University of Southern California Dissertations and Theses

The construct of integrative cognitive style and its measure as an indicator of advanced ego development

(USC Thesis Other)

The construct of integrative cognitive style and its measure as an indicator of advanced ego development

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE CONSTRUCT OF INTEGRATIVE COGNITIVE STYLE
AND ITS MEASURE
AS AN INDICATOR OF ADVANCED EGO DEVELOPMENT
by
Karen Lee MacDonald
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— Counseling Psychology)
September 199 0
Copyright 1990 Karen Lee MacDonald
UMI Number: DP25349
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
Dissertation Publishing
UMI DP25349
Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346
UNIVERSITY OF SOUTHERN CALIFORNIA
TH E GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFO R N IA 90007
This dissertation, written by
KarenttLeet M at c.Dona ] d..........................................
under the direction of Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
fc \$ lib ,
Ed
’ 90
3
DO CTO R OF PH ILO SO PHY
Dean of Graduate Studies
Date ... .S .£p.t.e/T tf}er.. X X *. J .9 .9 .Q .
DISSERTATION COMMITTEE
Chairperson
In gratitude to Marlene Picciolo,
a true friend and natural healer
Ill
ACKNOWLEDGMENTS
I would like to express my appreciation to those who
taught and influenced me in many ways while I was at the
University of Southern California. Frank Fox encouraged i
i
me to follow my own special interests and shepherded this |
i
research project throughout is various stages. Paul
Bloland, Rod Goodyear, and Dallas Willard provided me
. . . i
with critical assistance m spite of numerous competing
I
demands and commitments. My fellow students were [
consistently inspiring to be with, in the classroom and ^
I
l
in our many compelling hallway discussions. !
i
Philip Erdberg, Alan Karbelnig, Trudy Martin-Clark, |
f
and Karen McBride provided direction, computer services, |
and trained Rorschach scoring, upon which the soundness I
I
of this project in large part depended. I greatly i
i
appreciate the personal consultation and up-to-date i
I
materials provided by Kathryn Bobbitt of Washington
University in reference to ego development measurement
and theory, and by Martin Harrow of the Michael Reese
Medical Center in reference to overinclusive thinking and
the Object Sorting Test. I would particularly like to
acknowledge my special debt to Gail Stevenson, whose
iv
psychological understanding has significantly directed my
life both personally and professionally. Finally, I
would like to acknowledge the late David Lawrence, who
pursued intellectual adventures with gusto, passed his
I
pleasure on with enthusiasm, and supported my own
pursuits with a vision of what my future could hold.
This study was planned and the actual work of
i
writing and data collection begun with a prerequisite
sense of enthusiasm and hopefulness. Its completion is |
I
due to the inestimable encouragement, listening, and i
I
clarity that valuable friends so generously offered to me
as the work progressed, when I needed to believe that the
i
project was feasibile and my efforts worthwhile. From |
the obstacles I encountered I learned more than I
bargained for, but I gained even more from those who
i
helped me to persevere. In particular, I would like to j
thank Victoria Thoreson, Bob Hankins and Connie Hankins,
Marlene Urbaitis and Sue Harris (personal friends who |
were also valuable consultants and competent coders of
the Sentence Completion Test and Object Sorting Test,
respectively), Pamala Weistling, Phil Reich, Madonna
Brown, Maria Mora, Jorge Alonso, Nancy Verona, and my
brother Brian MacDonald, for the inspiring variety of
nourishment they provided: scholarly support, music,
food, company, and darn good advice.
V
Table of Contents
Page
Dedication....................................................ii
Acknowledgments ........................................... iii
List of T a b l e s .....................................viii
List of Figures......................................ix
Chapter
1. Introduction and Preliminary Review
of the Literature..................... 1
Ego Development........................ 13
The Concept of E g o ..................13
The Concept of Ego Development .... 17
Polar Aspects of Ego Development . . . 23
Descriptive Aspects and Measurable
Signs of Ego Levels......................2 5
Definition of Cognitive Style ............. 3 3
Field Dependence-Independence ............. 3 7
Global versus Articulated
Cognitive Styles.......................... 3 9
Psychological Differentiation ......... 40
Autonomy: Internal versus
External Referents .................... 4 2
Points of Comparison ...................... 4 4
r
VI
Critique of the FDI Construct
and Preliminary Conclusions ............ 4 8
Dimensional Characteristics ............ 4 8
Autonomy and Interpersonal Relations . 56
Differentiation and Integration .... 58
Cognitive Structuring
and Restructuring........................61
Theoretical Inconsistencies ............ 62
Conclusions................................. 64 j
2. The Construct and Measure j
of an Integrated Cognitive Style............ 68
Relations to Broad Universal Processes. . 69
Phenomenology: Self and Object .... 69 !
Gestalt Psychology: The Whole
and its Parts....................... 7 6
General Systems Theory: Differentiation |
and Integration....................... 7 9 :
Categories and Information Integration. 81 I
Abstract versus Concrete Thought. . . . 88 i
Abstract versus Integrative Thought . . 91 |
I
Relations to Narrower |
Cognitive Constructs.....................9 6 j
Relations to Creative Thought ............ 10 0
Complexity and Synthesis. . . . . . . . 101
Field Dependence-Independence ......... 102
Hemispheric Specialization..............105
Disordered Thought....................... 106
Personality Correlates...................112
Optimal Functioning
and Cognitive Style...................114
A Unified Definition
of Integrative Cognitive Style......... 116
Conceptual Dimensions
of the ICS Construct...................119
Operationalizing
the Integrative Style Construct . . . . 12 3
vii
The Goldstein-Scheerer OST........... 124
The Rorschach Test..................... 127
Theoretical Summary and Overview
of Empirical Aspects................... 137
3. Empirical Methodology ........................ 142
Research Design
and Control Procedures.............. 142
Selection and Classification
of Subjects.......................... 144
Instrumentation .......................... 148
Procedure...............................157
Data Analysis............................17 6
4. Analysis and Interpretation of Results. . . 180
Ego Level
and Integrative Cognitive Style . . . .181
Hypothesis 1............................ 181
Hypothesis 2......................... 192
Follow-up Analyses Applicable
to Hypotheses 1 and 2 .............. 201
Ego Level
and Analytic Cognitive Style..........2 08
Convergent and Discriminant Validity. . .210
5. General Discussion, Conclusions,
and Recommendations..................... 215
Ego Development and Cognitive Style . . 221
Measurement of ICS..................... 226
Conclusions............................ 230
Recommendations for Future Research . . 231
References................................................ 238
Appendixes
A. Object Sorting Test Items..................2 57
B. Ogive Scoring Rules for the 18-Item
Sentence Completion Test ................ 2 58
C. Summary of Key Abbreviations and Codes. . . 259
Tables
Page
Some Milestones of Ego Development
Indices of Interrater Agreement by Items
for the Sentence Completion Test (SCT),
Six Rorschach Coding Categories, and
Two Object Sorting Test (OST) Scales. . . 160 j
1
Description and Scoring
of Three Rating Scales for the I
Object Sorting Test (OST) ................ 166 j
Hypothesized Rorschach Integrative
Style (ICS) Index...................... . 174 '
Descriptive Statistics of Selected
Rorschach Variables for High and Low |
Ego Development (ED) Groups Compared 1
to Normative D a t a ..........................18 2 1
Chi-square and t-test Analyses
of Hypothesized ICS Indices for High j
and Low Ego Development (ED) Groups . . . 19 0 I
Results of Multiple Discriminant Analysis |
of Integrative Cognitive Style (ICS) t
Differences Between Low and High !
Ego Development Groups (N = 4 0 ) 194 |
I
Hit Rates Using a Discriminant Function to I
Predict Level of Ego Development 199 !
Chi-square and t-test Analyses
of the Group Embedded Figures
Test (GEFT) Scores for High
and Low Ego Development (ED) Groups . . . 2 09
Intercorrelations Among
Integrative Cognitive Style (ICS),
Field Independence (FI),
and Ego Development (ED) Variables. . . . 212
r
Figure
1.
