tensions [p- 310]‘
James had previously pointed out that he felt no mortifica
tion at being ignorant of Greek, but that he would feel
humiliated if found ignorant in the field of psychology.
James also proposed a social self which was the
recognition one receives from his peers. He wrote:
Properly speaking, a man has as many social selves
as there are individuals who recognize him and carry
an image of him in their minds. To wound any one of
these images is to wound him [p. 29^].
The theory of the self set forth by James would
indicate that both one’s level of aspiration and one's
belief regarding what others think of one will affect one's
self-esteem. In turn, a person expects what others think
24
of him to be affected by all that he calls his own. His
body* his achievements, his social successes, his friends,
his family, and all other possessions.
Sanford (1 9 7 0) discussed the issues in personality
theory. He stated:
Psychologists who accent higher mental processes give
a large place to the concept of self. In the recent
literature of research and theory, the concept has
attained very considerable prominence (Allport, 1943;
Bertocci, 19^-5; Chein, 1944; Hilgard, 1949; Rogers,
1951, 1959; Sarbin, 1952; Sherif and Cantril, 1947;
Smith, 1950; Combs and Snygg, 1959; Stephenson, 1953;
Symonds, 1951; . . •) [p. 2 7].
One of the current Issues in personality theory according
to Sanford (1970) is the clash between behaviorists and
self-theorists. The behaviorists believe that the study
of simple processes and observable events will develop em
pirical laws that permit an understanding of the more
complex processes. The self-theorists assume that simple
processes are always in part determined by the larger pat
terns and purposes within which they have a place. Sanford
described these situations as follows:
For a long time psychologists have doubted their abil
ity to analyze anything so global as the whole person
ality. . . . To judge by recent writings, however,
there is now less objection to conceiving of the whole
self or the whole ego or to studying them by empirical
means [p. 3l].
He then went on to say:
Whereas psychologists differ with respect to their
views of the role and function of the phenomenal self,
not many deny that such a self exists or that studies
of its contents and vicissitudes are useful [p. 3 2].
25
Wiggins, Renner, Clore, and Rose (1971) also re
ferred to the philosophical differences between behavior
ists and self-theorists. They stated: "Self-theory is
thus a reaction against the past 70 years of behaviorism
[p. 5^3]." Self-theory, they emphasized, accepts a philo
sophical position which requires an internal, rather than
an external, view of man. These authors elaborated fur
ther by saying, "Self-theory is the current label given to
a philosophical view of man which emphasizes his capacity
for self-direction and freedom of choice [p. 5 6 0]."
Citing Self-Theories Specific to
the Pour Measures under Analysis
Gordon and Gergen (1 9 6 8) noted that psychology and
sociology had accounted for over 2 ,0 0 0 publications con
cerning the self. For the.purposes of this research it
seemed best to identify and then to discuss only those the
ories of the self that are directly related to the four
measures of self-concept that have been selected for analy
sis. In a chapter entitled "Theoretical Formulations,"
Coopersmith (1 9 6 7) discussed the works of several theorists.
These included James (1 8 9 0); Mead (193-^); Cooley (1 9 2 2);
and Rogers (1951* 196l); three neo-Freudians— Sullivan
(19^7)* Horney (Lindzey & Hall, 1 9 6 5)* and Adler (1 9 2 7)*
and Rosenberg (1 9 6 5). After discussing the views and
theories of these investigators, Coopersmith concluded
that four major factors contribute to the development of
26
self-esteem. These factors may be paraphrased and summar
ized as follows:
1. The amount of respectful, accepting, and con
cerned treatment that an Individual receives
from significant others in his life;
2. The history of an individuals successes and
the status and position that the person holds
in the world;
3. The way experiences are interpreted and modi
fied in accord with the individual's values
and aspirations; and
4. The manner in which an Individual responds
to devaluation.
The factors just listed would seem to form the theoretical
base for the questions included in Coopersmith's Self-
Esteem Inventory (SEl).
In the manual for the PHCSCS, Piers (19&9) indi
cated that the pool of items for the scale was developed
from Jersild's (1952) collection of children's statements
telling what they liked and disliked about themselves. The
manual presents no further discussion of the theory of self
that is the basis for the test.
In the revised manual for the TSCVT, Thomas (1 9 6 9)
presented an extensive theoretical justification of the
content of this test. In a section entitled "A Theory of
the Self" the author discussed the formulations by Perkins
27
(1958)* Jersild (1 9 5 2), Rogers (1951), and English and
English (1 9 5 8) as well as the special attention given to
the theory of self-concept by Freud, Adler, Sullivan, and
the phenomenologists. Thomas concluded the section of the
manual on self-theory with a rather detailed discussion of
the work of Mead. He stated:
The present instrument is intended to combine the
theoretical basis of the self provided by Mead and
Sullivan with the theoretical basis for the value
concept offered by this writer [p. 1 3].
In the manual for the Tennessee Self-Concept Scale
(TSCS), Fitts (1 9 6 5) indicated that the original pool of
items for the scale was derived from other self-concept
measures including those developed by Balestar (1956),
Engel (1 9 5 6), and Taylor (1953). No further discussion of
self-theory is presented in the manual. Fitts and his as
sociates (Fitts, Adams, Radford, Richard, B. Thomas, M.
Thomas, & Thompson, 1971) identified themselves with self-
theorists like Maslow (195^), Rogers (1 9 6 1), and Seeman
(1959).
Exposition of Self-Theories
Basic to the Tests Selected
for Critical Analysis
The self-theory of James (1 8 9 0) has already been
discussed. In order to provide a general background for
analyzing the theories fundamental to the four measures of
self-regard that have been selected for critical review,
28
the views of the following additional theorists will be
presented: Mead., Cooley, Sullivan, Jersild, Snygg and Combs,
Rogers, and Maslow.
Mead (1934) believed that the self was not present
at birth. It develops, he felt, out of social interaction.
In time the individual is able to assume the role of an
other person and to look at himself as an object from this
vantage point. The self-concept develops as the individual
begins to form his own perception of the way he believes
others perceive him. The self arises, then, out of inter
action with the society and with the members of the society
in which a child grows and develops.
Cooley (1922) described the social self as a re
flected or looking-glass self:
Each to each a looking-glass
Reflects the other that doth pass [p. 184].
The individual, Cooley felt, has some conception of his own
appearance. This is followed by some imagination of how
others judge his appearance. Finally, some sort of feeling
such as pride or mortification follows the individual's
conception of how he believes others see and judge him.
The looking-glass reflection is not simply a mechanical
transaction. It is followed by an imagined judgment. This
perceived judgment of another person is the result of the
reflection of a self upon the mind of another. The in
dividual is pleased or displeased, happy or sad, proud or
29
ashamed as he imagines what is going on in the mind of an
other person when that person perceives him.
Sullivan (1947) developed an interpersonal theory
of psychiatry. He believed that personality could not he
isolated from interpersonal situations. For Sullivan, psy
chiatry and social psychology were closely allied. Like
Mead, he accepted the social origins of personality. Ac
cording to Sullivan, anxiety is a product of interpersonal
relations. Parents and siblings are important as signifi
cant others who are particularly influential in the develop
ment of the self-system. The self-system is the guardian
of one's security. Although the function of the self-
system is essential in helping the individual to cope with
anxiety, it is often distorted because it develops in a
society that has many irrational aspects. Sullivan's theory
is notable for its exclusive emphasis on the role of inter
personal relationships in the development of the self.
Jersild presented his own definition of the self
as follows:
When we speak of the self, we mean, among other things,
a system of ideas, attitudes, appraisals, and commit
ments to one's own person. The person experiences
these as distinctly belonging to him, and all of them
together constitute the person's awareness of his in
dividual existence and his conception of who and what
he is [1951. P- 1241.
For Jersild, then the self is a composite, a collection of
many features. He and his colleagues gathered compositions
from several hundred children from the fourth grade through
high school on "What I like about myself," and "What I
don't like about myself." These were grouped into the
following categories: (a) Physical Characteristics and
Appearance; (b) Clothing and Grooming; (c) Health and
Physical Soundness; (d) Home and Family; (e) Enjoyment of
Recreation; (f) Ability in Sports, Play; (g) Ability in
School, Attitudes Toward School; (h) Intellectual Abilities
(i) Special Talents (music, arts); (j) Just me, Myself; and
(k) Personality, Character, Inner Resources, Emotional
Tendencies. Although Jersild (1952) accepted the influence
of others on the self as being extremely important, he
stressed also the Importance of the child's own qualities
as another salient factor in the development of the self.
By the child's own qualities Jersild meant such things as
temperament and disposition. He saw these characteristics
as different for each individual but nevertheless as in
fluential in their effect on interpersonal relationships.
Combs and Snygg (1959) stressed the need for a
phenomenological system of psychology. They pointed out
that behavior may be viewed from two distinct points of
orientation. It may be studied objectively by an outside
observer or It may be studied phenomenologically. That is,
It may be studied from the point of view of the behaving
organism. The perceptions of the behaving individual play
31
an integral part in the theory of Combs and Snygg. They
say, "All behavior, without exception, is completely deter
mined by, and pertinent to, the perceptual field of the
behaving organism [p. 20]." They also explained what they
meant by the perceptual field. It is, they said, "the
entire universe, including himself, as it is experienced by
the individual at the instant of action [p. 20]." The
perceptual field is also called the phenomenal field.
Turning to the subject of self, Combs .and Snygg explained
their position as follows:
All perceptions of the self a person has at a particu
lar instant we call the phenomenal self or the per
ceived self. It is the phenomenal self which each
human being is forever seeking to maintain [p. 44],
At the center of the phenomenal self, these theo
rists placed the self-concept which in their theory they
saw as even more stable than the phenomenal self. The
self-concept, which is extracted from the phenomenal
field, includes those concepts of the self so fundamental
to the individual that they seem to be "he in all times and
at all places [p. 127]." The self-concept of Combs and
Snygg is developed out of the perceptions of the individ
ual, and these perceptions are only understood accurately
from the internal frame of reference of the individual
doing the perceiving.
Rogers (1951) emphasized the significance of the
self-concept in human behavior. His theory of personality
32
is basically phenomenological in character and relies
"heavily upon the concept of the self as an explanatory
construct [p. 532]." Rogers declared that there is no need
for hidden or unconscious motivation in explaining behavior.
Rogers took the position that self-reports are useful for
obtaining information about the person. In Rogers' think
ing, the individual can be understood by listening to what
he says about himself.
Rogers and his colleagues, Butler and Halgh (195^)j
hypothesized that compared with other forms of counseling,
client-centered counseling in which self-Judgments were
minimized would result in a closer congruence between a
client's [real] self-image and the client's ideal self-
image. Using Q methodology of Stephenson (1953).> they de
veloped Q,-sort items for a study. The items included
statements such as "I am an impulsive person," "I am a
hard worker," and "I am a submissive person." They gave a
self-sort and an ideal-sort to three groups of people. One
group that asked for therapy were tested again after ther
apy. The correlation between self-concept and ideal self
increased significantly for this group. Another group
asked for therapy but the beginning of therapy was post
poned. This group did not show an increased correlation
between real-self and ideal-self during the waiting period.
A third group consisted of persons not seeking counseling.
33
They made no changes in pre- and posttest correlations
between real-self and ideal-self measures. This group,
however, had a continuing positive correlation between
current self-concept and ideal self-concept. This, the
investigators felt, proved that those not seeking therapy
were much better satisfied with themselves than the group
that requested counseling assistance.
The type of study just described demonstrates the
importance of the self-report in Rogers' system of person
ality. In Rogers' theory the self, which grows out of the
phenomenal field of the Individual, is made up of the con
scious perceptions of what each person believes himself to
be.
Maslow (195^j 1 9 6 2) has been classed as a phenome-
nologist. He believed that all human perception is a prod
uct of the human being. He stressed that the individual is
Innately good and that this capacity for positive potential
is reached in a self-actualized person. The self
actualized person has peak-experiences. These are char
acterized by such qualities as wholeness, perfection,
aliveness, uniqueness, effortlessness, self-sufficiency,
beauty, goodness, and truth. Maslow believed nearly
everyone has some peak-experiences, but that self-
actualized people as compared with those not self
actualized have these experiences more consistently. At
34
the time of a peak: experience, according to Maslow, per
ception is very strongly idiographic. Whether or not this
perception has external validity is not important. It is
real for the individual and it may be a more nearly true
picture of reality than people ordinarily imagine.
Maslow also developed a hierarchy of motives.
These include physiological needs, safety needs, belong
ingness and love needs, esteem needs, and the need for
self-actualization. The lower needs he calls D-motives
(Deficiency motives) and the higher needs are B-motlves
(Being motives). The higher motives generally come into
play only when the lower motives have been satisfied. In
discussing the need for self-esteem Maslow stated:
Satisfaction of the self-esteem need leads to feel
ings of self-confidence, worth, strength, capability,
and adequacy of being useful and necessary in the
world. But the thwarting of these needs produces
feelings of Inferiority, of weakness, and of help
lessness. These feelings give rise to either basic
discouragement or else compensatory or neurotic
trends [p. 3 8 2].
This discussion of self-esteem needs indicates those per
sonality characteristics 'that Maslow would see as indica
tive of both high and low self-esteem. Maslow, however,
would focus on high esteem. He believed that psychologists
had given too much attention to man's deficiencies and that
the emphasis should have been turned to helping people in
the development of their positive qualities.
• 1
35
Although the philosophical positions of many other
self-theorists might be discussed,, the theories just pre
sented seem to be basic in the thinking underlying the four
tests submitted for critical analysis in this research. In
the opinion of many psychologists* all measures of con
structs such as self-esteem or self-concept must be based
upon a set of sound theoretical postulates. In Chapter IV*
where the evaluation findings of this study are set forth*
references will be made to the adequacy or inadequacy of
the theory that supports each of the four tests under con
sideration.
The Development of Measures for Self-Regard
In the first part of this section of the review of
the literature, the difficulties that have hampered the de
velopment of reliable and valid measures of self-regard
are set forth. These concerns need to be presented so that
a sound judgment can be made on the way they were handled
in the four tests selected for analyses in this research.
The second part of this section reviews many of the in
struments that have been used to measure self-concept.
Such a review permits meaningful comparisons to be made
with the tests that are subjected to careful scrutiny in
this study.
36
Difficulties in the Measurement
of Self-Coneep¥
The difficulties of measuring self-concept have
"been described in other works (Lowe* 196lj Wylie, 1961;
J. Micha-el, 1970). Although these concerns are not easily
resolved, some authors have made valuable suggestions.
Those of many authors, including Cronbach and Meehl (1955)
and Campbell and Fiske (1959) are cited but not discussed
in a later section of this chapter which deals with spe
cific procedures for establishing test reliability and
validity. In some instances, however, the discussion of
the difficulties of measurement in this section naturally
includes suggestions for resolving the difficulties.
One of the most widespread obstacles to the meas
urement of self-concept is the lack of adequate theoretical
foundations. Without an adequate set of theoretical pos
tulates, it is difficult to select or design items to
measure the soundness of the construct being investigated.
A theory of self-concept is ultimately tied to the defini
tion of self-concept proposed by the theory.
A multitude of definitions for the self exists.
Wylie (1 9 6 1), who reviewed 463 articles on self-concept
stated: "An examination of empirical studies makes it
apparent that ambiguities in the measuring instruments can
be traced to inadequacies in the theorists’ definitions of
their terms [pp. 3-4]." La Benne and Greene (1 9 6 9) con
37
tended that there has been so much lack of agreement on the
definition of self-concept that the term itself is about
all that many studies have in common.
The term self-concept is, however, not universally
used. There are many ways of referring to an Individual's
feelings regarding himself. One researcher discussed sev
eral different types of self-reference:
There are, then, a number of different self-pictures
which may be psychologically important to the individ
ual: his present self-image; his committed self-image
(the type of person he would like to be if unencum
bered by reality); the ego-ideal (the type of person
he feels he should be); the future or possible self
(the type of person he feels he may become); and the
idealized image (the type of person he most enjoys
thinking of himself as) [which may include components
of his present self, his ego-ideal, his future self,
etc.]. In addition, one would wish to learn about
the "presenting self"— the picture of the self that
the individual attempts to set forth to the world
[Rosenberg, 19S5» p. 2751*
Coopersmith (1959) suggested that there are ac
tually four types of self-esteem: what a person purports to
have, what he really has, what he displays, and what others
believe he has. Other terms of self-reference often used
include the inferred self-concept which is based upon ob
served behavior; the phenomenal self as a measure of con
scious self-regard, the nonphenomenal self as a measure of
unconscious self-regard, and self-esteem as an attitude of
positive feeling or negative feeling that an Individual
holds toward himself as an object.
This last definition often applies also to terms
38
like self-satisfaction, self-acceptance, self-favorability,
or difference between the real-self and ideal-self. Terms
like self-regard or self-perception seem somewhat less eval
uative, but all these terms are often used interchangeably
in the literature.
It is obvious that not all of these terms measure
the same variable. One especially knotty problem arises
in the use of the self-report. Since the phenomenological
theories of self mentioned earlier in this chapter define
the self as the individual’s conscious attitudes and feel
ings toward himself, it is logical to ask the person how
he feels about himself. This type of verbal self report is
beset with possible shortcomings. Combs and Soper (1957).>
Gordon and Combs (1958), Combs, Soper and Courson (1 9 6 3)*
Courson (1 9 6 5)* and Parker (1 9 6 6) have contended that self-
reports and the self-concept are actually different psycho
logical constructs.
Combs and Soper (1957) listed the following as fac
tors which may Influence an individual’s self-report: (a)
the clarity of the subject’s awareness, (b) the availabil
ity of adequate symbols of expression, (c) the social ex
pectancy, (d) the willingness of the subject to cooperate,
and (e) the individual’s feeling of personal adequacy and
freedom from threat (pp. 138-139)-
To avoid the types of errors introduced by these
factors the investigators have preferred to infer self-
concept from samples of behavior.
There are problems, however, in making inferences
about a person's feelings on the basis of overt behavior.
Murphy, Murphy, and Newcomb (1937) declared that actions
are no more valid than words. In fact, actions are almost
always subject to social pressures from others who are
present. When safeguards for anonymity are clear to the
respondent, his verbal behavior, they indicated, may be
more valid than other observable behavior which is subject
to courtesy, expediency, or other social pressures.
It seems to this writer that any single approach
to determining self-concept may lead to errors that could
be overcome by combining several types of measurement.
Verbal self-report, Inferred self-concept based on observed
behavior, and some measure of unconscious self-regard could
be combined to obtain a more valid measure of self-concept
than any one of these approaches would provide when used
alone. A way to avoid much of the confusion that now
exists when the term self-concept Is used would be for each
researcher to define carefully the type of self-reference
with which he is concerned and the method or methods used
in measuring the construct.
40
Sources of Invalidity in the
Measurement of the Self-Concept
In an earlier discussion of the work by Combs and
Soper (1957), several factors were listed which may in
fluence the self-report as a measure of self-concept.
Wylie (1 9 6 1) also discussed variables that may produce
Invalid results In self-concept studies. She called these
factors irrelevant response determiners.
In the following discussion the writer has at
tempted to summarize those variables most often considered
as possible sources of invalidity in the measurement of
self-concept.
It has been established (Edwards, 1957» Kenny,
1 9 5 6) that high correlations are found between the mean
probability of endorsement of self-report items and the
scaled social desirability values of the items. Wylie
(1 9 6 1) maintained, however, that even though a response may
be predicted on the basis of Its social desirability, such
a response may still be valid as an indicator of conscious
self-concept. First, Items may be paired according to
their scaled social desirability values, and then the
subject may be forced to choose between these values. Al
though this procedure reduces the measured effect of social
desirability, it introduces other errors such as negativity
on the part of the respondent who may become frustrated and
41
careless In making responses (Levonian, Comrey, Levy, &
Procter, 1959)*
The forced-choice method may he grouped with other
methods of measurement which reduce validity by restricting
the range of responses. The semantic differential, the
Q sort,, and rating scales are all techniques of measure-
ment that may circumscribe possible responses and thus
prevent the subject from giving an accurate report of his
feelings.
The design of measuring instruments may introduce
many irrelevant response determiners. These include the
tendency to acquiescence, the halo effect, or the tendency
for the subject to check one end of a scale or a restricted
range of a scale almost exclusively. In discussing re
sponse styles, Cronbach (i9 6 0) noted that the use of fixed
categories such as "yes," "agree," and "like" makes ques
tionnaires especially subject to individual response
biases.
Jackson and Messick (1 9 5 8) maintained that, when a
response set has considerable effect, it is difficult or
Impossible to Interpret self-reports as if the face content
were true. Earlier, Cronbach (1950) discussed response
sets In relation to test design and concluded that a
response set is an enemy to validity. He made the follow
ing procedural suggestions for minimizing response sets:
42
We should keep response sets from affecting the test
score by one of the following methods: designing test
items which prevent response sets, s,ltering directions
to reduce response sets, or correcting for response
sets [p. 21].
The evidence for the importance of assuring anony
mity to respondents when using a self-report to measure
conscious self-image is not entirely convincing. In one
study, Parker (1 9 6 6) found a rather high average correla
tion of .74 between signed and unsigned self-reports com
pleted by 30 sixth-grade children. In any case, anonymity
would seem to offer some control over the subject*s tend
ency to respond in a socially desirable fashion. To reduce
the chance that subjects will try to fake answers in order
to look "good," some test-makers have attempted to conceal
the purpose of the test. It should be noted that Cronbach
(i9 6 0) Indicated that using this ploy skirts the edge of
unethical practice and that it may, in the long run, lead
to greater evasiveness on the part of subjects who work
with psychologists.
Statistical procedures that fail to explain the
amount of common variance or independent variance con
tributed by each separate scale in a self-concept Inventory
confound the interpretation of all Investigations that make
use of the instrument. Most self-regard measures are used,
however, without such data. The difficulty of sound test
construction was emphasized by Kerlinger:
43
The construction of objective tests and scales is a
long and arduous task. There are no shortcuts. A
poorly constructed instrument may do more harm than
good, because it may lead the investigator to erron
eous conclusions [1 9 6 6, p. 492],
Alternative Approaches to the
Measurement of the Self-Concept
The measurement of self-concept has been attempted
by using several diverse approaches. The usual methods
used by researchers for the measurement of self-image have
included some type of self-report such as rating scales,
questionnaires, adjective check lists, Q sorts, or a form
of the semantic differential.
Some investigators have tried to tap self-concept
by the use of projective devices. The most commonly used
of these devices are the Draw-A-Person, the Thematic Ap
perception Test, and the Rorschach.
An indirect approach to the measurement of self-
concept that affords additional support for the validity
of self-concept instruments is provided by investigations
that involve the correlation of self-concept with related
variables. In other investigations, researchers have
demonstrated how certain experiences may affect the self-
concept. Examples of self-concept Investigations include
studies of the relationship of self-concept to anxiety
(Lipsitt, 1958j Coopersmith, 1959); the effect of public
commitment on self-concept (Lewin, 1958; Festinger & Carl-
smlth,1 9 5 9); the relationship between self-acceptance and
- 44
the acceptance of others (Fey., 1955)j the agreement between
the way the person sees himself and the way others see him
(Calvin & Holtzman, 1953; Coopersmith, 1959); the relation
ship of the self-concept and the ideal self (Rogers &
Dymond, 1954); the influence of self-concept on achievement
or vice versa (Coopersmith, 1959.; Bruck & Bodwin, 1962;
Sears & Sherman, 1964; Williams & Cole, 1 9 6 8); the effect
of child-rearing practices on the self-concept (Ausubel,
Balthazar, Rosenthal, Blackman, Schpoont, & Welkowitz,
1954); the stability of the self-concept (Lecky, 1945;
Brownfain, 1952; Engle, 1 9 5 8); and the relationship of the
level of aspiration to self-concept (Steiner, 1957);
Coopersmith, 1 9 6 8). The popular practice to validate self-
concept measures against objective personality tests has
been generally somewhat successful (Lowe, 1 9 6 1).
The problems of measuring self-concept and the
general methods used in assessing self-concept have been
presented. In Chapter IV, the findings of this research
are partially based on the information that has been dis
cussed in this section of Chapter II. That Is, the tests
selected for analysis In this study are examined and evalu
ated In the light of the material presented In this review
of the literature.
Procedures for Reporting Test
R e l i a b i l i t y an(f Validity
Procedures for establishing test reliability and
validity are covered in depth in standard textbooks such as
those by Adams (1964), Cronbach (i9 6 0), and Guilford (1 9 6 5).
In this review, only the Information basic to a critical
analysis and.subsequent evaluation of the four measures of
self-regard selected for study in this research will be
presented. It is assumed that the reader is acquainted
with the basic concepts of reliability theory:
Reporting evidence about test reliability. Although
it has been traditional to classify reliability coefficients
into various types such as a coefficient of stability, a
coefficient of equivalence, a coefficient of internal con
sistency, and a coefficient of equivalence and stability,
French and Michael (American Psychological Association,
1966) pointed out In the Standards that this terminology
breaks down as more adequate statistical analyses are ap
plied to determine sources of error variance. They sug
gested the use of multivariate reliability experiments that
have controls for such factors as those of content, time,
and mode of administration. Instead of using such a term
as coefficient of equivalence and stability, they recom
mended that test authors use suitable phrases to convey
the meaning of the coefficients they report, French and
46
Michael (1 9 6 6), for example, suggested: "The expression,
•the stability of measurements by different test forms as
determined over a 7-day interval, 1 although lengthy will be
reasonably free from ambiguity [p. 2 7]." They also empha
sized that "clearly labeled components of errors variance
is the most important outcome of reliability study [p. 2 6]."
Regarding evidence about test validity. The valid
ity of a test is related to the usefulness of the instru
ment in measuring the variable that it purports to measure.
Guilford (1 9 6 5) emphasized that it is not possible to dis
cuss the validity of a test in general. The validity must
be measured in connection with a specific purpose for the
test. French and Michael (American Psychological Associa
tion, 1 9 6 6) stressed this point in slightly different terms.
They stated: "Validity information indicates the degree to
which the test is capable of achieving certain aims [p.
12]." These authors then went on to explain that various
types of judgments require different types of validity in
vestigations .
They also differentiated among three types of va
lidity coefficients: content validity, criterion-related
validity (which is a combination of the well known terms
of predictive validity and concurrent validity), and con
struct validity. It is assumed that the reader is ac
quainted with this terminology that is described in many
47
introductory measurement textbooks. Furthermore, it is
assumed that the reader is also familiar with the funda
mental contributions of Cronbach and Meehl (1955) and
Campbell and Fiske (1959) to the understanding of construct
validity.
Research Studies Related to the Reliability
and Validity of Four Selected Measures
of Self-Regard
This section of the chapter attempts to summarize
the findings of those research studies that bear directly
on the reliability and validity of the instruments sub
jected to critical analysis in this study. In The Seventh
Mental Measurements Yearbook edited by Buros (1972), all
known references on each test are reported. These refer
ences include unpublished articles, books, and unpublished
theses on the construction, validity, use, and limitations
of each test. These references were investigated and re
viewed for information that might contribute to as complete
an analysis as possible of the reliability and validity of
the tests under scrutiny in this dissertation.
The PHCSCS '
The manual for the PHCSCS was published in 1969*
Consequently it was possible for the authors to include in
the manual the results o'f several recent studies that fur
nished data on the reliability and validity of their in
strument. Five of the eight references listed in Buros
48
(1972) are also mentioned In the test manual. The findings
of these five studies are reviewed in the following chapter
of this dissertation which is devoted to an objective pres
entation of the reliability and validity data reported in
the test manuals.
The three PHCSCS references listed in Buros (1972)
and not discussed in the test manual were investigated by
the writer to ascertain whether they might throw additional
light on the reliability and validity of the PHCSCS. In an
investigation by Ashby (1968), four groups of third-grade
Sunday school children were given the PHCSCS before and
after 16 weeks of instruction. However, there were no
significant differences on the average posttest scores
among the three control groups and the experimental group
taught by the investigator. Similarly, in a study of
changes in the self-concept of an experimental group of
third-grade pupils after eight group counseling sessions
as compared with those of a control group of children who
remained in their classrooms, Hugo (1 9 6 9) found no signifi
cant differences in average PHCSCS scores. Using the
PHCSCS to study possible differences in the self-concepts
of fifth-grade and sixth-grade white and Negro children in
a suburban school system that had been integrated for two
years, Sisenwein (1970) also obtained no significant mean
differences in the global self-concept of the two racial
groups.
49
One study not listed in Buros (1972) lent some sup
port for the validity of the PHCSCS. Bolea, Felker, and
Barnes (1971) used a set of 50 cartoon-like picture cards
to measure the self-concept of children in grades K to 4.
In a group of 63 children they found a correlation of .42
between their instrument and the PHCSCS. This relationship
was significant at the .01 level.
As indicated earlier the authors of the PHCSCS in
cluded most of the research results on their instrument in
the test manual. This research is discussed in the follow
ing chapter of this dissertation.
The TSCVT
The TSCVT is a relatively new instrument. It was
first copyrighted in 1967 and the revised manual was pub
lished in 1 9 6 9. Apparently the test has not been well
publicized until recently. Buros (1972)., who has pur
ported to report all known references on each test re
viewed, listed no articles, books, or unpublished theses
on the construction, validity, use, or limitations of the
TSCVT.
The information in the TSCVT manual that deals with
the reliability and validity of the test is presented in
Chapter III of this study. Although the author of the test
made considerable reference to disadvantaged children and
to head-start programs in the manual, Swinn (1972) con-
50
eluded that "the TSCVT has not reached the stage of psycho
metric development which would justify its current use as a
means of making major decisions related to such matters
[p. 1543."
In support of the proposition that self-concept
studies are appropriate for young children, the test author
referred to a review of self-concept studies made by Woolner
(1 9 6 6). The writer investigated the work by Woolner and
found that she did present evidence that the self-construct
concept could be measured in kindergarten children. She
also concluded that the five-year-old children in her study
were able to express attitudes toward themselves. Thus
some theoretical support was lent to the validity of the
TSCVT, since it attempts to measure the self-concept of
young children by having them express attitudes toward
themselves. Statistical evidence of relationships between
the TSCVT and other self-concept measures is apparently not
available at this time.
There is a very limited amount of information on
the relationships between self-concept as measured by the
TSCVT and other variables. One study on such relationships,
which is presented in the test manual, is described in the
next chapter as a part of a complete description of the
TSCVT and its manual.
The TSCS
The TSCS manual was published In 19&5• Several re
search studies relating to the validity of the test, which
are discussed in the manual, are reviewed in Chapter III,
which describes the test and the content of the manual.
The TSCS which has generated much interest, has been used
in many investigations since the manual was published.
The Seventh Mental Measurements Yearbook (Buros, 1972)
lists 118 references for the TSCS. Of these 118 refer
ences, 8 2 have been published since 1965. This writer in
vestigated all 118 references to assess their relevance in
a study of the reliability and validity of the TSCS. Since
it would be impractical and unprofitable to review each
TSCS reference In this study, only the findings of those
references which bear directly on the reliability and valid
ity of the test are summarized.
In one reference (Nunnelly, 1 9 6 8), a reliability
coefficient of .91 was reported for total P, which Is the
TSCS scale that best reflects self-esteem. This reliabil
ity measure was obtained by using a Kuder-RIchardson split-
halves technique. Thus, the coefficient supplements the
reliability data in the manual which reports no measure of
internal consistency.
In a factor analytic study of the TSCS, Rentz and
White (1 9 6 7) used 12 of the 29 TSCS scores to determine
52
whether or not the test measures several different aspects
of self-concept. The TSCS was given to 138 university stu
dents. The scales of the test studied Included: (l) Iden
tity,, (2) Behavior., (3) Self-Satisfaction, (4) Personal
Self, (5 ) Family Self, ( 6 ) Physical Self, ( 7) Moral-ethical
Self, (8) Social Self, (9) Self-Criticism, (10) Variabil
ity, (ll) Certainty, and (12) Conflict. These 12 scales
loaded on two factors in a varimax rotation of a principal
components solution. Rentz and White (1 9 6 7) concluded
that "the five major dimensions of self-concept in the
TSCS are aspects of only two independent factors [p. 118]."
In another factor analytic study Vacchlano and
Strauss (1 9 6 8) administered the TSCS to 260 college stu
dents, computed product-moment correlations among the 100
scale Items, and extracted 20 interpretable factors. The
investigators reported: "Factor formations were based on
items having rotated factor loadings above .40 [p. 324]."
Based on their findings, these investigators Indicated
that the TSCS measured five aspects of the self: (a) physi
cal, (b) moral-ethical, (c) personal, (d) family, and (e)
social. They concluded that "the factor analysis performed
would substantiate the construct validity of the TSCS
[p. 326]."
Data from a study by George (1 9 6 9) may be related
to the validity of TSCS. When subjects were asked to re
spond to test Items in terms of how they would like to be
53
In contrast to their actual self-concepts, the Self Criti
cism or Lie Score dropped almost one standard deviation and
the Defensive Positive Score (a more subtle measure of de
fensiveness than the Self-Criticism Score) increased by
about the same amount. These results afforded evidence of
the sensitivity of these scores to defensive distortion.
Jones (1966) studied the response bias of the TSCS.
He asked college students (73 males, 21 females) to respond
to items with four different sets: (l) as though they were
part of a research sample, (2) as if they were asking a
psychiatrist for help with their problems, (3) as they
would respond to a psychologist appointed by a Judge when
charged with law violation, and (4) as if they were com
pleting the form for a job application. The moral-ethical
self-concept seemed most subject to variability. On the
Self-Criticism or Lie Score, Jones (196 6) found subjects to
be almost completely truthful in the Help situation and
only slightly less truthful in the Research situation.
Subjects were less truthful in a Judge situation and most
deceitful in the Job situation. These data might seem to
support the sensitivity of the Self-Criticism Score as a
measure of the truthfulness with which subjects respond to
the TSCS. However, they could also be interpreted as evi
dence that response set does have considerable effect on
test scores. Such an effect would be expected to alter
test reliability and test validity.
54
In a monograph entitled The Self-Concept and Self-
Actualization, Fitts, et al.(197T) reviewed reliability and
validity data on the TSCS. The material In the monograph
paralleled and elaborated on the information presented in
the foregoing paragraphs. In the text of the monograph,
the writers also promised two future monographs that would
offer additional evidence to support the validity of the
TSCS.
The SEI
The SEI is not listed In Buros (1972) because the
test is not published commercially. There is, therefore,
no complete recent summary of the research studies that
have used the SEI. However, the test author, Coopersmith,
has distributed, upon request, a comprehensive bibliography
of research in the area of self-esteem. Using this bib
liography and extensive library research, the writer se
lected those findings on the reliability and validity of
the SEI which are reported In the following paragraphs.
These data supplement those found in Coopersmith1s (1 9 6 7)
book and in the memorandum which the author has sent to
those who request information on his test. The data from
these sources, Coopersmith^ book and his memorandum, are
discussed in Chapter III of this dissertation.
Dyer (1 9 6 3) cited a reliability coefficient of .6 5
for a sample made up of approximately 50 students in each
55
of 10 groups. The sample included boys and girls in grades
three, five, seven, nine, and eleven. In another study
bearing on test reliability, Campbell (1 9 6 5) tested 158
students and concluded that the question of internal con
sistency should be answered affirmatively for the SEI.
In an earlier study, Coopersmith (1959) gave a
description of the reliability sample which he later pre
sented in his book (Coopersmith, 1 9 6 7). In his book he
reported that the SEI had been administered to two fifth-
and sixth-grade classes of boys and girls and that five
weeks later it had been given to one of the fifth-grade
classes. For a sample of 30 fifth-grade children, test-
retest reliability was estimated to be .88. In his ear
lier work Coopersmith (1959) apparently described the same
sample in greater detail. He stated that the fifth- and
sixth-grade children, who were 10 to 12 years of age,
attended public schools in a small eastern city. The
group consisted of 49 girls and 53 boys who lived in or
near a neighborhood judged to be middle-middle to upper-
middle class. These additional data on the reliability
sample, if mentioned in the book (Coopersmith, 1 9 6 7),
would give test users an improved basis for judging the
reliability of the test for diverse research populations.
In the same study, Coopersmith (1959) cited cor
relations that might lend evidence for the validity of the
SEI as a measure of self-esteem. Using 87 students from
56
the sample described in the preceding paragraph, Cooper
smith found a correlation of .3 6 between the SEI and the
Iowa Achievement Test. Children in the four classes de
scribed earlier were asked to Indicate the three children
whom they would most like to have as their friends. These
sociometric choices correlated .37 with the SEI. If one
assumes that popularity or good peer relations and high
achievement are indications of high self-esteem, then these
findings, both significant at the .01 level, might be con
sidered as limited evidence for the validity of the SEI.
In another study, Coopersmith (1 9 6 8) related meas
ures of unconscious self-evaluation to self-esteem as meas
ured by the SEI. While the quotation which follows Is made
with almost no additional supporting data, It does show
that Coopersmith believed that the findings supported the
validity of his Instrument.
We employed psychological tests (the Rorschach and
the Thematic Apperception Test) that Indicate a per
son's unconscious self-evaluation. We found that in
more than 80 percent of the cases the ratings by these
Indexes were substantially in accord with the boy's
own estimate of his self-esteem [p. 9 8].
Trotter (1971) reported the results of a four-year
project in the central United States involving a sample of
3,800 students between the ages of 8 and 13. Comparing
scores on the SEI with those on Intelligence tests, an un
expected finding resulted. Students scoring above the
90th percentile on group intelligence tests did place
57
lower on the SEI than did students scoring between the
4oth and 6 0th percentile on intelligence tests. Trotter
quoted Trowbridge., who directed this research, as stating,
"that children of low socioeconomic status (SES) scored
higher than children of middle SES at all ages, in both
sexes, black and other races, and In rural as well as
urban areas [p. 131]." These findings, based on self
esteem as measured by the SEI, may be seen by some theo
rists as so unexpected that the validity of the measuring
instrument is called into question.
In a factor analysis of the SEI, Richmond and White
(1 9 7 1) reported that the responses of 204 children were
factor-analyzed by the principal components solution.
Five factors with eigenvalues greater than 1.0 were
extracted and rotated to the varimax criterion. The labels
given to the five SEI factors by these investigators were:
(a) self-rejection, (b) parental approval, (c) rejection
by authority, (d) social-self acceptance, and (e) a lie
scale. The authors summarized their findings by conclud
ing: "Five factors emerged from the SEI, but they bear
slight resemblance to the factor structure or labels re
ported by Coopersmith [p. 72]."
It was Indicated earlier that additional evidence
for the reliability and validity of the SEI is reviewed in
the next chapter of this dissertation. In addition to pre-
sentlng data from the hook by Coopersmith (1 9 6 7) and his
test memorandum, the following chapter also furnishes the
data on reliability and validity taken directly from the
test manuals for the PHCSCS, TSCS, and TSCVT.
CHAPTER III
THE CONTENT OF FOUR SELECTED MEASURES
OF SELF-CONCEPT
Introduction
Chapter III is devoted to an objective description
of the four measures of selfrconcept that have been se
lected for critical analysis in this research. The descrip
tion will touch upon several aspects of the self-regard
instruments including the content, materials, theoretical
underpinnings, administrative procedures, and test scales.
The four measuring instruments are discussed in the follow
ing order: (l) the Piers-Harris Children^ Self-Concept
Scale (PHCSCS) (Piers & Harris, 1 9 6 9)} (2) the Tennessee
Self-Concept Scale (TSCS) (Fitts, 1 9 6 5); (3) the Thomas
Self-Concept Values Test (TSCVT) (Thomas, 1 9 6 9); and (4)
the Self-Esteem Inventory (SEI) as described in The Ante
cedents of Self-Esteem by Coopersmith (1 9 6 7).
It should be emphasized that the information on
reliability and validity presented In this chapter is an
objective summary of the data provided by the authors in
the test manuals or in the case of the SEI from a book that
serves as a test manual. In all cases objective data are
59
6 0
summarized in this chapter and the critical analysis and
interpretation of information on reliability and validity
are presented in the following chapter.
The Piers-Harris Childrens Self-
Concept Scale '(PHCSCS)'
An Overview of the PHCSCS
The PHCSCS is a self-report instrument designed for
use over a wide age range. Students respond to 80 ques
tions by marking "yes" or "no1 ' on the test booklet. Sample
items are "I am obedient at home," and "I am dumb about
most things."
The manual recommends that questions be read aloud
to pupils below grade 7* A total test score is obtained
by using a scoring stencil. The manual presents norms for
grades 4 through 12. No separate sex or grade level norms
are given. The authors justify this omission by explaining
that no consistent significant differences have been found
between sexes or grade levels. The norms are based on a
sample of 1 ,1 8 3 public school children from grades 4, 6, 8,
10, and 12 from a small town in Pennsylvania. A table in
the manual enables the test user to convert raw scores to
percentile scores and to stanine scores.
A need for caution in the interpretation of test
scores is set forth in the manual. The scale was developed
primarily for research. The standard error of measurement
61
was estimated to be 6 points. It is recommended that in
dividual changes in score of less than 10 points be ignored.
A factor analysis of the scale for 330 sixth-grade
pupils was made. The factors accounted for 42 percent of
the variance. Six factors were large enough to be lnter-
pretable: (l) Behavior, (2) Intellectual and School
Status, (3) Physical Appearance and Attributes, (4) Anxiety,
(5) Popularity, and (6) Happiness and Satisfaction. The
number of the factor or factors as they are reflected in
responses has been marked on the scoring key. The sums of
these items answered in a given direction can be calculated
to obtain a cluster score. The authors emphasize again
that the most appropriate use for cluster scores is in fur
ther research.
Reliability of the PHCSCS
The reliability of the PHCSCS has been investigated
in a number of studies reported in the manual. After the
Kuder-Richardson Formula 21 had been applied to data from
six samples, resulting coefficients ranged from .7 8 to .9 3.
The Spearman-Brown odd-even formula was applied to samples
of sixth grade and tenth grade pupils with resulting coef
ficients of .9 0 and .87* respectively. Samples of children
from grades three, six, and ten were retested after four
months. The coefficients obtained were .7 2, .71* and . 7 2.
In a study by Wing (1 9 6 6) a group of fifth-grade
6 2
pupils was retested after two months and again after four
months. Both reliability coefficients obtained were .7 7.
The Conclusion presented in the manual is that the scale
was judged to have good internal consistency and adequate
temporal stability.
The test authors pointed out that group means on
retesting were consistently in the direction of higher self-
concept scores. They suggested that this change might have
been due to increased familiarity with items. Since scores
might have been higher on retest without treatment, the
authors stressed the need for using control groups before
any claims could be made regarding changes in self-concept.
Validity of the PHCSCS
Content validity for the PHCSCS was built into the
scale. Items were developed from Jersild!s (1 9 5 2) collec
tion of children’s statements regarding what they liked
and disliked about themselves. Jersild grouped children's
statements into 11 categories such as "Physical Character
istics and Appearance," "Ability In School," or "Attitudes
Toward School." Items were written to cover all 11 areas.
The authors began with 164 statements at least half
of which were negative in content (e.g., "I behave badly at
home") in order to reduce the possible effect of a tendency
toward acquiescence. Twelve "lie" statements such as "I am
always good" and "Sometimes I act silly," were included to
63
determine whether a reduction in test validity would occur
when children consistently responded in a socially desir
able fashion. The lie statements were later dropped when
it appeared that these statements did not markedly affect
test scores.
A pilot study with 90 third-* fourth-* and sixth-
grade pupils was conducted. Most items answered in one
direction by fewer than ten percent or by more than ninety
percent were dropped from the scale.
Following the pilot study a 140-item scale was ad
ministered to four third-grade classes* four sixth-grade
classes, and four tenth-grade classes. Statements were
classified by three judges as reflecting high or low self-
concept. From the sixth-grade sample of 127 students, the
30 highest and 30 lowest scores were identified. On each
item Cureton's chi-square test (Lindquist* 1951) was ap
plied to identify items that discriminated between the high
and low groups at the .05 level or better. These items
were retained. The final scale consists of 80 items or
statements that children can respond to with a "yes" or
"no" answer.
Information about how the PHCSCS correlates with
another self-concept instrument and with other traits re
lated to self-concept is presented in a section of the test
manual entitled "Validity." In a study by Mayer (1 9 6 5)
results on the PHCSCS correlated .68 with scores on Lip-
sltt's (1958) Children's Self-Concept Scale. Cox (1 9 6 6)
found correlations of .43 and .31 between the PHCSCS and
teacher and peer ratings, respectively. In the same study
negative correlations were obtained between the PHCSCS and
(a) Health Problems (-.48) and (b) Big Problems on the SRA
Junior Inventory (-.64). Cox also found a positive corre
lation of .40 between the PHCSCS and teacher ratings of
Children's Superego Strength. In a 1965 study with fourth
and sixth graders, Piers obtained correlations with teacher
ratings that ranged from .0 6 to .41 while correlations with
peer ratings ranged from .2 6 to . 4 9.
The authors reported the results of a study (Piers
& Harris, 1964) that they thought relevant to the construct
validity of their scale. In this investigation a predic
tion of relatively lower self-concept scores for institu
tionalized retarded females was confirmed.
In a separate section of the manual the test auth
ors discussed other correlates of self-concept that might
lend support to the validity of the PHCSCS. Correlations
with IQ measures ranged from -.04 to .50, and with achieve
ment tests the correlations ranged from .0 6 to .43. Millen
(1 9 6 6) found positive correlations significant at the .01
level between the PHCSCS and a children's social desirabil
ity scale. The same investigation yielded negative corre-
65
lations also significant at the .01 level between the PHCSCS
and the Children's Manifest Anxiety Scale. In the Cox
(1 9 6 6) study a measure of perception of parents as loving
rather than rejecting correlated .5 6 with the PHCSCS and a
measure of peer acceptance correlated .61 with the PHCSCS.
Both correlation coefficients were significant beyond the
.01 level.
The factor analysis reported in the manual indicates
that at least six factors might be present in the test.
The authors did not attempt to justify the presence of six
factors in a test designated to measure the supposedly
single trait of self-concept. It would seem that a single
set of norms might lose some useful information if the
scale should be truly six-dimensional.
The Thomas Self-Concept Values Test
(TSCVT)
An Overview of the TSCVT
This test Is an individually administered form of
self-report. The manual indicates that It can be used with
children, ages 3 to 9. The examiner takes a Polaroid pic
ture of the child and uses this picture throughout the test
to keep the child's attention on himself as the "object’ 1 in
the photograph.
The child responds to 14 bipolar value items for
each of four referents. The bipolar value items include
suoh socially desirable and undesirable traits as happy—
sad, smart— not very smart, and goodlooking— not goodlook-
ing. The four referents are self, mother, teacher, and
other peers. First the child is asked whether he is, for
example, happy or sad for all 14 value items. Scores are
recorded as +1 for a positive response and -1 for a nega
tive response. The scoring on two value items (big— little
and strong— weak) are reversed for girls. Next the child
is asked in order whether his mother, teacher, and other
peers think that he is happy or sad. Each time the child
responds to all l4 value items before the examiner intro
duces the next referent. In the administration of the
test, the use of personal pronouns Is avoided and the
child's name is repeated whenever a reference to the child
Is made in each question.
The examiner records responses on the Individual
Record Sheet as the test is administered. When the test
is completed, the child has responded to happy— sad with
four different referents (self, mother, teacher, peer).
The four raw scores associated with these referents are
summed to give one value score. Since there are 14 value
items, the test produces 14 value scores. The 14 items to
which the subject has responded with self as referent are
summed to give a referent score. The four referent scores
are self, mother, teacher, and peer. The totals for the
67
four referent scores are summed, to yield a total self-
concept score. Thus the scale furnishes 19 scores; 14
value scores, four referent scores, and one total self-
concept score.
The manual provides tables so that all raw scores
can be converted to standard scores. The standard scores
can then be plotted on the Individual Profile Sheet which
is on the reverse side of the Individual Record Sheet. The
examiner can see graphically how the Individual's scores
compare with the mean values for each of the 19 scores that
the test provides.
Reliability of the TSCVT
The manual presents the results of an item analysis
of the TSCVT. A tetrachoric correlation was computed be
tween each item and the remaining items in that respective
scale. Data for these item-analysis correlation coeffi
cients came from a random sample of 75 children selected
from a population of 750 children. For each of the 19 dif
ferent distributions subsamples of students placing in the
upper or lower 27 percent of the total score distributions
were used in the item analysis for each item In that scale.
The resulting correlations ranged from .10 to .93. One
hundred and nine of 126 item-scale coefficients were above
.50. The manual states that these Item-scale relationships
show that the test has internal consistency, and, thereby,
provides evidence of a type of reliability.
A test-retest reliability estimate was obtained by
having 3^ subjects retested by the same examiner from 9 to
14 days later,. A Pearson product-moment correlation was
computed for the test-retest score distribution. Eighteen
of the 19 coefficients were reported in the manual. (The
mother referent scale was omitted.) The range was from
.34 to .93. The median coefficient for the distribution
was reported to be .75. The reliability for the total
scale was estimated to be . 7 8.
The total self-concept score on the TSCVT is ob
tained by combining the scores on the four referent scales
(self,, mother., teacher, and peer). The scores of 50 chil
dren on the four referent scales were correlated through
use of the Pearson product-moment method. The paired com
parison referent scales were self vs. mother, self vs.
teacher, self vs. peer, mother vs. teacher, mother vs.
peer, and teacher vs. peer. These correlation coefficients
ranged from .6 0 to .80. The average (Fisher z_ transforma
tion) of the distribution was reported as .73. The inter
correlations among the four self-referent subscales were
thought by the author to provide evidence of the instru