2 .
ix 1
i
i
i
i
Figures
Page j
Distribution of Group Embedded Figures
Test (GEFT) Mean Scores Across Ego
Levels (ELs) for a Postconformist Group
and a Heterogeneous Comparison Group . . 6
Plot of Individual Discriminant I
Scores for High and Low j
Ego Development (ED) Subjects
Around their Own Group Centroids (D). . . 2 00 i
1
CHAPTER I
Introduction and Preliminary Review of the Literature
Theories about the nature of personal growth and
change and the search for their signs are at least as old
as the belief attributed to Socrates by Plato that evil is
supplanted by virtue with an increasing knowledge of the
self (Hawtrey, 1981; Plato, ca. 367BC/1965). In
successive transmutations of the topic, simple growth has
increasingly been recast as a dialectical process, and
change as a progressive psychic reorganization. In other
words, a distinctly developmental point of view has been
adopted, with origins in Darwin's evolutionary theory of
species change for more effective adaptation (Achenbach,
1978). A major contemporary investigator who has applied
this framework to the study of the self or ego and its
development is Loevinger (1966, 1969, 1973, 1976, 1979b,
1979c). She based her ego development (ED) theory on a
synthesis of her own and other pertinent theories as well
as extensive empirical data collected over a 3 0-year
research career, in turn constructing and refining the
Washington University Sentence Completion Test (SCT) and
its comprehensive scoring manuals (Loevinger & Wessler,
2
1970; Loevinger, Wessler, & Redmore, 1970). ED defines a
stage progression in which the ego moves from
preconformist to conformist to postconformist ego levels
(ELs), characterized in part by increasing cognitive
complexity, psychological autonomy, and rich interpersonal
relations, as summarized on Table 1. Manifest gualitative
changes coincide with underlying quantitative change
throughout these various realms of functioning.
Signs of such growth and change exhibit a wide
variation among individuals and contribute to an
understanding of personality differences. Cognitive
processes have become an important source of hypothetical
explanations for both nomothetic and idiographic behavior
during the past 3 0 years, as the study of mind has
increasingly regained its respectability within academic
psychology (Gardner, 1985). Included among the many
cognitive processes of interest is the construct of
cognitive style, which refers in general to varying
dispositions towards modes of information processing.
One cognitive style has engendered an especially
prolific line of research within the original research
group and spawned many hundreds of additional
investigations: field dependence-independence or FDI
(Witkin, Dyk, Faterson, Goodenough, & Karp, 1962/1974;
Witkin et al., 1954/1972). Originally conceived as a
3
Table 1
Some Milestones of Ego Development
Code Level Impulse Control,
"Moral" Style
Interpersonal
Style
Conscious
Preoccupations
Cognitive
Style
El Presocial
El Symbiotic
E2 Impulsive
E3 Self-
protective
E4 Tran
sitional
E4 Conformist
E5 Self-aware
E6 Conscien
tious
Impulsive, fear
of retaliation
Fear of being caught,
externalizing blame,
opportunistic
Obedience and
conformity to social
norms are simple
and absolute rules
Conformity to external
rules; shame, guilt
for breaking rules
Dawning realization
of standards,
contingencies,
self-criticism
Self-evaluated
standards,
self-criticism
Autistic
Symbiotic
Receiving,
dependent,
exploitive
Wary,
manipulative,
exploitive
Manipulative,
obedient
Belonging, help
ing, superficial
niceness
Being helpful,
deepened
interest in
interpersonal
relations
Intensive,
responsible,
mutual,
concern for
communication
Self vs. nonself
Self vs. nonself
Bodily feelings,
especially sexual
and aggressive
Self-protection,
wishes, things,
advantages, control
Concrete aspects of
traditional sex roles,
physical causation as
opposed to psycholo
gical causation
Appearance, social
acceptability, banal
feelings, behavior
Consciousness of the
self as separate from
the group, recogni
tion of psychological
causation
Differentiated
feelings, motives for
behavior, self-
respect, achieve
ments, traits,
expression
Stereotypy,
conceptual
confusion
Conceptual
simplicity,
stereotypes
Conceptual
simplicity,
stereotypes, cliches
Awareness of individ
ual differences in
attitudes, interests,
and abilities: men
tioned in global and
broad terms
Conceptual
complexity, idea
of patterning
E7 Individual- Individuality, coping Cherishing of
istic with inner conflict interpersonal
relations
E8 Autonomous0 Add: Coping with
conflicting inner
needs, toleration
Add: Respect
for autonomy
Communicat ing,
expressing ideas and
feelings, process and
change
Vividly conveyed
feelings, integration
of physiological and
psychological, psycho
logical causation of
behavior, development,
role conception, self-
fulfillment, self in
social context
Toleration for
paradox and
contradiction
Increased conceptual
complexity;
complex patterns,
toleration for
ambiguity, broad
scope, objectivity
E9 Integrated3 Add: Reconciling
inner conflicts,
renunciation of
unattainable
Add: Cherishing
of individuality
Note. Adapted from "Loevinger's Model and Measure of Ego Development: A Critical Review" by S. T.
Hauser, 1976, Psychological Bulletin. 83., p. 933.
“Add means in addition to the description of the previous level.
perceptual construct, FDI was delineated as a continuum
ofstable individual differences that is global on one end
and articulated at the other end. These cognitive styles
in turn are a reflection of degree of psychological
differentiation, and account for pervasive differences in
individual functioning. Witkin proposed that a more
differentiated ego would evidence a more articulated
(analytical or field-independent) cognitive style rather
than a global (diffuse or field-dependent) style. Witkin
further postulated that this difference would be expressed
in performance on a cognitive restructuring task known as
the Embedded Figures Test (EFT) or Group Embedded Figures
Test (GEFT) (Witkin, 1950; Witkin, Oltman, Raskin, & Karp,
1971), in which a simple figure must be located and
isolated within a complex design.
Because cognitive style is a consistently featured
aspect of Loevinger's stage sequence and because the
description of the FDI dimension paralleled that of
Loevinger's ED in many respects, MacDonald (198 6) proposed
that comparable performance on the GEFT would provide
further confirmation of Loevinger's theory, and a
construct validity study was designed. Particular
interest was focused on performance at higher
postconformist ELs, where high GEFT scores indicating FI
were expected.
5
However, when the GEFT and the SCT were administered
to a female student comparison group comprising a
heterogeneous range of ELs and a separate postconformist
ED group of women, an indication of a direct relationship
between FDI and EL in the comparison group was reversed in
the postconformist group. As Figure 1 illustrates, the
rate of increase in GEFT scores for the comparison group
began to drop off between the Self-aware and
Individualistic ELs. This suggested decline at
postconformist ELs in the comparison group became a
pronounced trend in the postconformist ED group. Instead
of supporting the proposition that postconformist ELs are
characterized by the greatest amount of FI, the data of
the study suggested that cognitive style Figure 1 becomes
increasingly FI up to a certain point in ED close to the
Conscientious EL, and then drops off. Another discrepancy
was a near 4-point difference in mean GEFT scores between
Individualistic subjects from the postconformist and
comparison groups. The difference placed these subjects
in the second and third quarters of the normative GEFT
distribution, respectively. Because Witkin makes no
specific prediction for the immediate score range, it is
not necessarily of theoretical significance. It may
suggest, however, that age effects be included when
considering the developmental implications of cognitive
6
Figure 1
Distribution of Group Embedded Ficrures Test (GEFT)
Mean Scores Across Ego Levels (ELs) for a Postconformist
Group and a Heterogeneous Comparison Group
CO
C D
o
o
CO
h~
LI
LLI
CD
c
C O
C D
13
• Postconformist group
O Com parison group
12
11
10
9
8
7
6
7 3 6 8 4 5
Ego Levels
(Self-protective [3] = preconformist EL; Conformist [4]
and Self-aware [5] = conformist ELs; Conscientious [6],
Individualistic [7], and Autonomous [8] = postconformist
ELs. Range of GEFT scores is 0-18; higher scores reflect
relatively less field dependence and more field
independence.)
7
style: average age for the postconformist ED group was 55
years versus 25 years for the comparison group. In the ED
model, age is a contributory but not determinative factor;
individual ages vary widely at each EL, and people
equilibrate more or less permanently throughout the
sequence.
The foregoing study (MacDonald, 198 6) presented
theoretical anomalies and measurement issues that
ultimately informed the concerns addressed by the current
study. It was unclear whether the drop in GEFT scores
represented the inadequacy of the FDI variable to indicate
EL, or a systematic association in an unexpected (reverse)
direction. If the postconformist ED group were in fact
FD, it would not be in accord with the assumption of the
Einheit or self-consistency of personality, a position
supported by both Loevinger and Witkin. Perhaps two
different kinds of cognitive styles were responsible for
low GEFT scores. One would be the FD style described by
Witkin (global, passive, relatively undifferentiated and
lacking in autonomy) whereas the second might be a
holistic style in which there is no lack of a
differentiated self or object, but in which integrative
processes receive primary emphasis.
In light of such confusion, the FDI construct needs
to be examined more carefully for possible sources of
8
internal inconsistency. When terms such as "global" and
"holistic," or "autonomy" and "independence," are treated
as synonymous and overapplied to a wide range of data,
they may be transformed into unexamined metaphors and
therefore distort interpretation of data.