ment ' s internal consistency. He concluded that the mean
value of the distribution of correlations was a useful
estimate of the total test reliability.
Validity of the TSCVT
The author used the first 16 pages of the manual
to review the literature and the research that he con
sidered pertinent to the development of the TSCVT. In
these pages he established that there was a paucity of
self-concept measurement data for preschool and primary
school children. He said that there were no factor analy
tic studies available on values at early childhood ages.
Prom literature and research he selected 14 factors that
he believed were culturally approved and necessary for
normal personality development. Some of these factors
were supported by research studies, but others were ap
parently included on the basis of cultural standards or
values proposed by an anthropologist. Thomas stated:
"The fourteen self-value factors, value items, are in
cluded in this instrument as a result of speculative deci
sion [p. 14 ]."
The use of the photograph and the selection of the
four referent scales are explained as follows. The test
is intended, according to the manual, to combine the theo
retical basis of the self provided by Mead (1934) and Sul
livan (19^7 ) with the value concept of the test author.
Mead believed that an individual becomes an ''object" to
himself by taking the attitudes of other individuals toward
himself. Sullivan stressed the Importance of significant
70
others In the development of the self-concept. It is ap
parent that the theoretical underpinning of the TSCVT
stresses the need for young children to assume the role of
significant others in replying to questions.
To support the construct validity of the TSCVT the
author summarized and defined 14 psychological constructs.
These constructs, he believed, furnished the basis for the
development of his instrument and lent support to the valid-
ity of the test. The constructs defined did seem to sup
port the test format and test content in general, but they
were not related to the selection of each individual test
item. The author concluded by stating: "The particular
item format, the logically selected Item content, the In
dependent scale scores are all cited as maximizing the pos
sible construct validity of the Thomas Self-Concept Values
Test [p. 34]."
For concurrent validity the manual presents the
results of a study with underprivileged children and privi
leged children. It was hypothesized that the privileged
children as compared with the underprivileged would have
higher self-concept scores on the TSCVT. In 14 of the 19
group analyses mean differences significant at the .05
level were found. These differences were all in the hy
pothesized direction, and, according to the manual, they
supported the conclusion that the TSCVT could discriminate
71
between two criterion groups in at least 14 self-value
factors.
The Tennessee Self-Concept Scale
(frscS)
An Overview of the TSCS
The TSCS consists of 100 self-descriptive items.
Ten of the items, which were taken from the MMPI Lie-Scale,
constitute the Self-Criticism Score. The other 90 items
assess the self-concept. The subject chooses one of five
response options from "completely true" to "completely
false."
The TSCS has two forms: a Counseling Form and a
Clinical and Research Form. Both forms use exactly the
same test booklet and test items. The Counseling Form,
which has fewer subscores, is appropriate for self
interpretation and feedback to counselors. The Clinical
and Research Form, which is more complex than the Counsel
ing Form in terms of scoring, analysis, and interpretation,
is not appropriate for self-interpretation by the subject.
The Counseling Form has 15 scores for interpreta
tion. These scores are described briefly below:
1. Self-Criticism Score (SC). Ten items from the
Lie Scale of the MMPI measure overt defensive
ness.
72
2. Total P Score. The Score on all 90 Items Is
summed to reflect the overall level of self
esteem.
3. Row 1 [yielding] P Score— Identity. Thirty
Items permit the subject to describe what he
is as he sees himself.
4. Row 2 [furnishing] P Score— Self Satisfaction.
Thirty items allow the test taker to tell how
he feels about the self he perceives.
5. Row 3 [providing] P Score— Behavior. This
score comes from 30 items that indicate the
individual's perception of his own behavior.
6. Column A— Physical Self. In these 18 items
the individual describes his body, his health,
his physical appearance, and his sexuality.
7. Column B— Moral Ethical Self. The score from
these 18 items reflects moral worth, relation
ship to God, and feelings about religion.
8. Column C— Personal Self. These 18 items tell
how the subject feels about his adequacy as a
person apart from his body and his relation
ships to others.
9. Column D— Family Self. This score derived
from 18 items reflects the counselee's feeling
of adequacy as a family member.
73
10. Column E— Social Self. These 18 items permit
the subject to express how he sees himself in
relation to other people in general.
11. Total V. This score represents the variability
from one area of self-perception to another.
12. Column Total V. This score measures the varia
bility of self-concept within columns.
13. Row Total P. This score is the sum of varia
tions in self-concept across the rows.
14. The Distribution Score (D). This score is a
count of the number of 4, 3, 2, and 1 re
sponses. It is a measure of how certain or
definite the person is about the way he sees
himself.
15. The Time Score. This score indicates the time
needed, to the nearest minute, to complete the
test. Compulsive, paranoid, and depressed
psychiatric patients do take longer, in general,
to complete the scale than do non-patients.
The. Clinical and Research Form of the TSCS has all
of the scores on the Counseling Form plus these additional
ones:
1. The True-False Ratio (T/F). This index is a
measure of response set or response bias. The
T/F ratio is the sum of the 5 and 4 responses
divided by the sum of the 21s and 1 1s.
Total Conflict Score. The scale is divided
into 15 sections with 3 positive and 3 nega
tive items in each section. The 3 positive
items for each section or cell are totaled
(i.e., one 5-point response, plus two 3-point
responses is 11 points). This sum is called
P for the cell. Negative responses, which are
totaled in the same fashion, represent N for the
cell. The total of P-N for all cells, which is
calculated non-algebraically, or without regard
to sign, is the Total Conflict Score. This
score represents general conflict or confusion
on the part of the subject.
Acquiescence Conflict. This score is obtained
by adding all the P-N scores algebraically.
When P-N is a positive number, it shows that
the subject has over-affirmed his positive
attributes.
Denial Conflict. When the algebraic sum of
P-N yields a negative number, the subject has
denied his negative characteristics.
Defensive Positive Scale (DP). This is one of
six Empirical Scales. The items in each of the
Empirical Scales were determined by an item anal
ysis. There is overlapping of items, since sev
eral of the items are used on more than one scale,
These Empirical Scales were derived from an analy
sis of item responses with the following groups:
Norm Group N = 626
Psychotic Group (Psy) N = 100
Neurotic Group (N) N = 100
Personality Disorder M
Group (PD) N = 100
Defensive Positive
Group (DP)
Personality Integration
Group (Pi)
N = 100
N = 75
The Defensive Positive Scale which consists of
29 items, does represent a more subtle measure
of defensiveness than does the Self-Criticism
or Lie-Score.
The General Maladjustment Scale (GM), The 24
items In this scale distinguish psychiatric
patients from non-patients. This scale pro
vides no clues that help to classify patients
into the psychotic, neurotic, or personality
disorder groups.
7. The Psychosis Scale (Psy). The 23 items of
this scale differentiate psychotic patients
from other groups.
8. The Personality Disorder Scale (PD). This
scale is composed of 27 items which distin
guish individuals in this broad psychiatric
classification from those who might be placed
in some other psychiatric category.
9. The Neurosis Scale (N). These 29 items identify
subjects who are similar to the neurotic group
of patients from whom the scale was derived.
10. The Personality Integration Scale (Pi). The
25 items on this scale Identify individuals
similar to a norm group composed of 75 people
who were judged as average or better in terms ■
of adjustment.
11. Number of Deviant Signs (NDS). This Is another
empirically-derived measure. It is simply a
count of the number of deviant features of all
other scores. This score Is the test's best
index of psychological disturbance.
The TSCS was normed on a sample of 626 persons.
The subjects in the sample ranged in age from 12 to 68 with
approximately the same number from both sexes. Although
both Negro and white subjects and representatives of all
77
social, intellectual, economic, and educational levels were
included in the sample, the norms were overrepresented in
the number of college students, white subjects, and persons
in the 12 to 30 year age bracket.
The scoring of the test involves a rather extensive
procedure. However, it is well organized in the accompany
ing answer sheet, score sheet, and profile package. The
Clinical and Research Form is somewhat more complex than
the Counseling Form, since it also requires the use of
scoring templates. If a large number of tests (50 or more)
are being administered, computer scoring is available di
rectly from the publisher.
Each form of the test (The Counseling Form and The
Clinical and Research Form) has a different score sheet and
profile sheet. After raw scores are plotted on the profile
sheet, both the T-scale value and the percentile score for
all scale and subscales of the test are clearly shown.
Reliability of the TSCS
The reliability data presented in the manual of
the TSCS were based on test-retest approach with 60 college
students over a two-week period. A table in the manual
shows the mean, the standard deviation, and the reliability
for each subscore of the TSCS. Reliability coefficients
were estimated to range from .60 for the Row Total Varia
bility Scale to .92 for the Total Positive Scale which
reflects the general level of self-esteem.
Although reliability coefficients are available for
the measures entitled Empirical Scales., these were taken
from a retest of 60 college students. No data on the re
liability of these scales are furnished based on patient
samples. The TSCS manual also presents no information on
the internal consistency of the scale or the subscales.
The manual does present a table showing a 29 variable
intercorrelation matrix in which many of the subscores cor
relate highly— up to .9 1.
Validity of the TSCS
The validity of the TSCS is discussed in the manual
under four different headings: content validity, discrimi
nation between groups, correlation with other personality
measures, and personality changes under particular condi
tions .
To establish content validity the author compiled
a large pool of self-descriptive items. These were taken
from other measures of self-concept and from self-descrip
tions written by patients and non-patients. After the
items were edited, seven clinical psychologists were em
ployed as judges. They evaluated each item as to whether
it was positive or negative in content. The only items
retained were those for which perfect agreement existed
among the judges.
79
The manual suggests that when such groups as de
linquents, psychiatric patients, normals, and psychologi
cally integrated persons can be correctly identified by the
scale a type of validity has been demonstrated. In one
study the scores of three groups were compared. One sample
was a personality integration group of 75 persons who were
considered average or better in terms of personality ad
justment. A second group was composed of 3 6 9 psychiatric
patients, and the third group was the 6 2 6 non-patients in
the norm sample. A table In the manual presents the mean
and the standard deviation for all three groups on each of
the scores and subscores of the test. Although the author
of the manual called attention to the fact that the means
for virtually every score substantiated original predic
tions, he presented no statistical treatment of differences
between means.
In a second study the author established cutoff
points so as to exclude about 10 percent of the normal
population. Then a comparison was made showing the per
centage of cases exceeding the cutoff points in the origi
nal norm groups and in four new norm groups composed of
(l) an Ohio State University student group, (2) a group at
the Ohio State Hospital, (3 ) a community mental health
center group, and (4) a Veterans Administration psychiatric
hospital group. The author presented these data in a table
80
and concluded that the original level of discrimination
held up rather well with the cross-validation groups.
Using the Number of Deviant Signs (NDS) to discriminate be
tween groups, the author reported that 80 percent of three
patient groups scored above the cutoff point as compared to
20 percent of the original norm group.
As additional evidence that the TSCS discriminates
between groups, the manual reports the results of a study
by Huffman (1964). The author analyzed the profiles of
four different groups (job applicants, paranoid schizo
phrenics, depressive reactions, and emotionally unstable
personalities) prepared from Huffman’s (1964) data. He
believed that his analysis of the profiles indicated that
the TSCS could be used to establish the type of disorder as
well as the degree of disorder.
Finally, as evidence that the TSCS discriminates
between groups of individuals, the author reported the re
sults of several studies that had used the TSCS and pro
duced group test score differences in the predicted direc
tion. Atchison (1958) studied delinquents and non
delinquents; Lefeber (1964) investigated juvenile first
offenders and repeated offenders; Boston and Kew (1964)
worked with unwed mothers; Gividen (1959) studied para
troopers who could stand up under stress and those who
could not; Wells and Bueno (1957) analyzed the test scores
81.
of alcoholics; and Piety (1958) used the test to discrimi- :
nate between groups of individuals with different personal
ity characteristics.
One of the four types of data presented in the
manual to support the validity of the TSCS is degree of
correlation with other measures. The author of the manual
presented correlations between TSCS scores and a number of
other instruments.
All profile variables of the TSCS are correlated
with scores on the Minnesota Multiphaslc Personality Inven
tory (MMPI). The correlations are based on tests from 102
psychiatric patients (McGee, i9 6 0). Sundby's (1 9 6 2) data
on 66 students from three different high schools provided
the scores on the TSCS which were correlated with scores
on the Edwards Personal Preference Schedule (EPPS). In
both of these studies the manual gives a Pearson product-
moment correlation coefficient as well as the correlation
ratios (eta). The manual describes eta as a generalized
measure of the strength of a relationship but not of di
rection. The author did not explain why eta was used. The
data on correlations between the TSCS and the MMPI as well
as the correlations between the TSCS and the EPPS are pre
sented in tabular form in the manual. All correlation
coefficients significant at both the .05 level and .01
level are noted.
82
The author of the manual also reported the correla
tion of the TSCS with some other tests which are less well
known than the MMPI and the EPPS. He indicated that all
the correlations with others measures lent support to the
validity of the TSCS.
The fourth type of data presented in the TSCS
manual as supporting the validity of the test involves
studies that report personality changes under particular
conditions. If enhanced or lowered self-esteem is pre
dicted under certain conditions, and if subjects score on
the TSCS as predicted, this outcome provides, according to
the author, additional evidence for the validity of the
test.
In an unpublished study with six female patients
In group therapy, the author of the manual predicted 88
changes In TSCS scores and subscores. Of the 88 predic
tions, 60 were correct (P < .001). In another study by
Ashcraft and Fitts (1964), 7 65 of 1,110 score changes were
correctly predicted. This study involved an experimental
group of 30 patients who had been in therapy for six months
and a control group of 24 patients who had been waiting
for therapy for an average of 6 . 7 months. The author of
the manual concluded that persons’ self-concepts do change
as a result of significant experiences. Since the TSCS was
judged to reflect these changes in predicted ways, he
83
maintained that this information constituted additional
evidence for the validity of his instrument.
The Self-Esteem Inventory (SEI)
An Overview of the SEI
The objective description of the SEI in this chap
ter is taken from two sources: (a) Coopersmith1s (1 9 6 7)
book, The Antecedents of Self-Esteem, and (b) a memorandum
from the author describing both the procedures he has de
veloped for assessing self-esteem and the methods of ad
ministration, scoring, and interpretation.
The SEI is composed of statements that the respond
ent marks either "like me" or "unlike me." Some sample
items follows:
1. "I often wish I were someone else."
2. "My parents expect too much of me."
3. "I often get discouraged in school."
There are two forms of the SEI. The long form con
tains 50 items to measure self-esteem plus 8 items in the
Lie Scale to measure defensive reactions. The long form
may also be broken down into four additional subscales [the
hyphenation is not consistent]: (l) General Self, (2)
Social Self-peers, (3) Home— parents, and (4) School—
academic. It Is not necessary, however, to score the sub
scales separately. On the long form all items marked with
responses indicating high self-esteem are multiplied by two
so that the total possible score is 100. The Lie Scale,
which is scored separately, helps the tester to interpret
the truthfulness with which the subject responds. The
short form of the SEI contains 25 items and no Lie Scale.
The short form, which was constructed on the basis of the
results of an item analysis of the long form, includes
those 25 items which demonstrate the highest item-total
score relationships. The two forms were shown to inter-
correlate .8 6. On the short form responses Indicating high
self-esteem are multiplied by four so that this form, like
the long form, has a possible maximum score of 1 0 0.
The memorandum received from the author indicates
that the scale may be used with populations ranging from
age 9 to adult level. The author reported that the wording
of several items might need to be altered for older groups
and that children younger than age 9 or pupils who have
less than average reading ability might require an individ
ual administration of the test. The same form is used for
both sexes on the basis of the author's reporting that most
investigators have found no significant differences between
the esteem level of the males and females tested.
In his memorandum the author supplied norms as well
as additional data that help the test user in interpreting
results.
85
Reliability of the SEI
Coopersmith (1 9 6 7) reported the results of two
reliability studies. In one study with a sample of 30
fifth-grade children, a test-retest reliability estimate
of .88 after a five-week interval was obtained. In another
investigation In which 56 children attending the public
schools of central Connecticut were retested after a three-
year interval, a reliability coefficient of .7 0 was found.
The only additional information on the nature of
the two reliability samples is limited to statements indi
cating that both sexes were included. The conditions of
the testing are not specified for the sample retested after
five weeks, but for the sample which was tested again three
years later the author explained that the pupils were
tested in their classrooms under the guidance and super
vision of his research staff.
Validity of the SEI
Most of the items In the SEI were based upon those
selected from the Rogers and Dymond (195^) scale, but sev
eral original items were also Included. .Regarding selec
tion of items, Coopersmith wrote:
All the statements were reworded for use with chil
dren ages 8 to 10. Five psychologists then sorted
the. items into two groups— those indicative of high
self-esteem and those Indicative of low self-esteem.
Items that seemed repetitious or ambiguous or about
which there was disagreement were eliminated. The
set of items were then tested for comprehensibility
with a group of 30 children [1 9 6 7, p. 3 0].
86
The 50 items finally selected were designed to measure
self-attitudes in four areas: peers., parents, school, and
personal interests.
The SEI was administered to 1,748 children in the
Connecticut public schools. Then the teachers of all these
children were asked to rate each child on a 13 item, five-
point scale on behaviors presumed to be related to self
esteem. For the Behavior Rating Form (BRF), each teacher
judged the child's standing on such items as his reaction
to failure, his self-confidence in a new situation, or his
need for encouragement. The author indicated that the BRF
measures behaviors which are external manifestations of the
person's prevailing self-appraisal. The author, however,
presented no correlations between scores on the SEI and the
BRF. In fact, 85 subjects selected for intensive study
were divided into 5 groups, each consisting of 17 students,
on the basis of their different scores on the two instru
ments. The five groups were selected as follows: (l) those
earning scores at or above the 75th percentile on both
scales, described as the High-Highs, (2) those obtaining
scores on both scales within the interquartile range desig
nated as the Medium-Mediums, (3) those receiving both
scores at or below the 25th percentile, defined as the
Low-Lows, (4) those making scores above the 75th percentile
on the SEI, but below the 25th percentile on the BRF, called
87
the High-Lows, and (5) those registering self-reported
scores below the 75th percentile on the SEI but obtaining
observed behavior scores (BRF) above the 75th percentile
termed the Low-Highs. Although the author presented no
correlations between scores on the SEI and the BRF, he did
point out that extreme divergence on the two scales was
likely to occur in less than 10 percent of the cases.
In his volume, Coopersmith described an intensive
investigation of the family background of the 85 students
previously mentioned and reported on a series of experi
ments to determine the students' reactions to such factors
as level of aspiration, success and failure, pressures
toward conformity, and situations involving stress. He
did not consider the findings to be additional evidence for
the validity of his instrument. Instead he assumed the
classification into groups such as High-Highs and High-Lows
to be valid and discussed other findings in terms of how
the 85 individuals in his sample scored on the SEI and the
BRE.
Coopersmith (1 9 6 7) devoted an entire chapter to a
discussion of the theoretical foundations of self-esteem.
In Chapter II of this dissertation, Coopersmith's theory
of the self was reviewed. In brief, after discussing the
views of several theorists including James, Mead, Cooley,
Rogers, Fromm, Sullivan, Horney, Adler, and Rosenberg,
Coopersmith concluded that four major factors contribute
to self-esteem: (a) the amount of acceptance an individual
receives from significant others, (b) the person*s present
status based on past successes, (c) the way experiences
are interpreted by the individual in accord with his as
pirations, and (d) the manner in which an Individual re
sponds to devaluation. These factors appear to be basic
to Coopersmith*s definition of self-esteem. There is no
clearly defined attempt in the book, however, to explain
how these factors were incorporated into the items of the
SEI. It does seem clear, though, that Coopersmith has
adopted the position that a person can know himself as an
object and that his self-esteem is a measure of his atti
tudes about his own worth.
In this chapter the four self-regard instruments
selected for critical analysis in this research have been
objectively described. In the following chapter each of
these instruments is evaluated in the light of the validity
and reliability standards required by the Standards.
CHAPTER IV
AN EVALUATION OP FOUR SELECTED MEASURES
OP SELF-CONCEPT
Introduction
The present chapter undertakes a critical examina
tion and evaluation of the validity and reliability of the
four measures of self-concept selected for study in this
research. These tests, (l) the Piers-Harris Childrenrs
Self-Concept Scale (PHCSCS), (2) the Tennessee Self-Concept
Scale (TSCS), (3) the Thomas Self-Concept Values Test
(TSCVT), and (4) the Self-Esteem Inventory (SEl) are to be
judged In the light of the criteria provided by the Stand
ards (American Psychological Association, 196 6).
It was indicated earlier that the Standards are
divided into six major divisions. Division C— Validity—
consists of seven major standards, Cl through C7. Division
D— Reliability— includes six major standards, D1 through
D6. This same notation Is followed in this chapter in
making references to the standards or criterion In the
Standards.
The format for this chapter Is to present one at a
time each of the seven major standards for validity and
89
then one at a time each of the six major standards for re
liability exactly as they are given in the Standards.
Then evidence from pertinent literature relevant to the
validity and reliability of each of the four self-concept
measures is evaluated with respect to the standard or cri
terion under consideration. In addition the 82 corollary
standards on reliability and validity from the Standards
are also used as supplementary guides for analyzing data
that would support an evaluative judgment regarding each
major standard. The order for evaluating the four tests
on each standard is as follows: (l) PHCSCS, (2) TSCS, (3)
TSCVT, and (4) SEI.
After examining all the evidence for a particular
major standard and for a particular test, the writer en
deavored to render an unbiased judgment that indicated how
satisfactorily the test met the standard. As mentioned
previously, evaluations were classified in four categories:
"satisfactory," "unsatisfactory," "questionable," and "not
applicable.1 1 These classifications were, of necessity,
subjective estimates. These estimates, however, were
based on objective data that indicated whether or not, in
the writer’s opinion, the weight of the evidence supported
the particular criterion set forth in the Standards.
91
Evaluation of the Four Self-Concept
Tests with Respect to the
Standards of Validity
The four self-concept tests may he evaluated within
the framework of the seven following criteria or standards— :
all termed essential— (Cl through C7) pertaining to valid
ity :
Standard Cl
Cl. The manual should report the validity of the
test for each type of inference for which it is recom
mended. If its validity for some suggested interpre
tation has not been investigated that fact should be
made clear.
An evaluation of the PHCSCS with regard to validity
standard Cl. The manual of the PHCSCS presents data in a
clear and direct fashion. The authors, who are cautious in
making claims for the test, suggest that the instrument
should be used primarily for research. They recommend that
test results be interpreted only by those who are knowl
edgeable in measurement and statistics, the psychology of
adjustment, and self-theory.
The test is recommended by inference as a guide
for use in determining the self-concept of groups or in
dividuals. However, the authors have suggested certain
safeguards in making judgments about test results. They
recommend that individual changes of less than 10 points
be ignored. Because of the possible danger of faking
"good," they indicate that very high scores may be invalid.
92 :
They suggest, though, that very low scores can be assumed
to reflect truly negative self-attitudes. In the interpre
tation of group scores, the manual warns that control
groups must be used to reduce the possibility of spurious
results, since all groups typically score higher on re
testing than on Initial testing. The manual also indicates
that test results should be interpreted in the light of the
evidence that variability of scores tends to decrease as
grade level increases.
The manual suggests that the test can be used to
obtain cluster scores and that these scores can be employed
in correlational studies. Here, tooi the authors have been
cautious in their recommendations. They have said that it
would be premature to provide cluster score norms. More
over they have indicated that the proper use for the cluster
scores depends upon results from additional research.
The PHCSCS is also recommended for use in clinical
or counseling situations. In these Instances, the manual
suggests that very high scores should be evaluated In
terms of possible defensiveness, but that low scores can be
considered as a confirmation of the need for help. The use
of responses to individual items is recommended as an aid
to diagnostic Interviewing or counseling. A warning of the
possibility of extreme error when individual items are
used is not specifically given. However, the intent of
93
corollary Cl.21 from the Standards is at least partially
met by the recommendation that individual test items should
be used only in clinical or counseling settings.
Weighing the evidence presented that relates to
validity standard Cl and its corollary statements, the
writer concludes that the CSCS manual should be judged
"satisfactory" on standard Cl.
An evaluation of the TSCS with regard to validity
standard Cl. The TSCS manual recommends the use of all 29
scales and subscales. The scales of the Counseling Form are
said to be appropriate for self-interpretation and for
feedback to counselees. The manual suggests that persons
using the Clinical and Research Form for clinical assess
ments should have substantial training and supervised expe
rience in the clinical use of tests.
The manual implies that both forms of the test can
be used to discriminate between groups or to carry out
group measurement. Certainly, both forms are also recom
mended for individual counseling. Unfortunately, no warn
ing is given that some scales may not be so valid for in
dividuals as they are for groups. The author has recom
mended the TSCS for "counseling, clinical assessment and
diagnosis, research in behavioral science, personnel
selection, etc. [p. l]."
94
In the discussion of test validity, the author of
the manual has reported all suhscale scores in his review
of correlations with other tests as well as in his reports
of studies that appear to show that TSCS results can be
used to discriminate between groups of people who have
different personality characteristics. To assist the test
user in evaluating the subscale results, all raw scores are
converted to T^ scores, the values for which are visually
presented on a profile sheet.
Although the manual fails to caution the user
against overinterpretation, especially in applying subscale
results to individual cases, it presents evidence for the
validity of all 29 scales. The author has also recommended
substantial training for those who use the test for clini
cal diagnosis and assessment. The author has reported that
the test cannot be used for subjects unless they are at
least 12 years of age and unless their reading level is at
least at the sixth grade.
Considering the cautions and limitations for using
the test that the manual contains, it seems that the weight
of the evidence would limit misinterpretations of test re
sults. For this reason the TSCS is judged to be "satis
factory" with regard to validity standard Cl.
An evaluation of the TSCVT with regard to validity
standard Cl. The manual of the TSCVT recommends using the
test to compare an individual’s scores on each of the
test’s 19 scales with the same individual’s scores on
other scales. For such a comparison all raw scores are
converted to a T scale with a mean of 50 and a standard
deviation of 10 points. In addition to showing how an
individual’s scores vary among the scales, the norm aver
ages are also built into the standard T scale. This design
permits the test user to compare the scores of an individ
ual with the scores of the reference or norm group. When
scores are plotted on a T-scale graph, it would seem help
ful in interpretation of the scores that the points one
standard deviation above and one standard deivation below
the mean be specially noted. This procedure has not been
Incorporated within the TSCVT profile sheet.
The manual specifically cautions the test user that
the question to ask is, "How valid is this test for the de
cision I wish to make?" Then, the author has proceeded to
explain the basis for the validity of the test as a whole
and of the self-values In general. Although no validity
coefficients are given, the author has Indicated that he
believes that the test has both construct and concurrent
validity. The evidence for this validity Is theoretical
at present. In the opinion of the writer, since there is
insufficient mention in the manual of the possibility that
some or all of the tests scales may be invalid, caution in
96
interpretation is necessary pending further data on test
validity.
The manual does not recommend using separate items
or responses for personality assessment. Since this use
of the test has not been recommended, no caution against
the possibility of error in such interpretations of test
results may have been needed. Therefore, no violation of
corollary standard Cl.21 is evident.
Although the author of TSCVT has presented consid
erable evidence of test validity, this writer concluded
that the validity of the TSCVT for all suggested interpre
tations has not yet been sufficiently investigated. The
need for further studies is not made clear in the test
manual. For these reasons the TSCVT is judged as ’ ’ques
tionable" with regard to standard Cl.
An evaluation of the SEI with regard to validity
standard Cl. Since the SEI has no formal manual, the eval
uation of the instrument's validity and reliability is
based on data in the book by Coopersmith (1 9 6 7) and on a
memorandum he has supplied to those who request permission
to use the test in their research studies. By inference,
of course, the test is only recommended for use in research.
In judging validity standard Cl, one finds it nec
essary to determine whether the author has recommended
97
using the test to make inferences for which the validity of
the instrument is not clearly outlined.
Coopersmith's test memorandum suggests that the
test has five subscales: (l) General Self, (2) Social
Self-peers, (3) Home— parents, (4) Lie Scale, and (5)
School— academic. After noting the items in each scale,
the author has seemed to imply that these subscales may
have some valid use. However, in his book, Coopersmith has
failed to explain how these items were selected for each
scale. He has pointed out in his book, however, that his
studies revealed no significant differences in the re
sponses of students to these different areas of self-
evaluation.
In the judgment of this reviewer, it seems that the
lack of data on the validity of the test's subscales is not
clearly admitted. Therefore, the SEI is rated "question
able" with regard to validity standard Cl,
Standard C2
C2. Item-test correlations should not be pre
sented in the manual as evidence of criterion-related
validity, and they should be referred to as item-
discrimination indices, not as item-validity coeffi
cients.
An evaluation of the PHCSCS with regard to validity
standard C2. The manual of the PHCSCS makes no presenta
tion of item-test correlations. Two studies of test items
were used in developing the scale. After a pilot study
98
most items answered in one direction by fewer than 10 per
cent or by more than 90 percent were dropped from the test.
An item analysis was made to determine those items that
discriminated between high and low scoring groups at or
beyond the .05 level. Items not meeting this criterion
were dropped.
Since the manual makes no claim that item-test cor
relations are available or that they are evidence of
criterion-related validity, its contents are judged to be
"satisfactory" with regard to validity standard C2.
An evaluation of the TSCS with regard to validity
standard 02. The six scales of the TSCS (referred to in
the manual as Empirical Scales) were derived by item analy
sis. The test was given to such groups as psychotics,
neurotics, and normals. Those items that distinguished one
group from all other groups were then used to make up the
scale for that group. The manual shows the correlations of
these scales with the total test. The manual also pre
sents other part-whole correlations, but no item-test cor
relations.
The author of the manual has not referred to any of
the part-whole correlations as evidence of criterion-
related validity. Since item-test correlations are not
available, there is no violation of standard C2. The TSCS
99
is, therefore, judged to be "satisfactory" with regard to
this standard.
An evaluation of the TSCVT with regard to validity
standard C2. The manual of the TSCVT presents an extensive
item analysis. Each of the 14 items in the four self-
referent scales was correlated with the total score of the
corresponding self-referent scale. Similarly, each of the
four items in the 14 value scales was correlated with the
total score for its respective value scale. For purposes
of correlating item responses and scale responses, the
number of responses agreeing and disagreeing with each
item was tabulated. Then, a tetrachoric estimate of the
correlation between item response and scale response was
calculated. These correlations ranged from .10 to .93
with most relationships above .50.
The author of the manual in no way has inferred
that these data on item analysis constitute evidence of
criterion-related validity. He has correctly mentioned in
two separate places that this information is a measure of
the internal consistency of the test.
The manual of the TSCVT is judged to be "satisfac
tory" with regard to standard C2.
An evaluation of the SEI with regard to validity
standard C2. The short form of the SEI is composed of
those 25 items which showed the highest item-total rela-
100
tionships with scores obtained on the long form.
In the memorandum on the use of the SEI, a correla
tion of .86 between the two forms of the test is cited. It
is suggested that this correlation makes the validity data
from Coopersmith's book equally applicable to both forms
of the instrument. In no way, however, does the memorandum
or book imply that item-test correlations are evidence of
criterion-related validity. Item-test correlations are not
referred to as item-validity coefficients.
Evidence from Coopersmith's book and from his test
memorandum shows that he has not improperly used item-test
correlations as evidence of test validity. Therefore, the
SEI is rated "satisfactory" with regard to standard C2.
Standard C3
C3. If a test performance is to be interpreted
as a sample of performance or a definition of per
formance in some universe of situations, the manual
should indicate clearly what universe is represented
and how adequate is the sampling.
An evaluation of the PHCSCS with regard to validity
standard C3. The manual indicates that the universe to be
measured consisted of 11 areas established by Jersild
(1 9 5 2) from children's statements telling what they liked
and disliked about themselves. Some of the areas were (a)
Physical Characteristics and Appearance, (b) Clothing and
Grooming, and (c) Health and Physical Soundness. The
manual states,
101
Items were written to cover all these areas but
during the Item analyses non-discriminating items
were dropped so that the final scale no longer covers
every area to the same degree [p. 5].
This statement leaves the reader with no knowledge of how
adequately the test samples each of the 11 areas.
Corollary standards C3.1 and C3.ll call for infor
mation not clearly presented In the manual. The manner in
which items were selected and the qualifications of the
experts who selected the items are not given. Further, the
manual says simply that the "statements were classified by
three judges as reflecting adequate (high) or Inadequate
(low) self-concept [p. 3]." It does not report the extent
of agreement or the specific directions under which they
made their judgments.
In regard to validity standard C3, the manual ne
glects to report sufficient information for judging the
adequacy of the test In sampling the defined content uni
verse. In the writer*s opinion the weight of the evidence
requires that the manual be judged "unsatisfactory" In
regard to standard C3.
An evaluation of the TSCS with regard to validity
standard C3. The content validity of the TSCS was care
fully selected. A large pool of self-descriptive state
ments was derived from a number of other self-concept
measures including those developed by Balester (1 9 5 6),
Engle (1956), and. Taylor (1953). Other items were derived
from written statements by patients and non-patients. The
items were submitted to seven clinical psychologists who
classified each item as to its fit with defined constructs.
They also judged each item as to whether it was positive or
negative in content. The final 90 items included in the
test are those on which the judges showed perfect agree
ment. Ten additional items composing the Self-Criticism
Score were taken from the Lie-Scale of the Minnesota Multi-
phasic Personality Inventory (MMPl).
The manual of the TSCS complies with corollary
standard C3.1 by describing the professional qualifications
of the experts and the directions under which they worked
in making their judgments regarding item selection. The
request in corollary standard C3.ll that the extent of
agreement among judges be specified is also satisfied.
The author of the TSCS manual has made it perfectly
clear that some of the part scales of the test are based on
empirical findings. In fact., six of the scales in the test
are called Empirical Scales. The author has distinguished,
as suggested in corollary standard C3.3* the content of the
test that is based on logical analysis from the content of
the scales based on empirical findings.
Practices followed in the TSCS manual, which are
consistent with procedures recommended in standard C3 and
103
with its corollary standards, are therefore judged to be
"satisfactory."
An evaluation of the TSCVT with regard to validity
standard C3. The TSCVT purports to measure values. A
value has been defined by the author of the test as, "A
normative, conceptual standard of the desirable that pre-
disposltionally influences individuals in choosing among
personally perceived alternatives of behavior [p. 33]."
In his discussion of the 14 values that make up the test,
the author has quoted the results of selected research and
has cited some professional opinions that bear on the
values under consideration. Supposedly, the 14 values in
cluded in the test are a sampling from all values. The
adequacy of the sample is not made clear. In fact, the
author has stated that the 14 self-value factors selected
for the Instrument were included as a result of "specula
tive decision."
Corollary standards C3.1 and C3.2 Imply that, in
the absence of factor analytic studies or other statistical
data, items for a test might properly be selected by a
panel of judges or experts in the field. It would have
been desirable for the author to have submitted the value
Items he selected to other experts in the field in order to
obtain a further analysis of their relevance for the test.
In the judgment of the writer the content validity
104
of the TSCVT Is not thoroughly explained. Further, the
items chosen for the test may be an inadequate sample of
the universe of value items from which they were selected.
Therefore, the TSCVT is judged to be "questionable" with
regard to standard C3.
An evaluation of the SEI with regard to validity
standard C3. Coopersmith’s (1 9 6 7) book and his memorandum
both indicate that the test was designed to measure self
esteem in four areas: peers, parents, school, and personal
interests. Yet, in neither source is there evidence pre
sented that shows how items were selected for these areas.
The adequacy of the selected items for sampling self
attitudes in the four areas is not treated.
In discussing the selection of items for the SEI,
Coopersmith has stated that most of them were based upon
the Rogers and Dymond (1954) scale. These items were sup
plemented with several of Coopersmith1s original items.
All items were sorted into two groups— those indicative of
high esteem and those indicative of low esteem— by five
psychologists. Corollary standard C3.1 asks that the pro
fessional experience of experts involved in item selection
be described and that the directions under which they made
their judgments be explained. The data given by the test
author at least partially met the criterion set forth in
this corollary standard.
105
Since Coopersmith. has reported that items about
which there was disagreement were eliminated, it would ap
pear that the five psychologists employed as experts in
item selection reached perfect agreement on all items fi
nally included in the SEI. This outcome completely satis
fies corollary standard C3.ll* which asks that the extent
of agreement among judges be reported.
The evidence presented on the content validity of
the SEI is substantial; it is not, however, complete.
The test author has failed especially to explain the basis
for the selection of items included in the four subscales
of the test that measure different areas of self-esteem.
It seems logical, therefore, to rate the SEI as "question-
able" with regard to validity standard 03.
Standard C4
C4. All measures of criteria should be described
completely and accurately. The manual should comment
on the adequacy of the criterion. Whenever feasible,
it should draw attention to significant aspects of
performance that the criterion measure does not re
flect and to irrelevant factors that are likely to
affect it.
An evaluation of the PHCSCS with regard to validity
standard C4. A criterion-related variable is an external
variable different from the test. For a self-concept test,
a criterion variable would be a separate variable that
could be used as a direct measure of self-concept. Since
no generally recognized direct measures of self-concept are
106
available, measures that by theory or practice correlate
highly with self-concept are often used as evidence of
convergent validity. The PHCSCS classifies students into
two groups: low self-concept and high self-concept. Stu
dents could also be classified into the six factors re
vealed by a factor analysis study.
The manual makes no attempt to suggest a specific
degree of accuracy for the test as an instrument for pre
dicting performance on the six factors: (l) Behavior,
(2) Intellectual and School Status, (3) Physical Appear
ance and Attributes, (4) Anxiety, (5 ) Popularity, and (6)
Happiness and Satisfaction. The manual does report that
Eastman (1 9 6 5) found substantial correlations (.4-3 and
.50) between cluster scores for Factor 2. of the Scale
[Intellectual and School Status] and the WISC Full Scale
and Verbal Scale. The manual makes no suggestion, however,
that these IQ scores can be predicted from a child's per
formance on Factor 2.
The manual reports correlations ranging from -.54
to - . 6 9 between the PHCSCS and scores on the Children's
Manifest Anxiety Scale. It is suggested that this corre
lation may be spuriously high because both instruments
Involve self-reports and because one of the factors on the
PHCSCS, which is labeled Anxiety, and the Self-Concept
Scale contains items similar to those In the Anxiety Scale.
107
Although the manual suggests no specific predic
tions of performance on other variables that can be based
on PHCSCS test scores, there are implications that correla
tions between PHCSCS scores and other traits might be ex
pected in certain instances. In presenting their data,
the test authors have not complied with corollary standard
C4.4 and have not reported the elapsed time between the
administration of the PHCSCS and the administration of the
other measures with which it was correlated. However, they
have given references to original studies so that these
data could be obtained by an interested investigator.
The data on criterion-related validity are not
easy to extract from the test manual. It seems that the
test authors may have felt no need to provide statistics
on criterion-related validity and instead may have classi
fied all correlational studies as evidence of construct
validity.
Although the manual contains little information on
criterion-related validity, the authors have made no un
justified claims for their instrument. Therefore, the
manual is Judged to be "satisfactory" with regard to
standard C4.
An evaluation of the TSCS with regard to validity
standard C4. The relationships that exist between scores
on the TSCS and other external variables are discussed at
108
some length in the manual. In discussing the test norms
and the advisability of applying them to samples other than
the norm group, the author has stated that in his opinion
additional expansion of the original standardization group
is unnecessary. He has reported that demographic variables
such as sex, age, race, education, and intelligence have a
negligible effect on test scores.
In the discussion of the previously mentioned Em
pirical Scales which classify test takers into such groups
as Psychotic, Neurotic, and Personality Disorder, there is
no information in the manual on the qualifications of the
judges who selected the original norm groups. Likewise
there is little or no Information on the definitions used
by the judges in making their classifications. These omis
sions violate both corollary standard C4.6, which calls for
diagnostic terms and categories to be specifically defined
and clearly described and corollary standard C4.61, which
recommends that the training, experience, and professional
status of psychiatrists or psychologists as well as the
nature and extent of their contacts with patients be re
ported.
Since all the standardization data on which the
test scores are based are not clear, the interpretation of
the test is also not clear. For example, it is Impossible
to know exactly what it means when a person receives a
109
high score on the Neurotic Scale. In the writer’s opinion
the weight of the evidence requires an "unsatisfactory"
rating for the TSCS with regard to standard C4 and its
corollary standards.
An evaluation of the TSCVT with regard to validity
standard C4. The author of the TSCVT manual has stated:
"It was impossible to select another instrument to provide
a reliable and valid criterion index of personal self
values [p. 35] . ' 1 In his discussion of the results of two
studies, the author has indicated that the concurrent va
lidity of his test has been demonstrated. These studies
involved external variables that theoretically were thought
to be related to the variable measured by the test. Al
though this type of relationship tends to be more often
associated with construct validity than with criterion-
related validity, the author of the test manual has con
tended that the relationships existing in the studies he
reviewed might serve as a substitute for the typical
criterion-related validity studies.
Relative to a study listed under concurrent valid
ity, the manual describes TSCVT results on two groups of
50 children. One group was from highly privileged homes
and the other from underprivileged homes. On 14 of the 19
test scales the highly privileged group did perceive them
selves more positively than did the underprivileged group.
110
In a separate study, relationships between self-concept and
(a) the number of siblings, (b) the father's education,
(c) age, and (d) intelligence were investigated. Only 16
of 58 relationships investigated were found to be signifi
cant. The author concluded that the test apparently could
discriminate among children with selected concurrent char
acteristics. In the view of the writer, this statement in
its context is open to possible misinterpretation. The
caution that the test could discriminate among groups of
children with certain characteristics, but that it did not
discriminate among individual children, would seem to be
Inappropriately omitted. Likewise, a warning that the
results of these studies might not apply to other different
populations should also be Included.
After weighing the evidence presented, the writer
concludes that the manual of the TSCVT should be judged
"questionable1 ' with regard to standard C4.
An evaluation of the SEI with regard to validity
standard C4. Coopersmith has not presented relationships
between the SEI and other variables or tests as evidence
of the validity of his instrument. Instead, he has assumed
his test of self-esteem to be valid. He has discussed his
findings by making such conclusions as the following:
Self-esteem is not related to height and physical
attractiveness, two widely respected attributes in
middle-class American society, and it Is only rather
Ill
weakly related to social status and academic perform
ance [Coopersmith, 19&7j P* 242].
In his research Coopersmith did use a Behavior
Rating Form (BRF) as well as his SEI to measure student
self-esteem. The BRF is a 13-item, five-point scale of
behaviors presumed to be related to self-esteem which was
completed by teachers. Although not stating a correlation
coefficient between these two Instruments the author has
said that a marked discrepancy between scores on the two
measures is relatively rare.
Neither Coopersmith's book nor his test memorandum
give any other data that relate scores on the SEI to scores
on other measures of self-esteem. Not only are data on re
search findings with other instruments that might be used
as criterion measures largely omitted, but also in the
judgment of this writer there is the implication that the
test author has doubted the need for such validating data.
It is the judgment of this reviewer, therefore, that the
SEI should be rated as "unsatisfactory" with regard to
validity standard C4.
Standard C5
C5. The sample employed in a validity study and
the conditions under which testing is done should be
consistent with the recommendations made in the manual.
They should be described sufficiently for the user to
judge whether the reported validity Is pertinent to
his situation.
112
An evaluation of the PHCSCS with regard to validity
standard C5. The sparsity of criterion-related data in the
PHCSCS manual was pointed out earlier in this chapter in
the discussion of standard C4. The correlational studies
that are presented would suggest that the trait of anxiety
would have a negative correlation with PHCSCS scores and
that the trait of intelligence could be predicted to have a
positive correlation with scores on the PHCSCS.
Since the manual gives only partial data on age,
sex, socioeconomic status, and level of education for some
of the groups, it thus fails to meet corollary standard
C5.2. References to original sources, however, make these
data available to the serious investigator.
Since the authors have made no recommendations for
the use of their instrument as a predictive device, and
since the data on the nature of the sample are available,
the manual is judged to be "satisfactory" with regard to
standard C5.
An evaluation of the TSCS with regard to validity
standard C5. The author of the TSCS has reported correla
tions between the various scales of his test and two other
well known measures of personality adjustment. The manual
gives a separate table of correlations between the TSCS
and the MMPI and between the TSCS and the Edwards Personal
Preference Schedule (EPPS). In presenting these data the
113
manual describes the sample taking the MMPI as a group of
102 psychiatric patients. The sample used in the EPPS cor
relation study is defined simply as 66 students from three
different high schools. Such a description of a sample
would ordinarily be considered inadequate. However, the
manual does supply references to the original studies
(McGee, 1960j Sundby, 1 9 6 2) where additional information
on the nature of the samples is available.
The basic sample statistics that are essential to
report in a test manual according to corollary standard
C5.1 are measures of central tendency and variability.
Although these statistics are not cited, they are available
in the original studies.
The demographic data on the MMPI and the EPPS sam
ples are omitted from the manual. The need for such data
is set forth in corollary standard C5.2. Although this
information might be obtained from the original studies,
the lack of such data makes it difficult for the test user
to evaluate the statistics presented.
Although several data are available in other refer
ences, considerable information on two important comparison
samples is omitted from the TSCS test manual. Other data
on the samples which were used to establish some of the
previously mentioned Empirical Scales are also inadequately
presented in the manual. This inadequacy was explained
Il4
earlier in the discussion of the TSCS with reference to
standard C4. In view of these shortcomings, it is the
judgment of the writer that the TSCS should be rated as
"unsatisfactory" with regard to validity standard C5.
An evaluation of the TSCVT with regard to validity
standard C5. The two studies reported as evidence of the
concurrent validity of the TSCVT were discussed in the
evaluation of standard C4. The descriptions of the
samples used in these two studies now need to be considered
more carefully.
In the study of the scores of privileged and under
privileged children on the TSCVT, the manual supplies con
siderable information on the nature of the samples. The
50 highly privileged children were from homes where both
the father and mother were college graduates, the father
was a professional or a high level executive, the family
income was in excess of $20,000, and the home was valued
above $25,000. The underprivileged children were from
homes where neither parent finished high school, the
father was employed (if employed and if living at home)
as an unskilled or semi-skilled worker, and the home
valuation was under $10,000. This information would seem
sufficient for the test user who wished to know how his
population compared with the criterion group.
115
In a second study described in the manual as pro
viding evidence of concurrent validity* the data on the
nature of the sample are incomplete. The age* sex* and
social status of the sample population are not presented.
A test user would have difficulty in relating the findings
on this poorly defined sample to his own findings on a
different sample of children.
It apparently can be assumed, although It is not
clearly stated In the manual* that the samples employed and
the conditions under which the testing was done were con
sistent with recommendations made in the manual. One
sample was adequately described In the manual* the other
was not. Since the evidence is somewhat contradictory*
the writer judges the TSCVT to be "questionable" with
regard to standard C5.
An evaluation of the SSI with regard to validity
standard C5. It was indicated in the discussion of the SEI
in relation to validity standard C4 that the only criterion-
related validity study reported by Coopersmith Involved the
use of the BRF completed by teachers. Corollary standard
C5.1 requires test authors to report measures of central
tendency and variability for the validation sample. These
data are given in Coopersmith's book. Teacher ratings are
reported in the basic sample as ranging between 23 and 100.
A mean of 68.4 and a standard deviation of 15.4 were ob-
116
talned. The mean rating for boys was 65.0; the standard
deviation,1 6.2; for girls the mean was 71.3; the standard
deviation, 13.6.
The description of the validity sample falls short