The problems associated with FDI multiply when the
GEFT is used as a criterion measure and construct validity
conclusions are derived from its scores, as in the FDI-ED
study (MacDonald, 1986). In particular, more data are
needed to refine and sharpen the psychological meaning of
high ED; but confounding variables are many and valid and
reliable criterion measures are hard to find. Many
researchers have directed efforts toward establishing the
predictive and construct validity of the SCT. A diverse
assortment of relationships and behavior patterns
predictable from the ego development model have been
investigated, including helping behavior (Cox, 1974),
responsibility-taking (Blasi, 1972), conformity (Hoppe &
Loevinger, 1977), interview behavior (Lucas, 1971), coping
strategies (Haan, 1977), moral development (Lambert, 1972;
Sullivan, McCullough, & Stager, 1970), and social
desirability (Redmore & Waldman, 1975). The one specific
study of construct validity of cognitive processes in ego
development provided a modest correlation between the SCT
and a cognitive complexity scale (Sullivan et al., 1970).
9
Indicators of more complex development are
correspondingly more difficult to untangle. Previous
evidence for the particulars of postconformist functioning
have been equivocal, due to the frequent use of younger
subjects and the invariably restricted score distribution
on the SCT. The strongest construct validity evidence for
postconformist ED comes from three studies in which a
postconformist group differed from the others in a manner
consistent with Loevinger's model. Erickson (1977) found
that subjects at these ELs viewed roles and rights of
women with greater equality and complexity, and Candee
(1974) described postconformist subjects as reporting more
complex political ideologies. Lawrence and MacDonald
(1984) found that a group of career-changers, none of whom
sought a change primarily for financial reasons, scored in
the postconformist range of ED. These findings concur
with the documentation of work attitudes throughout the
developmental sequence, in which work becomes less onerous
and more representative of the search for self-
fulfillment. No studies were found, however, that have
examined the nature of the cognitive processes that might
result in such conceptions.
In an extensive critical review of investigations
on ED reliability and validity, Hauser (1976) examined
various strategies for establishing validity and their
10
corresponding degrees of success. He criticized studies
that employ a single polar variable like IQ to examine its
overall correlation with the ED score. He recommended
that any program of future construct validation give
careful consideration to the theoretic derivation of the
variable under study, to the type of trend predicted
between it and ED, and especially to the location and
prediction of relationships between salient variables at
specific ELs. Thus he suggested sampling from varying ELs
and a comparison by stages design, with predictions based
on a description of specific ELs.
Given such an improved research design, however, a
dilemma remains that no appropriate method for measuring
cognitive processes hypothesized to operate at high ELs
are readily apparent, due in part to the incomplete
conceptualization of such processes. Although it is
possible that high ED samples can exhibit a FI cognitive
style under certain altered measurement conditions, the
theoretically-based expectation of the presence of a
separate cognitive style seems especially convincing. The
current study was therefore designed to address these
conceptual and methodological concerns, consolidated into
a general working thesis as follows: Witkin left out an
integrative cognitive style (ICS) in his theory and his
measure, but this style is how people typically process
11
information at higher levels of ego development. Specific
objectives that need to be met in substantiation of this
thesis are:
1. The generation of procedural evidence to test the
adequacy or inadequacy of the FDI construct, along with a
review of empirical work. Procedural evidence, according
to Rychlak (1981), results from the submission of a
theoretical proposition to a cognitive or conceptual
method. When a proposition makes sense and seems
implicitly true after testing it conceptually, this
evidence permits us to proceed, based on our best judgment
that the tested line of thought is correct.
2. The generation of procedural evidence, through a
critical review of relevant literature, to support the
proposition of the existence of ICS as distinct from FDI,
and to provide the source of its conceptual definition.
3. The derivation of an empirical measure of ICS as
distinct from the GEFT, based on its conceptual
definition.
4. The generation of validating evidence (empirical
support) for the link between high ED and ICS, by testing
a high ED group and a low ED group with the GEFT and ICS
measures and analyzing the scores statistically for the
presence of differences in the direction expected from
Loevinger's model and the ICS construct.
12
It was hoped that results of this study would serve
to clarify the nature of cognitive processes at
postconformist ELs by eliminating some of the sources of
confusion and unintentional bias that often accompany the
study of personality and cognitive constructs. The ICS
hypothesis seeks to address the relative neglect of the
topic of advanced development by empirical research, in
the form of tangible support for the ED theory. Several
overlapping theoretical accounts of continuing self
development now exist, but empirical data can lend
structure and applicability to theory while providing
necessary attention to its limitations. This is important
to clinicians and academic psychologists alike, both of
whom use an implicit if not explicit conception of
advanced development to frame an understanding of
psychopathology. The diverse characteristics of
individuals who equilibrate at different developmental
levels should therefore interest any professional who
hopes to foster therapeutic growth and change, or to
assist clients by pinpointing strengths and anticipating
associated weaknesses. In particular, obtaining a clearer
picture of the direction in which efforts towards self
enhancement at high ELs are aimed could strengthen the
practitioner's ability to offer accurate mirroring and
pacing to clients at all ELs.
13
Background information pertaining to the theoretical
and empirical support of this study (i.e. the
substantiating objectives listed above) follows next.
First, the model and measure of ego development and its
precursors are presented, including specific examples of
cognitive processes at various ELs. After explaining the
concept of cognitive style in general, the evolution of
the FDI construct and its characteristics are considered.
At the conclusion of this background information, key
points of comparison of the two theories are summarized,
and the FDI construct is critiqued. These sections
elucidate the subsequent search for the integrative style
construct and the derivation of its measure, which in turn
prepares the way for the presentation of the empirical
hypotheses and research design. An explanation of the
empirical methods and research results are reported in
following chapters, concluding with a general
discussion.
Ego Development
The Concept of Ego
Loevinger rarely uses the term ego by itself,
separate from the phrase ego development. The regular
combination of the two words indicates her emphasis: the
ego is more like an act than an entity. In fact, her
usage of the term is close to the original conception of
14
Freud, whose own term the I (das Ich) was purposely chosen
to enhance its personal, agentic connotation. A
hypothetical construct such as ego has the advantage of
explanatory power, but a certain degree of accompanying
ambiguity is required by this same strength (Loevinger,
1979b). Loevinger is nonetheless able to specify several
key characteristics of the term's conceptual range:
The ego is a process, not a thing. Attempts to
define the ego in spatial terms are misguided because it
exists, like life itself, as a process of interchange with
the environment. Neither is it merely a collection of
adaptive functions or "apparatuses" of learning and
perception, a position of one prominent school of
psychoanalytic ego psychology (Hartmann, 1939/1958).
Instead of the synthetic function of the ecro being one
among many equally important functions (Nunberg,
1931/1948), the act of synthesizing, or the search for
coherent meanings, is the essence of ego (Fingarette,
1963; Loevinger, 1969).
The ego functions holisticallv. Because the
synthetic function, or organization, is what ego is,
organic unity is the substance of Loevinger's field of
inquiry. The person is not divisible. A holistic
paradigm is ascribed to by many, but it has led more often
to ontological mind/body discussions than to actual
15
research programs, where individual traits continue to be
analyzed separately. Loevinger (1976) asserts that a
holistic approach can also accommodate the psychoanalytic
concept of inner conflict, but without resorting to the
additional "entities" proposed in other interpretations:
although the ego functions as a whole, some memories or
experiences are retained outside that frame of reference.
The ego is structural. The striving, holistic ego
seeks self-consistency, or a structural organization. It
can thus be thought of as representative of one's whole
frame of reference, an idea anticipated by Adler
(Ansbacher & Ansbacher, 1956), who conceived of the unity
of the personality and its general approach to problems as
one 1s style of life.
Sullivan (1953) preferred to call the ego the self-
system. and his account of its maintenance is central to
Loevinger1s approach. Through selective inattention,
observations inconsistent with one's current state are
screened out, thereby preventing the anxiety that is
aroused whenever events or interactions are present that
the self-system cannot construe successfully. If
incompatible data were not excluded or amended, the
existing self-system might itself have to be revised, a
further cause for painful anxiety.
16
This theory of ego stability can account for the
stability of any type of mental structure. Assimilation
(Piaget 1936/1952), dissociation (Merleau-Ponty,
1942/1963) and cognitive dissonance (Festinger, 1957) all
describe maintenance processes that seem much like the
ones used to stabilize scientific paradigms, as well
(Kuhn, 197 0). An equally significant, structurally-
oriented account exists of the relationship between ego
and anxiety (Fingarette, 1963), summarized by Loevinger
(1976) as follows:
Failure of the striving for meaning is
itself what constitutes anxiety. Just as
meaning is not so much what the ego seeks
as what it is, so anxiety is not what the
ego experiences but is the opposite of
ego: disorganization or meaninglessness.