of supplying the data required by corollary standard C5.2.
In his review of Coopersmith's book, Sears (1 9 6 9) aptly put
this shortcoming as follows:
Ultimately, 1748 "children" in Connecticut schools
were tested and the final 85 boys came from this popu
lation. The reporting of the nature of both the total
population and the final study group is woefully in
adequate. One cannot even determine the age range of
the 8 5 boys. Presumably they were from fifth and sixth
grades, but this is by no means certain, since the in
ventory was designed to be appropriate for children of
eight to ten years of age. When an author invites you
to go exploring in the dark, slippery, treacherous
field of naturalistic data on self-esteem, he ought at
least to provide the common demographic benchmarks for
you to cling to [p. 147]I
The author of the SEI has not attempted to validate
his instrument through the use of other measuring devices.
His description of the sample he had used was not adequate.
The weight of the evidence requires a rating of "unsatis
factory" for the SEI with regard to validity standard C5
and its corollary standards.
Standard C6
C6. Any statistical analysis of criterion-
related validity should be reported in the manual
in a form from which the reader can determine what
confidence is to be placed in judgments or predic
tions regarding the individual.
117
An evaluation of the PHCSCS with regard to validity
standard C6. The PHCSCS manual makes no claim that pre
dictions or judgments regarding an individual's performance
on other variables can be derived from the total score or
from cluster scores obtained on the PHCSCS. Since no spe
cific criterion-related validity is reported in the manual,
there is no inadequate or inappropriate statistical analy
sis.
Although.it might be maintained that the PHCSCS
manual does not contain sufficient data on criterion-
related validity, it does not seem appropriate to say that
the manual makes claims contrary to standard C6. For this
reason the judgment rendered is that the PHCSCS manual is
"satisfactory" with regard to standard C6.
An evaluation of the TSCS with regard to validity
standard C6. The TSCS manual presents Pearson product-
moment correlation coefficients to show the relationships
between the scores of the test and scores on the MMPI and
the EPPS. The correlation ratio (eta) is also used as a
generalized measure to show the strength but not the di
rection of a relationship. The manual cites no reference
as to where a further explanation of this statistic can be
found.
The correlation tables in the manual are keyed to
118
show the correlations that are significant at both the .05
and the .01 level.
Corollary standard C6.12 requires that whenever a
test is suggested for differential diagnosis* evidence be
included regarding the capability of the test to place
individuals in diagnostic groups rather than merely to
separate diagnosed abnormal cases from the normal popula
tion. The Number of Deviant Signs Score (NDS) is provided
as an index of psychological disturbance. Although this
score identifies deviant individuals with 8 0 percent ac
curacy* it does not place them in categories. The manual
presents a profile sheet showing data from Huffman’s (1964)
study. The TSCS is shown to discriminate between three
specific diagnostic categories. These three categories*
however* are not the same as any of those which are a
part of the regular diagnostic scales of the TSCS.
A table in the manual shows the upper and lower
cutoff points for each scale of the TSCS. This table
shows also the percentage of cases in each test category
that exceeds the cutoff points. The manual fails to show*
however* the proportion of patients who might be falsely
included or excluded from a particular category.
There is a great deal of statistical evidence in
the test manual that supports the author's contention that
the scale can be used for making predictions about individ
119
uals. Evidence for the degree of confidence with which
persons can be placed in diagnostic categories is lacking.
Since the evidence is inconclusive, the judgment is ren
dered that the TSCS should be rated as "questionable" with
regard to standard C6 and its corollary standards.
An evaluation of the TSCVT with regard to validity
standard C6. The main theme of standard C6 and its corol
lary standards is that the test manual should report
criterion-related validity in statistical terms that can
be easily understood and interpreted by the test user. Two
concurrent validity studies are reported in the TSCVT
manual in lieu of standard criterion-related validity
studies. These studies were described in the discussion
of standard C4.
Mean differences between two different groups of
children, highly privileged and underprivileged, are
analyzed in one study. Fourteen of 19 mean differences
are reported as significant at the .05 level. In the
Standards it is pointed out that if variance is large,
classification may be inaccurate even if means differ
significantly. This criticism may be pertinent in this
case, since only 8 of the 19 mean differences exceeded
one standard deviation. In addition all the scores by the
underprivileged group were within one standard deviation
of the standard score mean.
120
In the second study reported in the TSCVT manual
as evidence of concurrent validity, results are presented
in terms of Pearson product-moment coefficients. Scores
on the TSCVT are correlated with the number of siblings,
the father's education, age, and intelligence. This in
formation satisfies the part of corollary standard C6.ll
which calls for results to be reported by the use of
familiar correlation coefficients. The manual seems to
fail, however, to meet the intent of corollary standard
C6.ll, which asks that the manual present information on
validity in such a manner that it communicates clearly
with readers who have limited statistical knowledge. In
her review of the TSCVT J. Michael stated: "Pew students
with master's degrees— even in educational measurement—
would be able to comprehend the technical aspects of this
manual [1 9 7 2, pp. 371-37^]."
After carefully considering the analysis of the
evidence presented In support of the concurrent validity
of the test, the writer judges the TSCVT manual to be
"unsatisfactory" with regard to standard C6.
An evaluation of the SEI with regard to validity
standard C6. The relationship between scores on the SEI
and scores on the criterion instrument, the BRF completed
by teachers, is expressed only in general terms. That is,
Coopersmith (1 9 6 7) has stated: "Extreme divergence is
121
likely to occur In less than 10 percent of the cases [p.
l4].'7 This type of reporting does not meet the require
ments of standard C6. The test user is not able to deter
mine how much confidence he can place in the test for
making judgments or predictions regarding an individual.
This single study provides insufficient evidence
of criterion-related validity to support the assumption of
validity for the SEI that is made by the test author.
Likewise, the statistical data reported for this lone
criterion measure are Inadequate. The judgment of this
reviewer is that the test must be rated "unsatisfactory"
with regard to validity standard C6.
Standard C7
C7. If the author proposes to interpret the
test as a measure of a theoretical variable (ability,
trait, or attitude), the proposed interpretation
should be fully stated. The interpretation of the
theoretical construct should be distinguished from
interpretations arising under other theories.
An evaluation of the PHCSCS with regard to validity
standard C7. The PHCSCS manual identifies the construct of
self-concept which it purports to measure as representing
concerns that children have about themselves. These dbn-
cerns were developed from Jersild’s (1952) study of
statements by children that revealed what they liked and
disliked about themselves. The authors have failed to
distinguish between self-concept and other related con
122
structs such as self-confidence, self-esteem, or good per
sonality adjustment.
Corollary standard 07.11 indicates that it is very
desirable that studies bearing on the theory of the test
be summarized in the manual. The data from the test
manual, which were presented in Chapter III of this study,
lend support to the contention that the authors have
satisfied this corollary standard.
The manual also cites correlations between the
PHCSCS and other relevant tests such as the Children's
Manifest Anxiety Scale and measures of intelligence and
achievement. This information appears to satisfy corollary
standard C7.12. Although generally accepted measures of
self-concept are not yet available, the authors have re
ported a correlation between the PHCSCS and another experi
mental self-concept scale (Lipsitt, 1958)- This would
seem to satisfy corollary standard C7.13.
The manual reports the results of a study by
Millen (1 9 6 6) who used Crandall's (1 9 6 5) Social Desirabil
ity Scale for Children and found correlations with the
PHCSCS ranging from .25 to .45. These data satisfy, at
least in part, the corollary standard C7.23 that calls
for information on the extent to which test scores are
susceptible to an attempt by the examinee to present a
false or unduly favorable picture of himself.
123
Corollary standard C7.24 suggests that it is de
sirable to report any evidence of a set to acquiescence for
scales using "yes— no" responses. The authors of the manual
have reported that they balanced the number of positively
and negatively worded items in their inventory. They also
have eliminated negative words such as "don't" or "not."
Although the manual fails to distinguish clearly
the construct of self-concept from other related variables,
the weight of the evidence rests with the data in the
manual that support the criteria suggested in standard C7
and its corollary standards. Therefore, the PHCSCS is
judged to be "satisfactory" with regard to standard C7.
An evaluation of the TSCS with regard to validity
standard C7. The TSCS manual defines the nature and mean
ing of all test scores and subscores. It is necessary,
however, to consult at least three places in the manual to
interpret adequately what each score means. In addition
to studying the section on the meaning of scales, the
writer found it necessary to read the scoring instructions
as well as the section on test interpretation before what
was being measured by each score began to be reasonably
well understood. Although most scales are well defined,
some are not adequately treated. It was pointed out ear
lier that the test supplies no specific meaning for some
of the previously mentioned Empirical Scales beyond a
124
generalized title such as Neurotic or Psychotic.
In one section of the manual the author has re
viewed several studies and their implications for his the
ory of self-concept. These studies were summarized in
Chapter III of this dissertation. The recommendation of
corollary standard C7.ll., which calls for such theoreti
cally oriented data, seems to be followed rather clearly.
In presenting separate correlations between his
test and the MMPI and between his test and the EPPS, the
author has met the requirements of corollary standard
C7.12 which calls for citation of correlations with other
relevant t ests.
Corollary standard C7.13 indicates that correla
tions of the test be made with other generally accepted
measures of the same attribute. Since there is no other
"generally accepted measure of the same attribute," the
author has reported correlations with two not so widely
accepted measures of self-concept. He has reported a
correlation of .68 between total P (the TSCS’s best general
measure of self-concept) and the Izard Self-Rating Positive
Affect Scale (Wayne, 1 9 6 3). Also cited was a tetrachoric
correlation of .5 8 between total P and the dual character
istic self-concept and family relations as measured by the
Kell-Hoefline Incomplete Sentence Blank (Searles, 1 9 6 2).
Corollary standard C7.23 Indicates that evidence
125
should he presented on the extent to which scores are sus
ceptible to an attempt by the examinee to present a false
or unduly favorable picture of himself. Two scales of the
test, both the Self-Criticism or Lie Score and the Defen
sive Positive Score, are measures of the defensiveness of
the test-taker. They can be used in interpreting test
results. In a study by Jones (1 9 6 6), which was reviewed
in Chapter II of this dissertation, it was indicated, how
ever, that the TSCS results are clearly influenced by the
mental set of the examinee.
Although the TSCS is not a timed test, the Time
Score is used in calculating the Number of Deviant Signs
(NDS) Score when subjects take more than 24 minutes to
complete the test. This use of time would not seem to
affect any of the other scales of the test. The manual
does present a table of Intercorrelations of test scores
which shows how time correlates with the other scales of
the TSCS. The author has certainly explained the effect
of speed on his test as is recommended by corollary stand
ard C7.25. The author has presented sufficient data so
that the interpretation of his theory can be distinguished
fro"m interpretations arising under other theories. It is
the judgment of the writer that the weight of the evidence
justifies a "satisfactory" rating for the TSCS with regard
to validity standard C7.
126
An evaluation of the TSCVT with regard to validity
standard C7. The author of the TSCVT manual has attempted
to define clearly differences "between his scale and other
self-concept tests. He has explained that in the term,
Self-Concept Values, the word self-concept is best consi
dered as an adj ective rather than a noun. The test measures
values which are then translated into self-concept referent
scores. For the author, self-concept has more reference
to the process of self-reporting than it has to the object
of the self-report.
Corollary standard C7.12 states that the manual
should report correlations between the test and other rele
vant tests for which the interpretation is relatively
clear. The manual reports one correlation with an intelli
gence test that was described as an invalid and unreliable
instrument. Correlations between reputable achievement and
intelligence tests and the TSCVT would seem to be a desir
able addition to the manual.
The manual indicates that the author believes that
his test is not a measure of self-concept, but rather a
measure of values— of self-concept values. The author has
indicated that it was impossible for him to select another
test that would provide a reliable and valid criterion
index of personal self-values. Nevertheless, it would
seem desirable for the manual to report some studies in
127
which the results of the TSCVT are compared with the re
sults of other self-concept tests.
Corollary standard C7.23 indicates that it is very
desirable for a personality measure to present evidence on
the extent to which scores are susceptible to an attempt by
the examinee to present an unduly favorable picture of him
self. This evidence is not provided by the TSCVT manual.
The TSCVT calls for an "agree— disagree" response.
In such cases it is desirable., according to corollary
standard C7.24, for the manual to report evidence on the
degree to which the scores reflect a set to acquiesce.
This evidence, too, is not provided by the TSCVT.
After considering all data presented in the TSCVT
manual that appear to relate to standard C7 and its corol
lary standards, it is the judgment of the writer that the
manual should be rated as "questionable" in the reporting
of construct validity information.
An evaluation of the SEI with regard to validity
standard C7. Coopersmith (1 9 6 7) carefully and cogently
presented the theoretical basis for his description of
self-esteem. Since this test author's description or
definition of self-esteem was reviewed in Chapter II, it
will not be repeated. It can be reemphasized that
Coopersmith has clearly differentiated his construct—
self-esteem— from other similar constructs. For this
128
theorist and test designer., self-esteem Is a personal judg
ment of worthiness that Is expressed in the attitudes which
the individual holds toward himself. Coopersmith has indi
cated that attitudes directed toward the self may be de
fined in the same way as attitudes directed toward other
objects.
In line with corollary standard CJ.l the work of
Rosenberg (1 9 6 5) has been reviewed by Coopersmith. The
results of Rosenberg's study tend to substantiate Cooper
smith 's view of self-esteem as an attitude that can be
measured in the same way as other attitudes are measured.
That the SEI was not correlated with other meas
ures of self-esteem constitutes a lack of the type of evi
dences of construct validity that should have been pro
vided for users of the test.
Corollary standards C7 .2 3 and C7.24 request test
authors to discuss the effects of defensive responses and
of a possible set to acquiesce on personality measures.
Although the author has discussed response sets and de
fensive postures, he had not given any hard data on how
these possible sources of error have affected the validity
of his test. Coopersmith has reported that he found more
marked social desirability set in the range of medium
self-esteem than in the range of low self-esteem. He has
theorized that those with medium self-esteem are uncertain
of their worth, hut that those with low self-esteem have
already judged themselves to be unworthy. While errors
Induced by response sets are recognized as legitimate con
cerns in the field of attitude measurement by Coopersmith,
he has indicated that this source of possible error does
not invalidate the measurement of self-regard attitudes.
After reviewing all the evidence on construct
validity presented by Coopersmith, it is the judgment of
this writer that the SEI should be rated "satisfactory"
with regard to validity standard OJ and its corollary
standards.
Evaluation of the Four Self-Concept
Tests with Respect to Standards
for Reliability
The four self-concept measures may be evaluated In
terms of the following six criteria or standards— all
termed essential— (D1 through D6) pertaining to reliabil
ity:
Standard D1
Dl. The test manual should report evidence of
reliability that permits the reader to judge whether
scores are sufficiently dependable for the recom
mended uses of the test. If any of the necessary
evidence has not been collected, the absence of
such information should be noted.
An evaluation of the PHCSCS with regard to relia
bility standard Dl. The manual of the CSCS recommends
using test results for group comparisons, for noting
130
changes in the score of an individual, and for obtaining
information helpful in individual interviewing and coun
seling. For all these uses the manual recommends certain
cautions. In the case of group comparisons over time, the
manual states that a control group is essential because
the scores of a single group tend to be higher on retest
ing. The manual stresses that the standard error of meas
urement must be considered in making comparisons between
children or in comparing two scores obtained by the same
child. For use in individual counseling, the manual cau
tions users to look for defensiveness in high scores, but
to consider low scores as a confirmation of the need for
help.
The manual reports both Kuder-Richardson Formula 21
and Spearman-Brown odd-even formula reliability coeffi
cients. In addition, test-retest reliabilities are cited
and an estimate of the standard error of measurement is
also presented. In line with corollary standard D1.5, the
manual states that a difference of at least 10 points in
test scores is required for statistical significance.
The manual suggests scoring some items separately
to obtain cluster scores. Although reliability data for
the items In each cluster are not presented, the recom
mended use of cluster scores Is clearly stated as appropri
ate pnly In further research.
131
Reliability data, which are reported in the manual
of the PHCSCS, follow the criteria recommended in standard
Dl and its relevant corollary statements. Therefore, the
test is judged to be "satisfactory" with respect to stand
ard Dl.
An evaluation of the TSCS with regard to reliabil
ity standard Dl. The TSCS is recommended for use in meas
uring group differences and for analyzing differences in
the scores of an individual over a period of time or after
some experimental treatment. For example, the author has
made reference to one study (Ashcraft & Fitts, 1964) in
which 1 ,1 1 0 score changes were predicted for 54 patients
in two groups. The reliability data in the test manual
were obtained on a test-retest of 60 college students after
a two-week interval. Although the reliability coefficients
vary for different scores, most are between .8 6 and .8 9.
Such values seem to be sufficiently large to warrant con
fidence in measurement of individual differences.
Corollary standard D1.3 requires as essential that
standards for reliability apply to every score, subscore,
or combination of scores. The TSCS manual reports relia
bility for all scores. Since eight of the scales have
reliability coefficients in the .6 0*5, caution is required
in the use of these scales for analyzing differences in
the scores of a single person on two administrations of
132
the test. The author has failed to mention in the manual
this necessary precaution in the use of test results. The
author also has not met the recommendation of corollary
standard D1.5, which indicates that the manual state the
minimum difference between two scores ordinarily required
for statistical significance.
The reliability reported in the manual is based on
a sample of 60 college students. The user of the test
might like to know how reliable the test scores are for
other samples, especially for patients in psychiatric
categories. The author failed to mention the lack of this
type of reliability information and in so doing has seemed
to be Implying that there is no need for this type of data.
It is difficult to make a judgment regarding how
satisfactorily the manual has provided the reliability
data needed for the recommended uses of the test. There
is evidence on reliability for all scores. The author has
neglected, however, to caution the test user against the
overinterpretation of scale scores with relatively low
reliability coefficients, and he has also failed to mention
that reliability coefficients might vary with other sam
ples. In the opinion of the writer the TSCS should be
rated as "questionable" with regard to reliability stand
ard Dl.
133
An evaluation of the TSCVT with regard to reliabil-j
ity standard. Dl. The manual of the TSCVT presents the
results of three studies that bear on the reliability of
the test for Its recommended uses. One study reports
Pearson product-moment correlation coefficients that were
calculated on each of the 19 pairs of test scores obtained
by the test-retest procedure. These coefficients range
from .34 to .93. The coefficient for the mother referent
scale was apparently inadvertently omitted.
Two reported studies lend support to the internal
consistency of the test. One of these studies contains an
item analysis with correlations calculated on the relation
ship between each item and the total scale in which the
item appears. These correlations, which are expressed as
tetrachoric estimates., range from .10 to .93* In a second
study of internal consistency product-moment correlation
coefficients were calculated showing the relationships
among the four referent scales (self, mother, teacher,
peer). These coefficients vary from .6 0 to .8 0. Use of
a z_ transformation furnished an estimate of .7 3 for the
mean of these distributions.
Corollary standard D1.3 indicates that reliabili
ties should be reported for each score, or combination of
scores recommended for other than tentative use. The data
supplied by the author of the TSCVT manual supply some
134
measure of reliability for each of the 19 separate test
scales.
The manual does not state the minimum difference
in subscale scores required for statistical significance.
However, raw scores have been converted to T-scale scores
with a standard deviation of 10 points. This information
would seem to satisfy the intent of corollary standard
Dl.5.
The practices followed in the manual of the TSCVT
do seem to comply with procedure recommended in standard
Dl and its relevant corollary standards. Therefore, the
TSCVT is judged to be "satisfactory" with regard to stand
ard Dl.
An evaluation of the SEI with regard to reliabil
ity standard Dl. This standard requires the test maker to
provide sufficient evidence of reliability for all the
recommended uses of the test. It was pointed out earlier
that the SEI Is not published commercially. It appears
that the test-author has been recommending this Instrument
only for use in research or pilot studies. By implication,
however, the test has been recommended as a valid and re
liable measure of self-esteem.
Coopersmith (1 9 6 7) has reported only the results
of two test-retest reliability studies. These studies were
both made with the long form of the SEI. In the test
135
memorandum Coopersmith has suggested the possible use of a
short form of his test. He also has provided information
on the subscales of his inventory. There is,, unfortunate
ly, no direct evidence for the reliability either of the
short form of the test or of the subscales of the SEI.
Corollary standard D1.5 suggests the need for in
formation regarding the minimum difference between two
scores ordinarily required for statistical significance.
Coopersmith has provided no data on the error of measure
ment for the SEI. The test-user must make decisions re
garding the significance of SEI scores from the data
available on the average mean and standard deviation of
the test.
It seems to the writer that Coopersmith has sug
gested the possible use of the short form of his test and
the use of test subscales without supplying sufficient
reliability data. He also has failed to provide users
with a standard error of measurement for his instrument.
For these reasons, the SEI is Judged to be "questionable1 1
with regard to reliability standard Dl.
Standard D2
D2. In the test manual reports on reliability
or error of measurement, procedures, and samples
should be described sufficiently to permit a user
to Judge to what extent the evidence Is applicable
to the person and problems with which he is con
cerned.
136
An evaluation of the PHCSCS with regard to relia
bility standard D2. The PHCSCS manual presents Kuder-
Richardson Formula 21 reliability coefficients for boys and
girls separately at grades three, six, and ten. Spearman-
Brown coefficients are given for sixth and tenth grade
samples containing both sexes. Test-retest reliability
coefficients are furnished for samples containing both
sexes at grades 5> 6, and 1 0.
The corollary standard D2.21 indicates that it
would be desirable for the manual to include demographic
information such as distributions of subjects with respect
to age, sex, socioeconomic level, intellectual level,
employment status or history, and minority group member
ship. The only data of this type in the manual are in the
form of information concerning the sex and grade level of
the children in the samples plus indications of the states
in which these children were attending public schools.
However, references are given in the manual to original
studies where additional data might be obtained.
A table of test results with a number of samples
is given in the manual. This table shows the means and
standard deviations obtained in samples at grades 4, 5.. 6,
8, 10, and 12. These additional data should assist the
test user in making judgments about the appropriateness of
the test for application to the problems with which and
the persons with whom he is concerned.
The weight of the evidence supports the conclusion
that the CSCS manual is "satisfactory” with regard to
standard C2.
An evaluation of the TSCS with regard to reliabil
ity standard D2. The TSCS reliability sample is described
simply as 60 college students. The sample is not de
lineated in terms of age,, sex, socioeconomic status., or
minority group membership. This lack of information is in
direct conflict with the recommendation of corollary
standard D2.1.
Corollary standard D2.4 states that a test em
ployed to make discriminations within various particular
categories of persons must have the reliability and error
of measurement within each category independently investi
gated and reported in the manual. The TSCS does attempt
to make discriminations within various categories. The
test results are applied to such diverse groups as psy
chotic s, neurotics, and individuals with better than aver
age personality Integration. Yet, no separate reliability
studies with such samples are reported in the manual.
It is indicated in corollary standard D2.42 that
it is desirable for the test manual to report whether the
error of measurement varies at different score levels.
For example, the manual might report the standard error of
138
measurement at the mean, one standard deviation above the
mean, and one standard deviation below the mean. Such
information is not available in the manual. The writer
clearly recognizes that it would be a momentous undertaking
to report such data for all 29 scales of the TSCS. How
ever, this kind of knowledge would help the test user in
the interpretation of Total P which reflects the overall
level of self-esteem.
The reliability data given in the manual are based
upon a limited sample which is inadequately described.
Whether or not the reported reliability would apply to
other samples that might differ in age, Intelligence,
minority group membership, or degree of personality adjust
ment is not clear. The test user cannot judge with assur
ance the extent to which the evidence on reliability
applies to the persons with whom he is concerned. The TSCS;
is rated as "unsatisfactory1 ' with regard to standard D2.
An evaluation of the TSCVT with regard to reliabil
ity standard D2. In the evaluation of the TSCVT with re
gard to standard Dl, three reliability studies reported in
the test manual were described. It is necessary to make