(p. 61)
Structure, then, is the central function of the ego, and
it is formed to help set one's course in the world. In
this sense, the terms "character structure" and "ego" are
nearly synonymous.
The ego is purposive. The preceding points emphasize
the premise that behavior is meaningful. In fact,
meanings and purposes direct the actions of the ego, "as
opposed to concepts such as psychic energy that pervade
certain psychoanalytic accounts" (Loevinger & Wessler,
197 0, p.3). An individual is more a subject who is drawn
along by self-discovered meaning than an object of
17
instinctual drives. This characteristic contributes to a
more phenomenological, humanistic perspective of ego, and
represents a counterpoint to mechanistic-deterministic
views of behavior.
The ego is social. In the tradition of Mead (1934),
Cooley (1902/1964), Baldwin (1897/1902), and especially
Sullivan (1953), the ego is seen to be fundamentally
social in character. Interpersonal relations play an
integral role in both structural and purposive aspects of
the ego as previously described, by fundamentally
constituting both intrapsychic schemes and purposes. This
view stands in distinct contrast to its alternative, in
which society is imposed upon an intrinsically selfish or
instinctual individual.
The Concept of Ego Development
Ego development is a sequence of steps along one
single continuum, each marked by its own distinct frame of
reference. Table 1 characterizes the various steps or
stages (ELs) in this progression by referring to major
aspects, such as impulse control, character, interpersonal
relations, conscious preoccupations, and cognitive
complexity. However, these are only aspects of the single
dimension of ED, which is itself more like the "master
trait" of personality (Blasi, 1976). Blasi describes ED
as the framing around which more specific traits are
articulated into personality. Progress along this
dimension from one EL to the next is marked by increasing
differentiation of one's inner life and interactions with
the world, and by increasing integration of complexity.
Another way of describing ED is as a transformation of
motives and values (Loevinger, 1973) .
Origins of the construct. Loevinger1s interest in
the mid-1940s focused on the domain of personality traits;
she devised a Family Problems Scale to assess mothers'
attitudes towards problems of everyday family life
(Loevinger & Sweet, 19 61). Although guided by hypotheses
derived from the psychoanalytic theory of psychosexual
stages and related personality characteristics, the item
content intentionally covered a far broader range. When
Loevinger and her coworkers clustered the items for
maximum discriminating power, they then obtained
unexpected but illuminating results.
The most general factor that emerged contained not
only the anticipated punitive-permissive dimension, but it
also contained items referring to the capacity for
conceptualizing a child's inner life and a penchant for
banal expressions of sentiment. The factor seemed to
delineate a more diffuse syndrome reminiscent of the
authoritarian personality described by Adorno, Frenkel-
Brunswik, Levinson, and Sanford (1950), though none of
19
Loevinger's items referred to political beliefs. A
capacity for inner life had also emerged in the Berkeley
research group's data as a more convincing correlate with
political beliefs than psychosexual development, contrary
to their original expectations. For these reasons, the
cluster was titled Authoritarian Family Ideology (AFI)
(Loevinger, 1962, 1976, 1979c).
A subsequent observation of great consequence was
that the immature extreme of the AFI continuum is not
authoritarianism, but a chaotic, impulsive, narcissistic
life-style. Looked at in this light, the act of
conforming to authoritarian order represents an advance,
and shifts the conceptual extreme point to a developmental
midpoint. "This insight was a turning point in my
intellectual history, changing me from a psychometric
psychologist into a developmental one, from a trait
theorist into a structuralist" (Loevinger, 1979c, p.7).
Ego development was the only term that Loevinger believed
could adequately encompass the many wide manifestations
measured by AFI. Her data had thus led her from the study
of sexuality to the study of ego psychology, a path many
psychoanalysts were following as well. She discovered
that others had also discerned a similar developmental
sequence, variously described as moral ideology (Kohlberg,
1964), cognitive complexity (Harvey, Hunt, & Schroder,
20
1961), and interpersonal integration (Sullivan, Grant, &
Grant, 1957) .
No one had as yet constructed a specific measure for
ego development. Based on the belief that only a
projective test could appropriately elicit the subject's
frame of reference as a sign or measure of EL, Loevinger
constructed the SCT (Loevinger & Wessler, 1970; Loevinger
et al., 1970). She began by using four stages delineated
by Sullivan et al. (1957): Impulsive, Conformist,
Conscientious, and Autonomous. As the data accumulated
and indicated levels beyond these initial parameters,
additional ELs were outlined and supplemented from a
variety of theoretical sources to fashion the current 10-
point scale.
Manor features of ego development. First, ED is
dialectical. At the heart of this assertion is the
distinction between polar variables and milestone
variables, which the identification of AFI's developmental
midpoint clarified. Trait theory and trait measurement
are based on logical polarities, so that the differences
between people are proportional. The application of this j
approach to the study of development produces a
measurement of how far someone has come along a path to an
endpoint, such as adult status. In the case of abilities,
development does essentially coincide with the
21
psychometric continuum, increasing monotonically with age.
In contrast, the developmental course of personality does
not conform to the logical polarities of its assembled
traits, but more closely resembles a series of
qualitatively different points whose ordering is given by
age. There are different signs for each stage— a
predominant trait or set of traits that increases in
probability up to a certain point and thereafter
diminishes, in favor of the next type of functioning.
This course of growth is better described as a sequence of
overlapping, curvilinear milestones (Loevinger, 1976,
1979c).
A milestone sequence is "built on the metaphor of
organism..., using structure as a key concept" (Blasi,
1976, p. 29). This model of development has been most
extensively investigated by Piaget (e.g., 1970) and Werner
(1940/1964). Although the qualitative changes between
milestones strongly suggest a discontinuous movement, it
is actually not known whether the changes occur in
disruptive jumps or by gradations. In fact, abilities as
well as personality traits have both polar and milestone
manifestations, depending on the perspective. For
instance, polar aspects of personality become curvilinear
milestones within the context of development.
22
Second, ED is typological, in the sense that each
developmental level can be found as a stable type in adult
life. This does not negate the deeper continuity of stage
development, the inner logic through which each level
builds on its predecessor. It does suggest, however, that
most people equilibrate far below the highest level, in
contrast to Piaget's cognitive development. If ED is akin
to an "exchange of paradigms," advancement may not always
seem favorable (Loevinger, 1976). When the manifestations
of ED are construed as polar variables, this ordering of
types is lost, for each type then seems to be
independently defined.
Third, ED implies structural change. Development is
more than a simple incremental process of addition,
indeterminately subject to conditioning. New structures
are built on new organizing principles, breaking from the
old ones. At the same time, continuity between the
successive structures is preserved by our predilection for
selectivity and integration of material. The relatedness
of the steps in the sequence therefore determines the
range of possibility, and the environment influences the
specific nature of the change (Blasi, 1976).
Fourth, ED is an abstraction— that is, a hypothetical
construct. As a stage sequence and a typology, it can
never be directly observable, but can only be inferred
23
from the empirical data that are gathered. Ego
development is independent of specific ages and
environmental conditions and represents laws of the
developmental process per se (Blasi, 1976; Loevinger,
1976).
Polar Aspects of Ego Development
Only a few measurable traits can be considered to be
polar aspects of the underlying quasi-quantitative
dimension of ED, in terms of constant progression from one
extreme to the other. One immediate issue is how to
characterize the high end of the continuum. In some
sense, the whole point of Loevinger's enterprise is that
there are many way stations along the course, all of which
are significant. Good adjustment can be experienced at
all ELs except the lowest ones, which may represent some
kind of maladjustment in adult life. Furthermore, the
attainment of all good things is not a requisite for the
highest ELs, where the acknowledgement of internal
conflict regarding needs and duties is first evidenced—
rarely at the Conscientious EL, and almost never below
that. Evidence of such conflict, however, would earn the
respondents a low rating on many good adjustment scales.
It is therefore equally incongruous to apply the term
optimum mental health to measures whose aims could more
24
accurately be described as the high levels of ED, although
the misapplication often occurs (Loevinger, 197 6).
With this qualification, Jahoda's (1958) summary of
current concepts of positive mental health seem to
correspond to the high end of the developmental continuum.
They include the regulation of behavior from within and an
independence from pressures from the immediate
environment— or the sense of identity and autonomy.
Because reality is shaped in an interpersonal context,
empathy and social sensitivity must increase, also, to
provide for accurate perception unencumbered by inner
distortion.