a judgment on how adequately the samples in these reliabil
ity studies are described and on how suitable the samples
are for the recommended uses of the test.
139
Test-retest data were obtained from a sample of 39
headstart children from a low socioeconomic white community.
Since the test is recommended for use with children aged
three to nine* it would seem that data based on a wider age
range might be desirable.
Internal consistency reliability data came from
two studies— an Item analysis and a correlation of scores
among the four referent subscales. The sample for the
item analysis was 75 children randomly selected from 750
students enrolled in headstart, in follow-through, in
private nursery schools., or in regular kindergarten and
first-grade classes. The manual describes the sample for
the study of relationships among the four referent sub
scales as 50 children randomly selected from existing
norm groups.
The descriptions of reliability samples are some
what incomplete. Data on the age, sex, socioeconomic lev
el, Intellectual level, and minority group membership may
be implied, but they are not clearly stated. The age range
of the samples seems to be restricted to children in the
first grade or to even younger children. This circumstance
would leave open to question the reliability of the test
when used with nine-year-old children. The use of a single
set of norms for children of ages three to nine makes it
difficult for the test user to interpret test results with
equal confidence at all age levels.
l4o
The writer concludes that the samples are not suf
ficiently well described and that a test user might have
some difficulty in determining to what extent the evidence
on reliability is applicable to the persons with whom and
problems with which he is concerned. The content of the
manual of the TSCVT is, therefore., judged to be "question
able" with regard to standard D2.
An evaluation of the SEI with regard to reliability
standard D2. The descriptions of the reliability samples
used by Coopersmith are not completely adequate. The
failure of the author to provide sufficient demographic
data on his samples was treated at some length in the dis
cussion of validity standard C5. In Chapter II an earlier
work by Coopersmith (1959) was reviewed. In this research
there was a more adequate description of the samples used
to establish the reliability of the SEI. It would have
been helpful if this additional information on the nature
of the samples had been included in. Coopersmith* s (1 9 6 7. )
book.
No error of measurement for the SEI.is available.
The author has provided mean scores and standard devia
tions for both sexes. In his test memorandum the author
has given the mean of the test for young adults (ages 16
to 23). He has not supplied the test-user with a separate
standard deviation for this older group.
141
The reliability samples used by Coopersmith are
not adequately described. He has not given sufficient
data for the test-user to determine how variable the test
scores may be at different levels. In the light of these
shortcomings, the Judgment is made that the SEI Is "un
satisfactory" with regard to standard D2.
Standard D3
D3. Reports of reliability studies should
ordinarily be expressed In the test manual in
terms of variances for error components (or their
square roots) or standard errors of measurement,
or product-moment reliability coefficients.
An evaluation of the PHCSCS with regard to reliabil
ity standard D3. Although the manual of the PHCSCS does
not include analyses of variance for error components, it
does present the standard error of measurement.
Since separate forms of the test are not available,
most of the data on reliability have involved the use of
methods designed to measure the Internal consistency of
the test. In the case of test-retest reliabilities, the
manual expresses results in terms of product-moment relia
bility coefficients.
Corollary standard D3.1 states that it is essential
for the manual to make it completely clear that measures
of reliability do not demonstrate the criterion-related
validity of the test. Although the authors of the manual
have made no claims or inferences that reliability measures
142
do demonstrate criterion-related validity, they have ne
glected to report that they have failed to offer any other
data designed to support specifically the criterion-
related validity of their instrument.
The evidence for judging the test in regard to
standard D3 is both satisfactory and unsatisfactory. Hence
the writer believes that the appropriate classification of
the manual with regard to this standard should be "ques
tionable."
An evaluation of the TSCS with regard to reliabil
ity standard D3» The reliability study reported in the
TSCS manual uses no unconventional statistics. The mean,
standard deviation, and product-moment reliability coeffi
cient are given for each score yielded by the test.
Corollary standard D3.1 requires the test manual
to make it clear that measures of reliability do not demon
strate the criterion-related validity of the test. The
manual of the TSCS is organized into sections with separate
headings for reliability and validity. In neither section
has the author implied that reliability data can take the
place of content, criterion-related, or construct validity.
The manner in which reliability data are presented
in the manual of the TSCS meets the recommendations of
standard D3. Consequently, the TSCS is rated as "satis
factory" with regard to this reliability standard.
143
An evaluation of the TSCVT with regard to reliabil
ity standard D3. The TSCVT manual reports reliability from
a test-retest study and from an analysis of the four refer
ent subsoales in terms of Pearson product-moment correla
tion coefficients. This information satisfies the main
emphasis of standard D3.
In reporting the results of an item analysis, the
author of the test has used a relatively common method to
obtain an estimate of the correlation between a single test
item and the scale in which the item is used. He has em
ployed the responses of the upper and lower 27 percent of
each distribution to obtain a tetrachoric estimate of the
correlation and has explained his statistical procedure.
In keeping with corollary standard D3.1, the
author has made no claim that measures of reliability
demonstrate the criterion-related validity of the test.
The writer concludes that the TSCVT manual should
be Judged "satisfactory" with regard to reliability stand
ard D3.
An evaluation of the SEI with regard to reliability
standard D3. This reliability standard requires that re
ports of reliability studies be expressed in standard sta
tistical terms. Coopersmith (1 9 6 7) cited the results of
two test-retest reliability studies. For 30 fifth-grade
children the test-retest reliability estimate after five
144
weeks was .88. For another sample of 56 children., the
test-retest reliability estimate after a three-year inter
val was .70. Both of these estimates are appropriately
reported as product-moment reliability coefficients.
Corollary standard D3.1 requires test authors to
make it clear that measures of reliability do not demon
strate the criterion-related validity of the test. Al
though Coopersmith has not mentioned this point, in the
judgment of this reviewer he has not implied in any way
that test reliability can be used to support the criterion-
related validity of his test.
The weight of the evidence requires a rating of
"satisfactory" for the SEI with regard to standard D3.
Standard D4
D4. If two forms of a test are published, both
forms being intended for possible use with the same
subjects, the means and variances of the two forms
should be reported in the test manual along with the
coefficient of correlation between the two sets of
scores. If necessary evidence is not provided, the
test manual should warn the reader against assuming
comparability.
An evaluation of the PHCSCS with regard to relia
bility standard P4. Since only one form of the PHCSCS has
been published, and since standard D4 and Its corollary
standards cannot be usefully applied, a rating of "not
applicable" is rendered.
145
An evaluation of the TSCS with regard to reliabil
ity standard D4. The TSCS has two forms, the Counseling
Form and the Clinical and Research Form. However, since
these forms both use the same items, the only difference '
in the two forms is in the scoring procedures and in the
interpretation. These two forms are definitely not alter
nate forms. Therefore, the standard D4 is "not applicable"
to the TSCS.
An evaluation of the TSCVT with regard to reliabil
ity standard D4. Only one form of the TSCVT has been pub
lished. Since data on comparability of forms are, there
fore, unavailable, a critical analysis of the test manual
with regard to standard D4 and its corollary standards is
"not applicable."
An evaluation of the SEI with regard to reliability
standard P4. The two forms of the SEI are not comparable.
The short form is made up of those 25 items from the long
form that showed the highest item-total test relationship.
Since the two forms of the SEI are not truly alter
nate forms, standard D4 must be rated as "not applicable."
Standard D5
D5. If the test manual suggests that a score is
a measure of generalized, homogeneous trait, evidence
of internal consistency should be reported.
146
An evaluation of the PHGSGS with regard to relia
bility standard D5. The manual of the PHCSCS reports
Kuder-Richardson Formula 21 reliability coefficients for
six separate samples and Spearman-Brown coefficients for
two samples. This information satisfies corollary standard
D5.1, which states that it is essential to report estimates
of internal consistency determined by split-half methods or
by methods of the Kuder-Richardson type, if these approach
es can be properly applied.
Apparently, since the PHCSCS is not a speeded test,
warnings that speed does not produce spuriously high esti
mates of internal consistency reliability are not needed.
Such warnings are not mentioned in the manual. There ap
pears to be no real violation of corollary standard D5.2
which reads as follows:
Whenever reliability coefficients based upon internal
analysis are reported, the test manual should present
evidence that speed of work has a negligible influence
on scores.
The PHCSCS manual and scoring key provide informa
tion for utilizing data from a factor analysis of the test.
The authors have suggested that items on six different
parts of the test may be scored separately. They have
cautioned, however, that information from the scores is
only tentative and that the most appropriate use for the
scores is in further research. Corollary standard D5.3
indicates that it is desirable to report correlations
147
between separately scored parts of a test along with rele- :
vant means and standard deviations. This is a desirable
rather than an essential standard. The authors have sug
gested that further research is needed on the part scores
of the PHCSCS. Therefore, no serious violation of the
major standard D5 seems to be involved.
The judgment is rendered by the writer that with
regard to standard D5, which calls for adequate evidence
on the internal consistency of the test, the manual of the
PHCSCS should be rated as "satisfactory."
An evaluation of the TSCS with regard to reliabil
ity standard D5. The TSCS is a measure of self-concept,
which is a generalized trait. Standard D5 calls for evi
dence of internal consistency for such a measure. Unfor
tunately the TSCS manual presents no information whatsoever
on the internal consistency of the scale or on any of the
scale subscores.
Corollary standard D5.3 suggests that it is de
sirable for the manual to report correlations between the
parts of a test when the parts are scored separately. The
author of the TSCS manual has reported intercorrelations
of the scores of the scale. The major subscores correlate
highly, up to .9 1. This degree of relationship would lead
one to suspect that internal consistency coefficients
would be quite high. There is now additional evidence on
148
the Internal consistency of the scale. In a study made
since the publication of the manual, Itfunnelly (1 9 6 8) found
a reliability coefficient of .91 for the TSCS through
using the Kuder-Richardson split-halves technique.
In discussing the intercorrelations of scores the
author has called attention to the fact that since some
subscores are derived from overlapping items these correla
tions are spuriously high.
Since the manual of the TSCS reports no statistical
evidence of the internal consistency of the.scale, it must
be rated as "unsatisfactory" with regard to reliability
standard D5 .
An evaluation of the TSCVT with regard to reliabil
ity standard D5. The TSCVT is designed to measure a gen
eralized homogeneous trait. It is, therefore, appropriate
that evidence of internal consistency be reported in the
test manual.
Corollary standard D5.1 requires as essential the
use of the split-half method or a Kuder-Richardson method
for estimating the internal consistency of a test if these
approaches can be properly applied. Since the TSCVT is a
combination of many subscales that are scored separately,
and since the number of items in the subscales is small (4
or 14 items), the typical methods of calculating Internal
consistency have not been applied.
1 4-9
The manual reports Internal consistency in terms
of an item analysis and through the use of correlation
coefficients that show the relationships among the four
referent subscales (self, mother, teacher, peer).
Corollary standard D5.2, which is concerned with
speeded tests, does not apply to the TSCVT. Corollary
standards D5 .3 and D5.31* which are both classified as
desirable in the Standards (1 9 6 6), appear to apply in a
critical analysis of the TSCVT. When a test consists of
separately scored parts or sections, the correlations
between the parts should be reported. The manual of the
TSCVT reports correlations among the four referent sub
scores, but fails to report correlations between the 14
value subscores and the other parts of the test. The
manual also fails to report correlations between the sub
scales and the total test score.
The manual of the TSCVT cites considerable evi
dence of the internal consistency of the test. However,
several correlations involving the relationships of the
value scores as well as the correlations of the subscales
with the total test score have been omitted. The evidence
in the TSCVT manual seems to justify a rating of "ques
tionable" with regard to standard D5.
An evaluation of the SEI with regard to reliability
standard D5. This standard requires evidence of internal
150
consistency to be reported for any instrument that is a
measure of a generalized homogeneous trait. Internal con
sistency is especially important when items are viewed as
a sample from a relatively homogeneous universe. Unfor
tunately, Coopersmith has presented no evidence of the
Internal consistency of his inventory. In Chapter II the
findings of Campbell (1 9 6 5) were reviewed. This researcher
concluded that the question of internal consistency for the
SEI should be answered affirmatively. Information of this
type needs to be presented by the author If the test Is to
be published commercially at some later date. The lack of
information on internal consistency In both Coopersmith1s
(1 9 6 7) book and In his test memorandum leaves most users
of his Inventory without essential data on test reliabil
ity.
Corollary standard D5 . 3 suggests that it is
desirable for correlations between test parts to be given
along with relevant means and standard deviations. The
subscales reported by Coopersmith in his test memorandum
are not accompanied by this type of statistical data.
It is the Judgment of the writer that the SEI
should be rated as "unsatisfactory" with regard to reliabil
ity standard D5 .
151
Standard D6
D6. The test manual should indicate to what
extent test scores are stable, that is, how nearly
constant the scores are likely to be if a test is
repeated after time has lapsed. The manual should
also describe the effect of any such variation on
the usefulness of the test. The time interval to
be considered depends on the nature of the test and
on what interpretation of the test scores is recom
mended.
An evaluation of the PHCSCS with regard to relia
bility standard D6. The manual of the PHCSCS reports test-
retest reliability data from four different samples. The
obtained correlation coefficients range from .7 1 to .77.
Information in the manual Indicates that these samples
were public school children from Pennsylvania and Oregon.
The manual also gives information on the grade level and
sex of the sample populations. All samples were retested
with the same test after four months. One sample was also
retested after two months. The one sample with the two
and the four month retest produced the same value of .77
for the reliability coefficient. The authors of the manual
have pointed out that scores are consistently higher on
retesting than on initial testing. They also have empha
sized the importance of using a control group when making
claims about changes in self-concept. The authors have
concluded that at least by age eight, self-attitudes have
a reasonable amount of stability.
The use of separate forms for determining the
stability of scores is recommended in corollary standard
152
D6.1. This procedure was not possible, since no alternate
form of the PHCSCS is available. Corollary standard D6.2
cites as essential the listing of means and standard devia
tions at each testing as well as the correlations. The
means and standard deviations at each testing are not given
in the manual. However, references are made to the origi
nal studies where additional data are available. Corollary
standard D6.3 states that data on stability be-provided
that indicate how long test scores can be considered useful
if the scores become a part of a written record. In this
regard the authors of the PHCSCS manual have suggested that
all individual test scores be considered tentative and that
these scores should not be made a part of a child’s perman
ent record. Corollary standard D6.31 requires as essential
that the test manual describe the relevant experience or
education of samples between testings. It Is implied, but
not specifically stated in the manual, that the experience
of all groups between testings Involved regular classroom
work with no special treatment designed to Influence self-
concept.
Practices followed in the manual of the PHCSCS
appear to be generally consistent with recommended proce
dures for reporting reliability data on the stability of
the test. The PHCSCS is judged, therefore, to be "satis
factory" with regard to- standard D6.
153
An evaluation of the TSCS with regard to reliabil
ity standard P6. The manual of the TSCS reports the re
liability of scores over a two-week interval. Corollary
standard D6.1 recommends the use of alternate forms to
minimize the recall of specific answers when the time
interval between testings is not long enough to assure
forgetting. While the use of alternate forms was not
possible, the writer has suspected that college students
might well remember some specific answers over a two-
week period. This circumstance would tend to make the
test appear to be more stable than it actually is.
The corollary standard D6.2 requires that the mean
and standard deviation of scores at each testing be given
as well as the period of time that elapsed between tests.
Although the manual of the TSCS cites elapsed time, it re
ports only one set of data showing mean scores and stand
ard deviations for the scales and subscales of the test.
If test results are to be made a part of a person’s
record, corollary standard D6.3 recommends that the length
of time for which these test results might be interpreted
as relatively unchanged by the passing of time ought to be
reported. Although the manual of the TSCS does not touch
upon this point, the author has stated that he has demon
strated that the distinctive features of individual pro
files are still present for most persons a year or more
later.
154
In reporting on the stability of test scores over
time* corollary standard D6.31 requires the test manual to'
describe the relevant experience or education of the group
between testings. The TSCS manual states that the sample
group was college students. While not specifically stated,
it can be presumed that no special experiences occurred
over the two-week interval that would have signifleantly
influenced test scores.
Although there are minor flaws in the treatment of
the reliability data which relate to the comparison of test
scores over time, the TSCS manual provides the user with
considerable evidence on the stability of the test. Prac
tices followed in the manual of the TSCS appear to be gen
erally consistent with the procedures recommended in
standard D6 and its corollary standards. In the writer's
judgment, the TSCS should be rated as "satisfactory" with
regard to these recommended procedures.
An evaluation of the TSCVT with regard to reliabil
ity standard P6. The evidence of the stability of the
TSCVT scores given in /the manual is based on one study
involving the retest of 39 children nine to fourteen days
later by the same examiner. Test-retest correlation
coefficients based on this study are given for 18 of the
19 test scales. The coefficient for the mother referent
scale is omitted.
155
The author of the test manual has pointed out that
personality characteristics are dynamic and changing. He
has cautioned that too long a time interval between testings
would produce indices of instability when in fact the test
is paradoxically more constant than the characteristic
being measured. In view of the fact that the author has
believed that scores might vary considerably in a period of
time longer than two weeks., it would have been appropriate
for him to include a warning against recording test scores
in permanent school records. This advice would help to
satisfy the intent of procedures recommended in corollary
standard D6.3.
Data in the manual give the test user knowledge
of the time that elapsed between testings. It is implied
that the children had no special experiences between test
ings that might have unduly affected their scores. In
supplying information about elapsed time between testings,
the manual appears to be consistent with corollary stand
ards D6.2 and D6.31.
In general, the author has supplied adequate data
for Judging the stability of the test. The procedures
recommended in standard D6 and its corollary standards are
followed in the test manual. The judgment is rendered that
the test should be rated "satisfactory" with regard to
reliability standard D6.
156
An evaluation of the SEI with regard to reliability
standard D6. The test-retest reliability studies reported
for the SEI document the stability of the inventory over
both a five-week span (r = .88) and a three-year interval
(r = .70).
Corollary standard D6.2 requires that the mean and
standard deviation of scores as well as the correlation be
reported at each testing. Coopersmith (1 9 6 7) has cited
only one mean and one standard deviation for each of his
reliability studies.
According to corollary standard D6.3j when test
scores are likely to be retained in a person’s record,
test-makers should indicate the length of time following
the test administration during which the scores may be used
effectively for recommended purposes. Coopersmith (1 9 6 7)
has not touched upon this point directly, but he has con
cluded "that at some time preceding middle childhood the
individual arrives at a general appraisal of his worth,
which remains relatively stable and enduring over a period
of several years [p. 5]."
Corollary standard D6.31 requires information from
test-makers that describes the relevant experience or edu
cation of the group between testings. Although these ex
periences have not been mentioned by Coopersmith, it would
not be unreasonable to assume that his reliability samples
had no unusual educational experiences between testings.
157
Although the author of the SEI has not supplied
completely all of the data requested in standard D6 and
its corollary standards, he has presented considerable
evidence to support the stability of his inventory. In
the judgment of the writer, the SEI should be rated as
"satisfactory" with regard to reliability standard D6.
Summary
The present chapter has undertaken a critical
analysis and evaluation of the validity and reliability of
four selected measures of self-concept relative to criteria
set forth in the Standards. In the light of available pub
lished evidence, a judgment was made indicating to what
extent the test authors took into account the recommenda
tions in the Standards for psychological tests and manuals.
Each test was rated on each of seven validity and each of
six reliability standards. Ratings were "satisfactory,"
"unsatisfactory," "questionable," or "not applicable."
These evaluative judgments are summarized in Table 1.
In the section entitled "Conclusions" in the next
and final chapter (Chapter V), the significance and gener-
alizability of the evaluative information in relation to
the development of measures of self-concept are set forth.
Implications for test development and practice are evident
in the four recommendations that follow the eleven con
clusions .
158
TABLE 1
SUMMARY OF FINDINGS SHOWING JUDGMENTS RENDERED
REGARDING THE RELIABILITY AND VALIDITY OF
FOUR SELECTED MEASURES OF SELF-CONCEPT
Validity
Standards PHCSCS TSCS TSCVT SEI
Cl SA SA QU QU
C2 SA SA SA SA
C3
UN SA QU QU
C4 SA ■ UN QU UN
C5
SA UN QU UN
C6 SA QU UN UN
07
SA SA QU SA
Reliability
Standards
D1 SA QU SA QU
D2 SA UN QU UN
D3
QU SA SA SA
D4 NA NA NA NA
D5
SA UN QU UN
D6 SA SA SA SA
Note: SA = Satisfactory
QU = Questionable
UN = Unsatisfactory
NA = Not Applicable
PHCSCS = Piers-Harris Children's Self-Concept Scale
TSCS = Tennessee Self-Concept Scale
TSCVT = Thomas Self-Concept Values Test
SEI = Self-Esteem Inventory
»
CHAPTER V
SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS
Summary
Purpose
The purpose of this study- was to analyze and evalu
ate evidence pertaining to the reliability and validity of
four selected measures of self-concept relative to criteria
or standards set forth in the Standards for Educational and
Psychological Tests and Manuals (American Psychological
Association, 196 6). The need for this type of research was
established by outlining the present uncertainty that sur
rounds results obtained from the use of available instru
ments. Improved measures of self-concept are needed as
research tools in studies of personality and child develop
ment and as devices for the assessment of the entering
behaviors of students commencing new learning experiences.
The evaluations furnished in this study were in
tended to provide the us'brs of self-concept measures with
additional scientific and well-documented information on
the advantages and disadvantages of four available self-
concept tests. This type of information might extend and
improve the use of these particular Instruments.
159
160
Finally, a relatively objective and critical evalu
ation of certain existing measures of self-concept may
serve to encourage both test authors and test publishers
to strive for even higher standards in the development of
instruments that are designed to evaluate self-attitudes in
public schools, in clinics, and in research endeavors.
Methodology
Prior to the explication of the research problem in
terms of two major questions to be answered, it was neces
sary to describe the methodological bases for the investi
gation, as the procedural strategies inevitably set the
constraints within which the research problem could be for
mulated. Since this investigation was designed to evaluate
critically evidence pertaining to the reliability and va
lidity of four selected measures of self-concept, it re
quired the writer to make three major procedural decisions.
First, what would be a relatively objective and authorita
tive set of criteria or standards against which the se
lected tests could be Judged? Second, which measures of
self-concept should be chosen for critical analysis and
evaluation? Third, what sources of information or evidence
should be used as a basis for evaluation?
Criteria chosen for evaluation. In response to the
first question, it was decided that the Standards for Edu
cational and Psychological Tests and Manuals (American
161
Psychological Association, 1 9 6 6), referred to as the Stand
ards , would establish the evaluative criteria. This pub
lication contains seven major standards termed as "essen
tial" and 54 corollary standards for test validity in Divi
sion C. There are six major standards and 28 corollary
standards for test reliability in Division D. In this
study each test selected for review was rated on each
major validity standard (Cl to Cj) and each major reliabil
ity standard (D1 to D6). Corollary statements of recom
mended practices were taken into account in making judg
ments about each major standard.
Seven criteria for evaluating validity. The seven
major standards or criteria employed in evaluating evidence
from test manuals, books, and other professional literature
relevant to the validity of the four selected measures of
the self-concept were as follows:
Cl. The manual should report the validity of the
test for each type of inference for which it is recom
mended. If its validity for some suggested interpre
tation has not been Investigated that fact should be
made clear.
C2. Item-test correlations should not be pre
sented In the manual as evidence of criterion-related
validity, and they should be referred to as item-
discrimination Indices, not as item-validity coeffi
cients .
C3. If a test performance is to be interpreted
as a sample of performance or a definition of per
formance in some universe of situations, the manual
should Indicate clearly what universe is represented
and how adequate is the sampling.
162
C4, All measures of criteria should be described
completely and accurately. The manual should comment
on the adequacy of the criterion. Whenever feasible,
it should draw attention to significant aspects of
performance that the criterion measure does not re
flect and to irrelevant factors that are likely to
affect it.
C5. The sample employed in a validity study and
the conditions under which testing is done should be
consistent with the recommendations made in the manual.
They should be described sufficiently for the user to
judge whether the reported validity is pertinent to
his situation.
C6. Any statistical analysis of criterion-
related validity should be reported in the manual in
a form from which the reader can determine what con
fidence is to be placed in judgments or predictions
regarding the individual.
C7. If the author proposes to interpret the test
as a measure of a theoretical variable (ability,
trait, or attitude), the proposed interpretation
should be fully stated. The interpretation of the
theoretical construct should be distinguished from
interpretations arising under other theories.
Six criteria for evaluating reliability. Similarly
the six major standards or criteria employed in evaluating
information from test manuals, books, and other profes
sional literature pertaining to the reliability of the same
four self-concept measures were as follows:
Dl. The test manual should report evidence of
reliability that permits the reader to judge whether
scores are sufficiently dependable for the recommended
uses of the test. If any of the necessary evidence
has not been collected, the absence of such informa
tion should be noted.
D2. In the test manual reports on reliability or
error of measurement, procedures, and samples should
be described sufficiently to permit a user to judge
to what extent the evidence is applicable to the person
and problems with which he is concerned.
163
D3. Reports of reliability studies should ordi
narily be expressed in the test manual in terms of
variances for error components (or their square roots)
or standard errors of measurement, or product-moment
reliability coefficients,
D4. If two forms of a test are published., both
forms being intended for possible use with the same
subjects, the means and variances of the two forms
should be reported in the test manual along with the
coefficient of correlation between the two sets of
scores. If necessary evidence is not provided, the
test manual should warn the reader against assuming
comparability.
D5. If the test manual suggests that a score Is
a measure of generalized, homogeneous trait, evidence
of internal consistency should be reported.
D6. The test manual should Indicate to what
extent test scores are stable, that is, how nearly
constant the scores are likely to be if a test is
repeated after time has lapsed. The manual should
also describe the effect of any such variation on
the usefulness of the test. The time interval to be
considered depends on the nature of the test and on
what interpretation of the test scores Is recommended.
Selection of instruments. The second major deci
sion required this researcher to determine which instru
ments should be selected for careful analysis. Three of
the tests selected were the only measures of self-concept
listed in the 1972 edition of the Mental Measurements Year
book by Buros. The three tests reviewed by Buros are the
Piers-Harris Children's Self-Concept Scale (PHCSCS) (Piers
& Harris, 1 9 6 9)j the Tennessee Self-Concept Scale (TSCS)
(Pitts, 1965)j and the Thomas Self-Concept Values Test
(TSCVT) (Thomas, 1 9 6 9). The fourth test selected for
evaluation was a well known noncommercial instrument, the
164
Self-Esteem Inventory (SEI) (Coopersmith, 1 9 6 7).
Sources of information. In order to render judg
ments on how well tests met the validity and reliability
standards established by the Standards, evidence was gath
ered from a number of sources. The review of the litera
ture included information on several topics: (l) "The His
tory of Self and Ego Theories," (2) "The Development of
Measures for Self-Regard," (3 ) "Procedures for Reporting
Test Reliability and Validity," and (4) "Research Studies
Related to the Reliability and Validity of Four Selected
Measures of Self-Regard." The primary sources for infor
mation about the tests being evaluated were the test manuals
of the commercially published measures and Coopersmith1s
(1 9 6 7) book, The Antecedents of Self-Esteem.
Data from these sources on the validity and reli
ability of the instruments were presented in Chapter III.
Also in Chapter III a complete description was given of
each test selected for analysis. All the evidence gath
ered was evaluated In the light of the criteria set forth
in the Standards. Then, the writer attempted to render an
unbiased judgment to indicate how satisfactorily each test
had met each of the major standards for reliability and
validity. Such judgments were classified as "satisfactory,"
"unsatisfactory," "questionable," or "not applicable."
Research Problem
The eight questions posed in Chapter I to explicate
the research problem may be subsumed under the following
two broad questions:
1. In light of the information provided by a manual
or book to accompany a test and the evidence
available in the professional literature, how
adequately or effectively does each one of the
four tests of self-concept meet each of the
seven major criteria of validity set forth in
the Standards?
2. In terms of the information provided by a manual
or book to accompany a test and the evidence
available in the professional literature, how
adequately or effectively does each one of the
four tests of self-concept meet each of the
six major criteria of reliability set forth in
the Standards?
Findings
The results of the evaluation of the pertinent
evidence regarding the validity and reliability of the
four self-concept measures may be summarized as follows:
1. Relative to each of the seven major standards
or criteria of validity Cl, C2, C3, C4, C5> C6,
and CJj, the four self-concept measures PHCSCS,
166
TSCS, TSCVT, and SEI, respectively, were judged
as falling in each of the four evaluative cate
gories of "satisfactory" (SA), ''questionable"
(QU), "unsatisfactory" (UN), or "not applicable"
(NA) as follows: Cl— SA, SA, QU, and QUj C2--
SA, SA, SA, and SA; C3— UN, SA, QU, and QU;
04— SA, UN, QU, and UN; C5— SA, UN, QU, and UN;
C6— SA, QU, UN, and UN; and CJ— SA, SA, QU, and
SA.
2. In terms of each of the six major standards or
criteria of reliability Dl, D2, D3, 334, 335, and
336, the four self-concept measures PHCSCS, TSCS,
TSCVT, and SEI, respectively, were judged as
falling in each of the four evaluative categories
cf "satisfactory" (SA), "questionable" (QU),
"unsatisfactory" (UN), or "not applicable" (NA)
as follows: Dl— SA, QU, SA, and QU; D2— SA, UN,
QU, and UN; D3— QU, SA, SA, and SA; 334— NA, NA,
NA, and NA; D5— SA, UN, QU, and UN; and 336— SA,
SA, SA, and SA.
Conelusions
The results of. the evaluations of the four se
lected measures of self-concept were considered to support
the following conclusions:
No instrument was judged to have met all the
reliability and validity standards recommended
by the Standards. It should be noted, however,
that each test evaluated has certain strong
points. _
The PHCSCS received more satisfactory ratings
than any one of the other tests evaluated. In
the judgment of the writer this test is an ex
cellent instrument when it is used and inter
preted in the light of the cautions expressed
by the authors in the test manual.
Since all four tests were rated as satisfactory
with regard to validity standard C2, it has been
demonstrated that these test authors avoided
the pitfall of claiming that item-test correla
tions are evidence of criterion-related validity.
Since only the TSCS was judged to be satisfactory
on standard C3 concerned with content validity,
it would seem that test authors have been having
difficulty in defining the universe of test items
from which Items for a particular scale can be
selected. This difficulty might be attributed
to inadequate definitions of self-concept and
to inadequate theories that do not permit the
test authors to develop a large pool of Items
168 :
from which they can sample "behaviors or attitudes
indicative of self-esteem.
5. Relative to the three validity standards, C4, G5,
and C6 that are concerned with criterion-related
validity, three of the four tests evaluated re
ceived no satisfactory ratings on these standards.
It would appear that the authors of these three
tests had considerable difficulty in finding an
external variable that might serve as a direct
measure of self-concept. This difficulty points
up the need for the improved measures of self-
concept .
6. Since three of the four tests evaluated were
rated "satisfactory" with regard to standard C7
concerned with construct validity, it would seem
that these test authors were careful to distin
guish their concept of the self from that in
other similar theories.
7. Since only one of the four tests received a
satisfactory rating on standard D2, which sug
gests a careful description of the samples and
procedures used in establishing test reliability,
and since this information is generally avail
able, it should be relatively easy for test
authors to meet this standard.
8. Since three of the four tests evaluated received
a satisfactory rating on reliability standard D3,
which is concerned with appropriate statistical
terms for reporting the results of reliability
studies, it would appear that the test authors
were capable of using proper terms when they
reported information about test reliability.
9. Since not one of the four tests evaluated had
a truly comparable or alternate form, as evi
denced by a rating of nnot applicable" on
standard D4, it is extremely difficult to develop
even one good measure of self-concept.
10. Reliability standard D5, which calls for evidence
of internal consistency for measures of a general
ized homogeneous trait, was met in a satisfactory
manner by only one of the four tests. Hence it is
clear that the failure to include such readily ob
tainable data is an unacceptable practice on the
part of designers of tests purported to measure
self-coneept.
11. Since all four tests were rated "satisfactory"
on standard D6 involving the demonstration of
relative constancy of measures over the passage
of time, it is evident that measures intended
to reflect the self-concept itself are relatively
stable.
170
Recommendat Ions
In light of the evaluations of the evidence for the
four self-concept measures, the following recommendations
were formulated:
1. Since evaluative information for standards C4,
C5, and C6 tended not to be satisfactory for
three of the four self-concept tests investi
gated, it would appear that greater attention
should be directed in validity studies toward
complete and accurate descriptions of perform
ance criteria, an explanation of sample selection
procedures and conditions of testing, and compre
hensible reporting of statistical procedures and
levels of confidence in judgments or predictions
made for individuals.
2. The lack of applicability of standard D4 dealing
with alternate form reliability because of the
absence of two forms for any one of the four
self-concept tests would suggest the probable
need for the construction of alternate forms
of these tests and for presentation of statis
tical data attesting to their comparability.
3. The absence of satisfactory evidence for standard
D5 regarding internal consistency estimates of
f- '
reliability for three of the four tests would
seem to suggest the need for encouraging test
in
makers to develop homogeneous univocal measures
of a self-concept construct.
4. Finally* as evidenced by a lack of satisfactory
judgments on all essential standards pertaining
to reliability and validity for the four tests
studied* research workers and test designers
should continue to strive for the development
of even more reliable and more valid measures
of self-concept than those now available.
These measures* when designed* can help research
psychologists to isolate with improved accuracy
the variables that are related to the self-
concept. With this knowledge it will be possible
for those who work with children to obtain more
nearly accurate measures of their self-esteem
and to have means for evaluating change in
degree of self-esteem as a function of exposure
to a variety of experiences.
R E FER EN CE S
172
REFERENCES
Adams, G. S. Measurement and evaluation in education, psy
cho logy and guidance. New York: Holt, Rinehart &
Winston^ 1965.
Adler, A. The practice and theory of individual psychol
ogy. New York: Harcourt, 1927.
Allport, G. W. The ego in contemporary psychology. Psy
chological Review, 1943; 50; 451-478.
Allport, G. W. Becoming: Basic considerations for a psy
chology of personality. New Haven: Yale University
Press, 1955.
American Psychological Association, Committee on Test
Standards. J. W. French & W. B. Michael, cochairmen.
Standards for educational and psychological tests and
manuals" Washington, D.C.: American Psychological
Association, 1966.
Angyal, A. Foundations for a science of personality.
New York: Commonwealth Fund, 1941.
Ashby, M. L. The effect of self-concept on children's
learning in religious education. Unpublished doctoral
dissertation, George Peabody College for Teachers,
1968.
Ashcraft, C., & Fitts, W. H. Self-concept change in psy
chotherapy. Psychotherapy, 1964, 1, 115-118.
Atchison, C. A comparative study of the self-concept of
behavior problem and non-behavior problem high school
boys. Unpublished doctoral dissertation, Indiana
University, 1958.
Ausubel, D. P., Balthazar, E. E., Rosenthal, I., Blackman,
L. S., Schpoont, S. H., & Welkowitz, J. Perceived
parent attitudes as determinants of children's ego
structure. Child Development, 1954, 25, 173-183.
173
174
Balester, R. J. The self-concept and juvenile delinquency.
Unpublished doctoral dissertation, Vanderbilt Univer
sity, 1956.
Bertocci, P. A. The psychological self, the ego and per
sonality. Psychological Review, 1945j 52, 91-99*
Bloom, B. S., Engelhart, M. D., Furst, E. J., Hill, W. H.,
& Krathwohl, D. R. Taxonomy of educational objectives 1
The cognitive domain! Handbook I. New York: Longmans,
1956.
Bolea, A. S., Felker, D. W., 8s Barnes, M. D. A pictorial
self-concept scale for children K-4. Journal of Educa
tional Measurement, Fall 1971, 8, 223-224.
Boston, G. N., 8s Kew, K. L. The self-concept of the un
married mother in the Florence Crlttenton Home, Nash
ville, Tennessee. Unpublished master's thesis, Univer
sity of Tennessee, 1964.
Brookover, W. B., Thomas, S., 8s Patterson, A. Self-concept
of ability and school achievement. Sociology of Educa
tion, 19^4, 37, 271-278.
Brownfain, J. J. Stability of the self-concept as a di
mension of personality. Journal of Abnormal and Social
Psychology, 1 9 5 2, 47, 597-606.
Bruck, M., & Bodwin, R. F. The relationship between self-
concept and the presence and absence of scholastic
underachievement. Journal of Clinical Psychology,
1 9 6 2, 18, 1 8 1-1 8 2.
Buros, 0. K. (Ed.) The seventh mental measurements year
book. Highland Park, N.J.: Gryphon Press, 1972.
2 vols.
Butler, J. M., 8s Haigh, G. V. Changes In the relation
between self-concepts and ideal concepts consequent
upon client-centered counseling. In C. R. Rogers 8s
R. Dymond (Eds.), Psychotherapy and personality change.
Chicago: University of Chicago Press, 19547 Pp. 55-75.
Calvin, A. D., 8s Holtzman, W« H. Adjustment and discrep
ancy between self-concept and inferred self. Journal
of Consulting Psychology, 1953, 17.* 39-44.
Campbell, D. T., 8s FIske, D. W. Convergent and discrimi
nant validation by the multitrait-multimethod matrix.
Psychological Bulletin, 1959, 56, 81-105.
175
Campbell, P. B. Self-concept and academic achievement in
middle grade public school children. Unpublished doc
toral dissertation, Wayne State University, 19^5•
Cattell, R. B. The scientific analysis of personality.
Chicago: Aldine Publishing Company, 1955.
Chein, I. The awareness of self and the structure of the
ego. Psychological Review, 1944, 51, 304-314.
Combs, A. W. (Chm.) Self-concept: Symposium presented at
the meeting of the American Educational Research Asso
ciation, Los Angeles, February 1 9 6 9.
Combs, A. W., & Snygg, D. Individual behavior. (Rev. ed.)
New York: Harper & Row, 1959*
Combs, A. W., & Soper, D. W. The self, its derivative
terms and research. Journal of Individual Psychology,
1957, 13, 134-145.
Combs, A. W., Soper, D. W., & Courson, C. C. The measure
ment of self-concept and self-report. Educational and
Psychological Measurement, 1 9 6 3, 23, 493-500.
Cooley, C. H. Human nature and the social order. New
York: Charles Scribners Sons, 1922.
Coopersmith, S. A method for determining types of self
esteem. Journal of Educational Psychology, 1959, 59,
87-94.
Coopersmith, S. The antecedents of self-esteem. San
Francisco: W. H. Freeman, 1957-
Coopersmith, S. Studies in self-esteem. Scientific Ameri
can, 1968, 2 1 8(2), 96-106.
Courson, C. C. The use of inference as a research tool.
Educational and Psychological Measurement, 1965, 25,
1029-1038.
Cox, S. H. Family background effects on personality de
velopment and social acceptance. Unpublished doctoral
dissertation, Texas Christian University, 1 9 6 6.
Crandall, V. C., Crandall, V. J., & Katkovsky, W.
A children’s social desirability questionnaire.
Journal of Consulting Psychology, 1 9 6 5, 29, 27-36.
176
Cronbach, L. J. Further evidence on response sets and test
design. Educational and Psychological Measurement,
1950, 10, 3-31.
Cronbach, L, J. Essentials of psychological testing. (2nd
ed.) New York: Harper & Row, i9 6 0.
Cronbach, L. J., & Meehl, P. E. Construct validity in psy
chological tests. Psychological Bulletin, 1955j 52,
281-302.
Diggory, J. C. Self-evaluation: Concepts and studies.
' * New York: Wiley, 198b.
Dyer, C. 0. Construct validity of self-concept by a
multitrait-multimethod analysis. Unpublished doctoral
dissertation, University of Michigan, 1 9 6 3.
Eastman, E. The relationship between self-concept and
intelligence in children. Unpublished manuscript,
Whitworth College, Spokane, Washington, 1965.
Edwards, A. L. The social desirability variable in person
ality assessment and research. New York: Holt, 1957.
Engel, M. The stability of the self-concept in adolescence.
Unpublished doctoral dissertation, George Peabody Col
lege, 1956.
Engel, M. The stability of the self-concept in adolescents.
Journal of Abnormal Social Psychology, 1958, 5 8, 211-
215.
English, H. B., & English, A. C. A comprehensive diction
ary of psychological and psychoanalytical terms.
New York: David McKay Company, 1958.
Festinger, L., & Carlsmith, J. M. Cognitive consequences
of forced compliance. Journal of Abnormal Social Psy
chology, 1959, 58, 203-210.
Fey, W. F. Acceptance by others and its relation to ac
ceptance of self and others. A reevaluation. Journal
of Abnormal Social Psychology, 1955^ 50j 274-276*1
Fink, M. B. Self-concept as it relates to academic under
achievement. California Journal of Educational Re
search, 1 9 6 2, 13# 57-62.
177
Jackson, D. N., & Messick, S. Content and style in per
sonality assessment. Psychological Bulletin, 1958* 55*
243-252.
James, W. Principles of psychology. New York: Holt,
1 8 9 0. 2 vols.
Jersild, A. T. Self-understanding in childhood and
adolescence. American Psychologist, 1951* 6, 122-
1 2 6.
Jersild, A. T. In search of self. New York: Teachers
College, Columbia University, Bureau of Publications,
1952.
Johnson, 0. G., & Bommarito, J. W. Tests and measurements
in child development: A handbook. San Francisco:
Jossey-Bass, Inc., 19734
Jones, M. H. The Influence of situational variables on
truthfulness of responses to a self-concept scale.
Paper presented at the meeting of the Inter-American
Society of Psychology, Lima, Peru, April 1 9 6 6.
Kenny, D. T. The influence of social desirability measures
between real self and ideal self. Journal of Consult
ing Psychology, 1956, 20, 315-318.
Kerlinger, F. N. Foundat ions of behavi oral res earch.
New York: Holt, Rinehart & WinstOn, 1 9 6 6.
Koffka, K. Principles of gestalt psychology. New York:
Hare ourt, 1935.
LaBenne, W. D., & Greene, B. I. Educational implications
of self-concept theory. Pacific Palisades, Calif.:
Goodyear Publishing Company, 1969*
Lecky, P. Self-consistency: A theory of personality.
New York: Island Press, 1945.
Lefeber, J. The delinquent’s self-perception. Unpublished
doctoral dissertation, University of Southern Calif
ornia, 1964.
178
Pitts, W. H. Manual for Tennessee self-concept scale.
Nashville: Counselor Recordings and Tests, 1965._
Pitts, W. H., Adams, J. L., Radford, G., Richard, W. S.,
Thomas, B. K., Thomas, M. M., & Thompson, W. The self-
concept and self-actualization. Dede Wallace Center
Monograph III. Nashville: Counselor Recordings and
Tests, 1971*
Preud, S. An outline of psychoanalysis. New York: Norton,
1 9 4 9. (First German edition, 1940).
George, F. H. The relationship of the self-concept, ideal
self-concept, values, and parental self-concept to the
vocational aspiration of adolescent Negro males. Un
published doctoral dissertation, North Texas State
University, 1 9 6 9.
Gwiden, G. M. Stress in airborne training as related to
the self-concept, motivation and biographical factors.
Unpublished master's thesis, Vanderbilt University,
1959.
Gordon, C., & Gergen, K. J. (Eds.) The self and social
interaction. Vol. 1. New York:’ Wiley, 196b.
Gordon, I. J., & Combs, A. W. The learner: Self and per
ception. Review of Educational Research, 1958, 2 8,
433-444.
Guilford, J. P. Fundamental statistics in psychology and
education. (4th ed.) New York: McGraw-Hill, 19o5.
Hall, C. S., & Lindzey, G. Theories of personality. New
York-i Wiley, 1957*
Hllgard, E. R. Human motives and the concept of the self.
American Psychologist, 1949> 4, 37^-382.
Huffman, T. B. Relationships between the self-concept and
three psychopathological types. Unpublished master's
thesis, University of Tennessee, 1964.
Hugo, M. J, The effects of group counseling on self-
concepts and behavior of elementary school children.
Unpublished doctoral dissertation, Ohio University,
1969.
179
Levonian, E., Comrey, A., Levy, W., & Proctor, D. A
statistical evaluation of Edwards personal preference
schedule. Journal of Applied Psycnology, 1959* 43,
355-359.
Lewin, K. Group decision and social change. In E. E.
Maccoty, T. M. Newcomb, & E. L. Hartley (Eds.),
Readings in social psychology. (3rd ed.) New York:
Holt, 1958.
Lindquist, E. P. (Ed.) Educational measurement. Washing
ton, D.C.: American Council on Education, T951.
Lindzey, G., & Hall, C. S. Theories of personality: Pri
mary sources and researcET New York: Wiley, 1 9 6 5.
LIpsitt, L. P. A self-concept scale for children and its
relation to the children's form of the manifest anxiety
scale. Child Development, 1958* 29* 463-472.
Lowe, C. M. The self-concept: Pact or artifact? Psycho-
logical Bulletin, 1 9 61, 5 8, 325-336.
Lundholm, H. Reflections upon the nature of the psycho
logical self. Psychological Review, 1940, 47, 110-127.
Maslow, A. H. Motivation and personality. New York:
Harper & RowJ 1954.
Maslow, A. H. Toward a psychology of being. Princeton,
N.J.: D. Van Nostrand Company, 1 9 6 2.
Mayer, C. L. A study of the relationships of early special
class placement and the self-concepts of mentally
handicapped children. Unpublished doctoral disserta
tion, Syracuse University, 1 9 6 5.
McGee, R. K. Personal communication, i9 6 0. In W. H.
Fitts, Manual for Tennessee self-concept scale.
Nashvilie: Counselor Recordings and Tests, 1 9 6 5•
Mead, G. H. Mind, self, and society. Chicago: University
of Chicago Press, 1934.
Michael, J. J. Alternate approaches to the measurement of
the self-concept. Paper presented at the 48th Annual
Conference of the California Educational Research
Association, San Francisco, February 12-13* 1970.
180
Michael, J. J. Review of W. L. Thomas, The Thomas self-
concept values test. In 0. K. Buros (Ed.), Seventh
mental measurements yearbook. Highland Park, N.J.:
Gryphon Press, 1972.
Millen, L. The relationship between self-concept, social
desirability and anxiety in children. Unpublished
master’s thesis, Pennsylvania State University, 1 9 6 6.
Murphy, G., Murphy, L. B., & Newcomb, T. M. Experimental
social psychology. (Rev. ed.) New York: Harper, 1937.
Murray, H. A., Barrett, W. G., Langer, W. C., Morgan,
C. D., Diven, K., Prank, J. D., Jones, E. C., Horn-
burger, E., Mekeel, H. S., White, R. W., Mackinnon,
D. W., Rosenzweig, S., Sanford, N. R., Wheeler, D. R.,
Beck, S. J., Christenson, J. A., Cobb, E. A., Inglis,
E., Kunze, K. R., Moore, M., Rickers-Ovsiankina, M.,
Peterson, R. T., Sears, R. N., Shevach, B. J., Smith,
C. E., Trowbridge, E. H., Whitman, E. M., & Wolf, R. E.
Explorations in personality. New York: Oxford, 1938.
Nunnelly, K. G. The use of multiple therapy in group
counseling and psychotherapy. Unpublished doctoral
dissertation, Michigan State University, 1 9 6 8.
Olds, J. The growth and structure of motives. New York:
Free Press, 195*5.
Parker, J. The relationships of self-report to inferred
self-concept. Educational and Psychological Measure
ment, 1 9 6 6, 2 6, 6 9 1-7 0 0.
Perdew, P. W. Criteria of research in educational history.
Journal of Educational Research, 1950, 44, 217-223.
Perkins, H. V. Teachers’ and peers' perceptions of chil
dren’s self-concepts. Child Development, 1958, 29,
203-220.
Piers, E. V.,-& Harris, D. B. Age and other correlates
of self-concept in children. Journal of Educational
Psychology, 1964, 35., 91-95.
Piers, E. V., & Harris, D. B. Manual for Piers-Harris
children’s self-concept scale. Nashville: Counselor
Recordings and Tests, 1 9 6 9*
181
Piety, K. R. The role of defense In reporting on the self-
concept. Unpublished doctoral dissertation., Vanderbilt
University, 1958.
Rentz, R. R., & White, W. P. Factors of self-perception
in the Tennessee self-concept scale. Perceptual and
• Motor Skills, 1 9 6 7, 24, 118.
Richmond, B, 0., & White, W. F. Predicting teachers’
perceptions of pupil behavior. Measurement and Evalu
ation in Guidance, July 1971, 4, 71-78.
Rogers, C. R. Client-centered therapy: Its current prac-
tice, implication and theory. Boston: Houghton Mif-
flin, 1951.
Rogers, C. R. A theory of therapy, personality, and inter
personal relationships. In S. Koch (Ed.), Psychology:
A study of a science. Vol. III. Formulations of the
person and the social context. New York: McGraw-Hill,
1959. Pp. 185-256.
Rogers, C. R. On becoming a person. Boston: Houghton
Mifflin, 195T:
Rogers, C. R., & Dymond, R. (Eds.) Psychotherapy and
personality change. Chicago: University of Chicago
Press, 1954.
Rosenberg, M. Society and the adolescent self-image.
Princeton, N.J.: Princeton University Press, 1 9 6 5.
Sanford, N. Issues in personality theory. San Francisco:
Jossey-Bass, Inc., 1970.
Sarbin, T. R. A preface to a psychological analysis of
the self. Psychological Review, 1952, 59, 11-22.
Searles, W. B. Personal communication, 1 9 6 2. In W. H.
Fitts, Manual for Tennessee self-concept scale.
Nashville: Counselor Recordings and Tests, 1965.
Sears, P. S., & Sherman, V. S. In pursuit of self-esteem.
Belmont, Calif.: Wadsworth Publishing Co., 1964^
Sears, R. R. A treatment of worth. Review of S. Cooper
smith, Antecedents of self-esteem. Contemporary Psy
chology, 1969, l4~j 146-147.
182
Seeman, J. Toward a concept of personality integration.
American Psychologist, 1959» 14, 633-637.
Sherif, M., & Cantril, H. The psychology of ego-involve-
ments. New York: Wlleyj 1947.
Sisenwein, M. A comparison of the self-concepts of Negro
and white children in an integrated school. Unpub
lished doctoral dissertation, Columbia University,
1970.
Smith, M. B. The phenomenological approach in personality
theory: Some critical remarks. Journal of Abnormal
and Social Psychology, 1950, 45, 516-522.
Steiner, I. E. Self-perception and goal setting behavior.
Journal of Personality, 1957.» 25, 344-355.
Stephenson, W. The study of behavior. Chicago: University
of Chicago Press, 1953.
Sullivan, H. S. Conceptions of modern psychiatry. Wash
ington, D.C.: William Alanson White Psychiatric Foun
dation, 1947.
Suinn, R. M. Review of W. H. Fitts, Tennessee self-concept
scale. In 0. K. Buros (Ed.), Seventh mental measure-
ments yearbook. Highland Park, N.J.: Gryphon Press,
1972. Pp. 367-369.
Sundby, E. S. A study of personality and social variables
related to conformity behavior. Unpublished doctoral
dissertation, Vanderbilt University, 1 9 6 2.
Symonds, P. M. The ego and the self. New York: Appleton-
Century-Crofts, 1951.
Taylor, D. M. Consistency of the self-concept. Unpub
lished doctoral dissertation, Vanderbilt University,
1953.
!'
Thomas, W. L. The Thomas self-concept values test.
(Rev. ed.) Rosemont, 111.: Combined Motivation Educa
tion Systems, Inc., 1 9 6 9.
Trotter, R. J. Self-image. Science News, August 1971>
1 0 0, 1 3 0-1 3 1.
Vacchiano, R. B., & Strauss, P. S. The construct validity
of the Tennessee self-concept scale. Journal of Clini
cal Psychology, 1 9 6 8, 24, 323-326.
183
Wayne, S. R. The relation of self-esteem to Indices of
perceived and behavioral hostility. Unpublished doc
toral dissertation, Vanderbilt University, 1 9 6 3.
Wells, W. S., & Bueno, L. P. A self-concept measure of
male and female alcoholics. Unpublished manuscript,
University of Tennessee, 1957.
Wiggins, J. S., Renner, E. K., Clore, Or. L., & Rose, R. J.
The psychology of personality. Menlo Park, Calif.:
Addison-Wesley Publishing Co., 1971.
Wilhelm, R,, & Jung, C. G. The secret of the golden
flower. New York: Harcourt, 1931.
Williams, R. L., & Cole, S. Self-concept and school ad
justment. Personnel and Guidance Journal, 1968, 46,
478-481.
Wilson, J. A. R., Robeck, M. C., & Michael, W. B. Psycho-
logical foundation of learning and teaching. New York:
McGraw-Hill, 1969.
Wing, S. W. A study of children whose reported self-
concept differs from classmates' evaluation of them.
Unpublished doctoral dissertation, University of
Oregon, 1 9 6 6.
Woolner, R. B. Kindergarten children’s self-concepts in
relation to their kindergarten experiences. Unpub
lished doctoral dissertation, Memphis State University,
1966.
Wylie, R. C. The self-concept. Lincoln: University of
Nebraska Press, 1961.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The Relative Efficiency Of Multiple Regression Analysis And Multiple Cutoff Analysis In The Prediction Of Academic Performance In A Selected Medical School