Another polar variable identified by many researchers
is the growth of realistic thinking in children. Various
distinctions of this basic dimension include
primary/secondary process, realistic/autistic thinking,
prototaxic/parataxic/syntaxic modes, and Piaget's stages
of intellectual development. An additional polar variable
from this domain is cognitive complexity, which depicts ED
as a succession of increasingly complex views of the world
(Loevinger, 1976). Each of these series of achievements
along the dimension of ED is marked by increasing
internalization, as well, in which the self is
acknowledged and relied upon as the principal author of
meaning, of choices, and of one's responses to them. "The
25
expansion of inner consciousness is simultaneously an
expansion and objectification of the world in which we
live and of our selves" (Loevinger, 1976, p. 139).
Descriptive Aspects and Measurable Signs of Ego Levels
Most responses to the original 3 6 items on the SCT
can be found in the categories of the scoring manual
(Loevinger et al., 1970), but there will always be
deviations from them in any new sample. The method of
rating a novel response must adhere to the underlying
logic of the construct: "not the specific content but the
thought structure it reveals is decisive" (Loevinger,
197 6, p. 235). Contrary to original expectations, in
fact, Loevinger discovered that subjects often revealed no
clear differences in the kind of content discussed at
different ELs. Instead, issues that appeared to be
logically equivalent on the surface were discussed with
diverse shades of emphasis. It was also discovered that
the ability to hold in mind contrasting, even
contradictory alternatives at one level of complexity may
assist in achieving the next higher level of complexity
and ED. In the following paragraphs, a general summary of
the quality of responses at each specific EL is presented,
in order to render the principles of ED elicited by the
SCT more precisely.
26
Impulsive. The Impulsive subject tends to
dichotomize the world into good or bad, mean or nice,
clean or dirty. Besides conceptual oversimplification,
conceptual confusion appears on Impulsive protocols;
causation carried by the word "because" is beyond these
subjects, and is understood only in concrete terms.
Tautological responses that echo the thought are frequent.
The subject seems unaware of reciprocity and considers
people to be sources of supply. Frank dependence is
expressed by going home as a solution to trouble, located
in a place rather than a situation. Impulses and affects
are often primarily body states, dysphoric moods, or bland
positive responses that are inappropriate in context.
Differentiated and cognitively shaded emotions are
missing.
Delta. This subject must be on guard at all times to
control the situation and to control herself. The world
is divided into those who take advantage of you and those
you can manipulate. Rules are to avoid hard work and look
for fun. The Delta subject displays hostile, callous
humor, especially on items that evoke sympathy or
idealization from other responses.
Conformist. Formulas for what happens in the world
tend to be stated in absolute terms, in sweeping
generalizations, and in absolute standards of right and
27
wrong. Sentimental idealization is common. Subjects
accept conventional social norms without personal
evaluation, especially conventional sex roles, and
belonging makes them feel secure. There is preoccupation
with physical appearance, and the concrete and external
aspects of life. Interpersonal interaction is described
in terms of behaviors rather than in terms of
differentiated feelings, motives, or traits. Inner life
is mentioned in terms of generalities and banalities. Sex
and love are treated as contrasting ideas, if they are
mentioned together at all, and love is taken in its
concrete meaning of taking care of— implying that love is
an aspect of dependence and security needs. Another
aspect of the Conformist*s concrete view is the
classification of people in terms of superficial,
demographic characteristics.
Self-aware. The Self-aware person sees multiple
possibilities, alternatives, exceptions, and comparisons,
though they are global and often banal. Self-
consciousness and rudimentary self-criticism is
prototypic. This subject has a stronger awareness of
feelings and individual differences than does the
Conformist person; the virtues emphasized are homely ones
and merge with elementary and vague social concerns.
Interpersonal relations are described in terms of
28
feelings, and behavior is seen as having a reason, but
this is stated in general. Elementary conceptions of
purpose, goals, and opportunities are present at this EL.
Conscientious. True conceptual complexity is shown
at the Conscientious EL, in contrast to the Conformist’s
conceptual simplicity and the Self-Aware*s multiplicity.
A response will often combine alternatives that are polar
opposites, but not so global, stereotyped, or evaluative
as at lower levels. The achievement motive and self-
improvement are at their height at the Conscientious EL,
and subjects have such a strong sense of responsibility
that they may have an excessive feeling of molding others.
They distinguish moral from aesthetic standards and means
from ends. Conceptual complexity also produces a richly
differentiated inner life and a vivid, realistic sense of
people and individual differences. Subjects display a
clear conception of interpersonal mutuality and more
differentiated ideas such as trust, respect, needs, and
emotional support. The Conformist reduces relations to
the behavior of talking, whereas the Conscientious person
thinks in terms of communication and expression. A clear
notion of psychological causality and a long time
perspective are evident. Appearances are distinguished
from underlying feelings and the physical from the mental,
whereas at higher ELs their integration and mutual
29
interdependence may be mentioned. The Conscientious
person is concerned with the problem of dependence-
independence, and does not see the inevitability and
desirability of emotional interdependence, as do subjects
at higher levels.
Individualistic. Above the Conscientious EL, most
responses are unique, though categorical aspects can be
pointed out. A frequent type of complexity at the
Individualistic EL consists of combinations of replicated
responses rated singly at lower levels, with at least one
aspect at the Conscientious EL. The subject has available
in one breath the thoughts that subjects at lower levels
select as alternatives. The Individualistic subject sees
a paradoxical contradiction in nature rather than polar,
incompatible opposites. Relations are seen as changing
over time, and their value may partially supplant the
value of ability and achievements at the Conscientious EL.
Psychological causality has become complex, contrasting
with vaguer references to "reasons" and "problems" at the
two lower levels. Dependence and independence remain a
problem but are seen in terms of emotional dependence or
emotional problems. A related problem is that of
individuality. Unlike the Conscientious subject who sees
a sharp dichotomy between mind and body, the
Individualistic subject sees interaction and integration,
30
and is often aware of conflicting emotions within the
self, rather than between one's needs and society's
requirements. The full force of intrapsychic conflict is
not felt until the next level, however. The
Individualistic subject goes beyond the perception of
individual differences to true toleration of others, and
will often reveal a broad view of life as a whole.
Autonomous. A distinguishing characteristic of
Autonomous protocols is the richness and variety of topics
found within one protocol. The Autonomous subject
considers conflicting alternatives to be aspects of many-
faceted life situations, and departs from the either-or
view of the Conscientious EL by offering composite
responses that include several possibilities in a
situation, such as appearances, actions and feelings. The
Autonomous subject feels the full force of inner conflict
and strives to cope with it. Whereas the Individualistic
subject is struggling with the problem of emotional
dependence and the attainment of individuality, the
Autonomous subject cherishes individuality and uniqueness
in herself and others. This person is concerned with
self-realization, which merges with enjoyment of life, a
dislike of stultifying work, and a cherishing of relations
with other people. The problem of identity appears here
in general form only, in contrast to the next EL.
31
However, there is a clear conception of roles and their
associated problems for women. Social interaction is seen
in its complexity and circularity.
Integrated. In most social groups one will barely
find 1% at the highest Integrated EL. Most people do not
display straightline growth from the lowest to the highest
EL, and only a few people reach the level of transcending
conflict and reconciling polarities that describes the
Integrated EL. Maslow (1954) describes such individuals
as self-actualizing. If the successive levels of ED
represented a series of problems to be solved, those at
the highest level might be considered to be the best
adjusted, but this is a distortion. There are well-
adjusted people at all ELs. Although one could make a
case for labeling those adults who remain below the
Conformist EL beyond childhood as maladjusted, some self-
protective, opportunistic persons become very successful.
Loevinger1s caveat regarding the nonequivalence of ED and
mental health is expressed in the following statement:
Certainly it is a conformist's world, and
many conformists are very happy in it,
though they are not all immune to mental
illness. Probably to be faithful to the
realities of the case one should see the
sequence [of ED] as one of coping with
increasingly deeper problems rather than
to see it as one of the successful
negotiation of solutions. (Loevinger &
Wessler, 1970, p. 7)
32
The above descriptions of EL indicate that the
dimension of concreteness-abstraction is inadequate on its
own as a discrimination of EL. For instance, some low-
level responses are vague, less precise versions of higher
levels. Advancement in these cases is marked by a
movement away from superficial terms (My mother and I—
"are said to look alike") to rather global terms (— "are
alike yet different"), and finally to specific renderings
("don't get along in our personal lives— only in
business"). Thus, a Conformist response to the stem,
"When she thought of her mother, she— " may mention
thoughts of "many good things," in contrast to thoughts of
"home, laughter, and good food" at the Conscientious EL,
in which elaboration has become more concrete and vivid
(Loevinger et al., 1970).
Another example of increasing specificity and
concreteness occurs in certain completions of the stem,
"At times she worried about— ." The vague response, "— my
sister," is an indication of the Conformist EL, one whole
step lower than the Conscientious response, "— my sister
on dates." Other responses to this same stem, however,
increase in abstractness. Compare, for example, the worry
about "money" at the Conformist EL to the worry about
"financial problems" at the Conscientious EL (Loevinger et
al., 1970).
33
Making such distinctions in both directions is well
within the intellectual capacity of most subjects, asserts
Loevinger, yet such spontaneous explicitness is not
ordinarily found at lower levels. For example, people
from all levels respond to the stem "A woman should
always— " by advocating compliance with a woman*s role.
At lower levels, however, the subject chooses first to
act, and then to feel like a woman; only at the Autonomous
EL is there a recommendation to look, act, and feel like a
woman.
Definition of Cognitive Style
The historical origins of the cognitive style
construct parallel the history of cognitive psychology
itself. Kogan (1971) has offered a general definition
that has guided much of the research in principle, if not
in practice: cognitive styles are individual variations
in modes of perceiving, remembering and thinking, or
distinctive ways of apprehending, storing, transforming,
and utilizing information. They differ from abilities in
that they give greater weight to the manner and form of
cognition rather than to the level of skill, or the "more
or less" of performance. However, investigators outside a
psychometric background have sometimes used a style label
for processes with an accuracy criterion, such as FDI.
Even when cognitive style performance is not based on
34
veridicality, constructs conceived to be bipolar often
carry value judgments imposed on either end. For
instance, when investigators observed individual variation
in preference for analytic and non-analytic groupings on a
sorting task, they placed greater value on the former
(Kagan, Moss, & Sigel, 1960, 1963). Although Kogan (197 3)
posits a third, intrinsically stylistic cognitive style
with no value judgment regarding directionality, it seems
unlikely that such conceptions remain value-free as
construct validation proceeds and extrinsic correlates are
examined.
The bulk of research on cognitive styles investigates
the age prior to young adulthood. Age comparisons that
have been pursued often find a score decline after young
adulthood in the direction of childhood scores. However,
the interpretation of these scores may well reflect an
ambivalence toward functioning in middle and late
maturity. Low scores are often attributed to regression,
without arousing research questions regarding possible
differences in the processes that produced similar scores
in different groups, or the different effects of test
conditions such as speed.
Later in life, a person secure in certain abilities
may be more willing to try out alternative response
options that offer maximal challenge, short of failure
35
(Kogan, 1973). Subjects may well formulate a personal
definition of their own capacity that changes with age.
But a cognitive disposition that is currently preferred to
one that has been replaced cannot be manifested on a test
with no response options, designed to register the latter.
Moreover, if subjects were capable of immediately
switching cognitive styles, then the notion of style
itself in terms of relative rigidity or pervasiveness
versus flexibility would be called into question. Taken
to extremes, an assumption of flexibility/availability
contradicts the idea of style, which seems to refer
primarily to a certain strength of preference over
flexibility of preference. If low scores on a cognitive
style measure represent "errors," then style in effect has
been treated as equivalent to ability, that is, as either
accessible or absent: the subject did not exhibit the
style because she or he was unable to do so. Using
response speed or accuracy as scoring procedures further
confounds the supposed distinction between style and
ability.
Kogan (197 3) points out that years of testing
patterns of stability and decline in the abilities have
not yielded an answer regarding the impact they have on
individual life experience or the ultimate value of
cognitive test scores in later portions of the life span.
36
Such decisions probably rest in part on value judgments of
what is lost or gained. Therefore, the current study of
cognitive style will not depend on an ability-style
distinction, but will consider cognitive style to be
subsumed by the analogous definition of personality as
"the various styles of behavior that different individuals
habitually reflect" (Rychlak, 1981, p. 2) . When
personality is taken as the larger frame of reference,
capacity-style distinctions become relative and moot, as
certain abilities come to be preferentially developed
through an individual's consistent manner of organizing
information. It is also possible that what develops first
as a capacity becomes a matter of preference later on.
Some support for this last proposition comes from two
separate studies (Kagan et al., 1963; Wachtel, 1968).
Fourth-grade children in the first study and college
students in the second study were given the same figure-
ground paired-associates task as a criterion measure of
analytic performance. A triads measure was also
administered, in which subjects may choose an analytic
sorting versus other sortings. The triads procedure
predicted performance on the analytic learning task for
the fourth-graders only. For the college students,
performance on the criterion measure was predicted by an
index of analytic capacity (the EFT) rather than analytic
37
choice. In other words, the option to choose analytic
concepts on the triads task did represent a capacity for
the ten-year-olds, but more of a stylistic preference for
young adults (Kogan, 1973).
Field Dependence-Independence
Witkin began his professional career as research
associate to Wolfgang Ktthler, one of the major Gestalt
psychologists, who was studying the effects of the
perceptual field on perception. In the mid 1940's, he
began to work with another eminent Gestalt psychologist,
Max Wertheimer, taking over the direction of their
research program upon Wertheimer's death (Messick, 1980).
In his earliest studies in space orientation (e.g., Asch &
Witkin, 1948; Witkin & Asch, 1948), Witkin and his
colleagues unexpectedly discovered that people differed
markedly in the manner of establishing the upright, and
that these differences were consistent in performance
across many similar orientation tasks. In everyday
experience, the vertical direction in space is
perceived by the framework of the visual field coinciding
with the bodily sensation of gravitational pull. When
these two referents are experimentally separated, reliance
on one or the other results in different locations of the
upright, and it is this feature which first revealed the
existence of individual differences in mode of
38
orientation. In the body-adjustment test (BAT) and the
rod-and-frame test (RFT), the visual framework was tilted;
visual cues had to be ignored to achieve a relatively
accurate adjustment of body or rod to the gravitational
vertical. In the rotating-room test (RRT), the force
acting upon the body was altered; in this case, reliance
on the body led to relatively inaccurate adjustments
(Witkin et al., 1954/1972). The tendency to rely
primarily on the external field was labeled field-
dependent (FD), and primary reference to the body was
labeled field-independent (FI).
Further study of the nature of the orientation tasks
convinced Witkin that they likewise involved the
separation of an item from an organized field. He
developed the EFT (Witkin, 1950; Witkin et al., 1971), in
which a simple figure must be located within a complex
design. A later modification was the GEFT (Witkin et al.,
1971), designed to facilitate group administration.
People who were labeled FI were said to find it easy to
overcome the influence of the organized embedding context
and locate the simple figure in the test items, whereas
the reverse was true for FD subjects. The FDI dimension
came to be conceptualized as a pervasive perceptual-
analytical ability. To perform adequately on Witkin's
tests, it was proposed that the individual must perceive
39
the environment in a discrete fashion to separate one item
from the total configuration. Using analogous
terminology, it is assumed that the FI person can withhold
attention from irrelevant items, whereas the FD or
relatively undifferentiated person cannot withhold
attention from the complex designs camouflaging the simple
figures.
Global versus Articulated Cognitive Style
Continued research broadened the scope of the
individual differences that Witkin was tracing. First, he
asked whether subjects would approach symbolic material
(thinking) in the same manner as they approached an
immediately present stimulus configuration (perception).
He ascertained that certain kinds of problem-solving
tasks, reguiring the removal of an element critical for
solution and the subsequent restructuring of problem
material, were more difficult for subjects identified as
FD in perception (Witkin et al., 1962/1974).
Second, he considered disembedding or analytical
ability to be based on a relatively more active than
passive manner of dealing with the field. He further
hypothesized that the tendency to act upon stimulus
material would also enable one to structure experience as
well as to analyze it. When the field is already
organized, an active perceiver would be able to
40
discriminate items as discrete from their background and
to reorganize that field, but the same individual could
also impose structure on a field without any systematic
order of its own. This expectation was supported in many
studies of perceptual and intellectual tasks (Witkin et
al., 1962/1974). Analysis and structuring were therefore
considered to be complementary abilities characteristic of
an articulated cognitive style. Experiencing an organized
field as given and an unorganized field as diffuse was
considered to be characteristic of a more passive, global
field approach.
Psychological Differentiation
Witkin continued to expand his theoretical conception
and to test the extent of self-consistency over broader
areas of psychological functioning. He hypothesized that
articulated cognitive functioning was just one of several
indicators from various domains of an organism-wide
process of differentiation. With this conceptual step,
Witkin found himself in the areas of personality and
development, proposing differentiation as the overriding
conceptual agent for FI and its accompanying personality
characteristics. As delineated in Werner's comparative-
developmental framework (Werner, 1940/1964), the process
of development produces an increase in differentiation and
hierarchic integration. Differentiation is a structural
41
or formal property of an organismic system, rather than
content. Primitive, developmentally early behavior is
global, diffuse, with no articulation among different
areas of activity. As differentiation increases during
ontogenetic development, the state of the system becomes
less homogeneous and more heterogeneous. The amorphous
whole is transformed as discrete, specialized functions
emerge and parts become discrete, discernible units, which
tend to produce a more complexly organized, hierarchical
system. Werner's theory stated the direction of all
psychological development, including perceiving thinking,
learning , language, and feeling (Werner, 1940/1964?
Werner & Kaplan, 1956; Witkin, Goodenough, & Oltman,
1979) .
It therefore seemed probable to Witkin that greater
or lesser differentiation would be consistent across a
wide variety of an individual's functioning, and he found
links between cognitive style and areas such as body
concept, nature of controls and defenses, and sense of
separate identity (Witkin et al., 1962/1974; Witkin et
al., 1954/1972). To account for these associations, a
hierarchical model was eventually proposed, with
differentiation at the apex and its three major indicators
immediately below it: self/non-self segregation,
42
segregation of psychological functions, and segregation of
neurophysiological functions (Witkin et al., 1979).
Each lower-order construct was investigated
separately in extensive research programs (reviewed fully
in Witkin & Goodenough, 1981). According to Witkin,
neurophysiological segregation manifests itself in
hemispheric specialization, and it is possible that the
psychological components of the differentiation cluster
are influenced by it. The construct of segregation of
psychological functions was studied by examining the use
of structured controls over impulse expression and the use
of specialized rather than global defenses against
disturbing experience. Witkin characterized the self
nonself segregation construct by the existence of definite
boundaries between an inner core and the outer world,
including other people, whereas a greater connectedness
between the self and others exists in a less
differentiated system. It was found in several studies
that people who function more autonomously of others in
ambiguous social situations also perform better on the RFT
and EFT (Witkin & Goodenough, 1981).
Autonomy: Internal versus External Referents
The evidence from the interpersonal domain inspired a
reorganization and consolidation of terms within the
self/nonself component of the differentiation cluster.
43
Although the designations FD and cognitive style were
originally meant to be distinct, they had long been
confused by customary usage in the literature. Witkin
regained some precision by intentionally applying both
terms to a new superordinate construct, autonomous
functioning in perception of the upright and in
interpersonal relations. Consequently, the combined terms
field-dependent and field-independent cognitive styles
reinforced the original research interest in the reliance
on self versus field, and aligned the concept of style to
its common meaning of manner of approach toward a goal,
rather than goal achievement. FDI was considered to be "a
pervasive feature of personality, and not of cognition in
the narrow sense" (Witkin, Moore, Goodenough, & Cox, 1977,
p. 15).
Cognitive restructuring ability as measured by the
EFT became a subsidiary of the FDI cognitive style
construct, regarded as a content variable in relation to
the process variable of style. Cognitive restructuring
was considered to be a skill through which the relatively
autonomous information-processing style of FI is
expressed. A person functioning with less autonomy, i.e.
more FD, would more likely adhere to information as given
rather than go beyond it, and would therefore obtain a
relatively low EFT score. However, the less autonomous
44
person's openness to external sources of information might
also stimulate the development of superior interpersonal
competencies, a derivative of greater connectedness with
others. Different cognitive styles, then, were said to be
associated with different sets of abilities and interests
(see, e.g., Witkin, Moore, Goodenough, & Cox, 1977;
Witkin, Moore, Oltman, et al., 1977).
Points of Comparison
Like Loevinger, Witkin has employed a single
comprehensive construct to explain individual differences
in personality across perceptual, intellectual,
personality, and social domains. Psychological
differentiation is a pervasive dimension of self-
consistency connected in its formation to the development
of the organism as a whole (Witkin & Goodenough, 1981), of
which FDI cognitive style is a major indicator. Witkin's
differentiation hypothesis addresses a range of individual
differences similar to Loevinger's developmental ordering
of levels in some respects, and his notion of self-
consistency is analogous to that of ego stability as
elaborated by Loevinger.
Many aspects of ego structure do seem to become
increasingly differentiated in successive levels of ED.
The individual continually shapes a separate identity,
marked by increasing internality and autonomy. For
45
Witkin, the segregation of self from nonself establishes
the boundaries necessary for an internal core of
characteristics and standards. An internally derived
perspective can be maintained in the face of contradictory
perspectives and relied upon more heavily than external
support or guidance (Witkin et al., 1979). A separate
identity, then, necessarily entails autonomy and
internality. In Loevinger1s terms, FD and FI are
different frames of reference.
Loevinger assumes that differentiation will be
manifested in the thought structure of verbal responses to
the SCT. Witkin assumes that psychological
differentiation will be manifested in the cognitive
restructuring of items on the GEFT. Each theorist, then,
proposes a substantial link between cognition and
personality functioning. Indications of their respective
constructs are also manifested through impulse control, in
a progression from diffuse anxiety and impulsivity to
structured controls and extensive coping ability (Hauser,
1976; Witkin et al., 1979).
Loevinger and Witkin share similar influences upon
their respective intellectual endeavors. For example,
both refer to Werner (1940/1964) as the originator of
their structural-organismic orientation. Like Loevinger,
Witkin also found parallels between his construct and the
46
authoritarian personality (Adorno et al., 1950), which he
considered to be support for his hypothesis of a
consistent relation between perception and personality.
He proposed that the authoritarian syndrome1s adherence to
external values, respect for authority, and intolerance of
ambiguity corresponded to the relatively dependent
attitudes and global perception of FD (Witkin et al.,
1962/1974).
Witkin considered the association of FDI with
psychoanalytic ego psychology to be an important direction
for further conceptual development. In the predominant
ego psychology at the time (e.g. Hartmann, 1939/1958), the
ego was defined as the seat of the defenses and a
mediating apparatus for the various functions of
perceiving, thinking, and learning. Witkin reasoned that
the ego functioning responsible for the articulation of
experience could be expressed in the FDI dimension, and
that there would be more or less differentiated egos
(Witkin et al., 1962/1974). The parallel to Loevinger1s
work is evident, though her definition of ego leans away
from an omnibus of functions towards an inherent
propensity for structure. Taken together, the various
points of similarity between the two constructs suggest
that an individual's standing on the FDI dimension could
be an index to EL.
47
Accordingly, a preliminary study was designed by this
investigator (MacDonald, 1986) to test one major
hypothesis and one related hypothesis about FDI as an
indicator of EL. The major hypothesis was that the
movement from FD toward FI would increase along with
successive levels of ED in a student sample with a
heterogeneous EL range, and this prediction did receive
moderate support. Contrary to expectations stated in the
related hypothesis, however, a homogeneous postconformist
ED group did not reveal the greatest amount of FI by a
disproportionate loading in the upper quarter of the
normative GEFT score distribution. Instead, their mean
GEFT scores were even lower than the mean scores of the
student comparison group computed at comparable ELs. In
fact, the average GEFT score for Self-aware comparison
subjects exceeded the mean score for postconformist ED
subjects at all ELs. Postconformist group GEFT scores
actually declined as ELs advanced, although there is some
indication of beginning decline in the comparison group at
higher ELs, as well. In the following section, the FDI
construct is critiqued in order to develop an explanation
for nonsignificant GEFT results in the postconformist ED
group.
48
Critique of the FDI Construct and Preliminary Conclusions
Dimensional Characteristics of FDI
Continuity and discontinuity. Although the
expectation of the MacDonald study (1986) was that the
GEFT would represent the fundamental continuous movement
from one EL to the next, data from both groups taken
together give the suggestion of FDI curvilinearity.
Should curvilinearity in fact be the case, Loevinger1s
explanation would point to the transmutation of structure,
whereas Witkin would have to suggest the occurrence of
psychological regression, because his is a bipolar
dimension. In fact, he has referred to this possibility
as a general "dedifferentiation during aging" (Witkin et.
al., 1979, p. 128), following the perspective of Werner
(194 0/1964) who also proposed regressive age changes over
the life span.
According to Kogan (1973), various studies using the
Witkin indexes of cognitive style have reported a distinct
U-shaped form. The performance of children and aged
subjects are similar, and both differ from young adults.
Kogan (1973) cautions that the decline in scores may prove
to be generational cohort-specific phenomena relating to
education and other demographic dimensions rather than
signs of developmental change. The explanation of low
scores by a cohort effect hypothesis, however, still views
49
low scores on a salient dimension to represent a lack, if
not a loss, of skill. It does not allow for the
possibility that mature individuals do exhibit
developmental change by replacing the measured dimension
with some other factor that is not being measured. In
other words, there is no room for the insight that
Loevinger describes as significant to her own professional
development— that the conceptual extreme point on a
continuum may be shifted to a developmental midpoint
(Loevinger, 1979c).
Witkin did in fact discuss development, but he
conceived individuals as moving along the FDI cognitive
style dimension in a decidedly polar progression from
dependence to the end-point of independence, in accord
with the organismic concept of increasing differentiation
and specialization. When FDI is placed within a larger
developmental context, however, a different emphasis
emerges, for the construct acquires pre- and post
manifestations as the ELs have done. Yet if something
else does succeed or replace the FI extreme, neither the
FDI cognitive style concept or the GEFT is designed to
grasp it, whereas there is room in the ego stage-typology
for multiple kinds of character patterns to emerge. So FI
could be a milestone manifestation of the more fundamental
variable of ED, increasing from FD up to a certain point
50
and then tapering off. The data from the FDI-ED study
(MacDonald, 1986) indicated a turning point near the
Conscientious EL, after which the GEFT scores for the
postconformist ED group decreased, depicted in Figure l's
plot of the study's mean GEFT scores. At the same
interpolated location on the comparison group line, rate
of increase appreciably diminished, as well.
The difficulty that emerges is one of aligning the
developmental narrative of qualitatively different levels
(i.e. milestones) in progressive development with a single
dimensional approach (i.e. the FDI polar variable). On
the measurable surface of the ED construct, sought-after
indices of EL most often turn out to be milestones.
Efforts to produce polar variables most often require the
inference of an underlying continuity, which can not be
measured directly. Although the ELs do consistently
increase, following the fundamental logic of building upon
their predecessors, many of their measurable
characteristics decrease along the way. In contrast,
polar manifestations of abilities like cognitive
restructuring are observable at a minimal inferential
level (Loevinger, 1976), and a deeper level of inference
is required to produce milestones or discontinuities, as
Piaget (e.g. 1936/1952, 1970) has shown.
51
Bipolarity and unipolaritv. When the GEFT was first
scored by mean log time, it measured field dependence
(more time = more FD). When the scoring changed to an
accuracy criterion, the measure became one of field
independence (more accuracy = more FI). The GEFT is
therefore unipolar in terms of increasing articulation,
but researchers have presumed field dependence to be
measured by default. The use of a default construct in
FDI theory reflects a broader problem in psychology
(Missler, 1986). To illustrate, global and/or synthetic
processing has not yet achieved a distinctive definition
in experimental cognitive psychology, either. Thus the
processing of a simple whole (in contrast to serial
processing) is defined "when reaction time does not vary
as a function of the number of parts" (Garner, 1979, p.
140) .
In Witkin's research, the default construct carried
over the theoretical dialectic to a measure designed to
tap the presence of only one style. The logical flaw
proceeded as follows: FD subjects obtain low scores on
the GEFT; this particular subject obtained a low score;
therefore, this subject is FD. Of course, by "affirming
the consequent," i.e. low scores, other alternative
hypotheses for their existence are neglected. No
"presence" is established— only an "absence." Many
52
possible conditions erroneously labeled FD could
conceivably be "there." As it stands, no theoretical
proposition about field dependence can be falsified on the
basis of the GEFT.
In effect, the conclusion of FD by default
methodologically entrenched errors that had already been
made conceptually. In the beginning stages of his
research, Witkin recognized an analytic-synthetic
distinction when factor evidence indicated a separation
between part-of-a-field and field-as-a-whole tests (Witkin
et al., 1954/1972). He also drew on the original research
of Thurstone (1944) and the subsequent work of his group,
in which two factors termed strength and flexibility of
closure eventually came to be referred to as synthetic and
analytic processes. In strength of closure, the subject
inhibits responding to irrelevant embedded items while
attending to relevant surrounds. In flexibility of
closure, the subject extracts items from the surrounds.
However, Witkin lumped the activity of both closures
together as a sign of FI manipulation of the field.
Passive registration of the field was ascribed to FD. In
effect, he applied a sensory Gestalt approach to the
definition of FD and a cognitive, logical Gestalt approach
to the definition of FI (see Rychlak, 1981).
53
Conceptual complications with bipolarity can be
exacerbated when additional pairs of terms are treated as
conterminous with the ipso facto bipolarity of dependence-
independence. Although Witkin lumped the closures
together for FI, most discussion and research of the FI
construct has centered on the disembedding imagery of the
GEFT. Thus, if independence engenders analysis, synthesis
as a naturally suggested opposite then becomes associated
with dependence, and all processing of wholes is
considered more rudimentary: "global” and "holistic" come
to connote "amorphous" and "diffuse." The process of
dialectical reasoning that produces a single theoretical
dimension can also present a temptation to continue
aligning other supposedly related bipolar concepts to the
original one. Eventually, shades of meaning can strain
the attempted parallels, one or more of the "opposite"
poles may begin to cross over, and the logic breaks down
into more contradictions as the implied dichotomies
dissolve. Thus, the procedural evidence for the FDI
construct is arguable.
Theoretical bipolarity cannot be assessed with a
unipolar measure, or limitations in the dialectic of its
procedural evidence remain undetected. The measure itself
must be constructed properly through a kind of unipolar
reasoning. As Rychlak (1981) states, "We simply have to
54
settle on a "this, and only this" . . . definition of our
concept" (p. 503). Such unidirectionality of thought is
achieved by the law of contradiction (Rychlak, 1981, p.
8): "A is not non A" . . . "something cannot both be and
not be." In other words, the only correct conclusion for
a low score on the GEFT is "no-differentiation," which
refers to existence-nonexistence rather than to the nature
of differentiation per se. The GEFT can only measure an
"articulated cognitive style" as Witkin defined it. By
itself, the GEFT can neither measure FD nor ascertain its
character, let alone identify the presence of anything
else. From a conceptual standpoint, then, each end of a
bipolar continuum must be separately operationalized with
diverging measures. Though the two unipolar definitions
appear to sum to bipolarity, the continuum will consist of
two-dimensional overlapping coordinates rather than the
ipso facto bipolarity derived through dialectical
reasoning.
Unidimensionalitv and multidimensionalitv. In the
process of theorizing about the nature of FD and FI
functioning, Witkin moved from a default definition of FD
(Witkin et al., 1979) to the inclusion of interpersonal
skills for FDs (Witkin & Goodenough, 1981), reasoning that
their limited self-nonself segregation would lead to an
adaptive orientation toward others. Given the subsequent
55
low magnitude of inverse relations between social skills
and cognitive restructuring ability, however, Witkin came
to the conclusion that both poles must be adaptive for
everyone, and therefore promoted individual "multilinear"
development "along both routes" (Witkin & Goodenough,
1981). Although the pertinent ability clusters do occur
in different domains, the cognitive styles that favor one
cluster over another are determined by the unidimensional
degree of differentiation. It is logically impossible to
get to both poles of a single continuum; to the extent
that people are more like one style, they are less like
another. It follows that the "interpersonal competencies"
described by Witkin seem to reguire dependence and
preclude independence, but Witkin supported this
involution by calling upon the fixity/mobility distinction
of Werner (1957), which accounted for regression in the
service of progression.
The FDI typology, though conceptualized as a complex
structure or pattern of traits, is nevertheless a
"degenerate" typology in the mathematical sense— defined
in terms of the two extremes. Intermediate cases are only
more or less of one pole or the other. A degenerate
typology is therefore formally equivalent to a single
trait dimension (Loevinger, 1976). Loevinger's typology,
56
however, proposes more than two types, one at each EL, and
is therefore in contradistinction to unidimensionality.
Autonomy and Interpersonal Relations
Because social and cognitive skills are opposed to
each other in the FDI model, people and ideas become
mutually exclusive concerns for those with FD and FI
cognitive styles. Basing his reasoning on the single
cognitive style dimension of increasing separateness,
Witkin defined what FD is in reference to a lack of
differentiation: "Interpersonal competencies [are] the
outcome of limited self-nonself segregation and, with it,
limited autonomy" (Witkin & Goodenough, 1981, p. 50).
Included among these competencies are paying selective
attention to social cues, maintaining ready access to
external sources of information to help structure
ambiguous situations (which FD individuals are purportedly
unable to do "on their own"), and "getting along" with
others (Witkin et al., 1979). Conversely, according to
Witkin (e.g. Witkin et al., 1962/1974; Witkin &
Goodenough, 1981), the self-reliance of FI subjects
fosters an impersonal orientation.
In contrast, says Missler (198 6), idealists