PDF

The Relationship Between Student Evaluations And Self-Evaluations Of Teachers In The School Of Communications And Professional Studies At A California State University

PDF

The Selection, Training And Evaluation Of School Bus Drivers In California

PDF

Relationships Of Variables Identified With Success In College Clothing-Construction Courses: A View Toward Advanced Placement And Improved Learning

PDF

A Factor Analysis Of The Symbolic-Evaluation Abilities

PDF

Relationship Between Difficulty Levels Of Assigned English Texts And Reading Ability Of Community College Students

PDF

Nietzsche'S Philosophy Of Education: A Critical Exposition

PDF

Development And Analysis Of Some Tactual Measures Of Intelligence For Adolescent And Adult Blind

PDF

Actual And Desirable Campus Environment As Perceived By Allied Health Students Classified Into Four Creative Ability Groups

PDF

The Stability Of The Self-Concept In Junior College Students

PDF

Relationships Between Motor Proficiency And Self-Concept Of Sixth-Grade Boys

PDF

A Comparison Of The Attitudes Of Outstanding College Teachers And A Non-Selected Group Toward Four Psychological Variables

PDF

An Investigation Of Selected Aspects Of The Doctoral Degree In Physical Education

PDF

Student Perceptions Of Selected Innovations In Secondary Education

PDF

A Critical Evaluation Of The Undergraduate Accounting Curriculum In The Philippine College Of Commerce

PDF

The Relationship Of Selected Predictive Variables To Foreign Student Achievement At The University Of California, Los Angeles

PDF

A Study Of Undergraduate Education In Selected Seventh-Day Adventist Colleges

PDF

The Academic Performance Of Community College Transfer Students: An Analysis Of Consistency

PDF

The Decision Process In College Attendance And Cognitive Dissonance

PDF

An Historical Analysis Of The Origin And Development Of The College Of Medical Evangelists

Asset Metadata

Creator Shreve, Elbert Earl (author)

Core Title A Critical Analysis And Evaluation Of Evidence Regarding The Reliability And Validity Of Four Selected Measures Of Self-Concept

Contributor Digitized by ProQuest (provenance)

Degree Doctor of Philosophy

Degree Program Education

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag education, educational psychology,OAI-PMH Harvest

Language English

Advisor Michael, William B. (committee chair), Lasswell, Thomas E. (committee member), Pullias, Earl Vivon (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-810728

Unique identifier UC11364156

Identifier 7318841.pdf (filename),usctheses-c18-810728 (legacy record id)

Legacy Identifier 7318841

Dmrecord 810728

Document Type Dissertation

Rights Shreve, Elbert Earl

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA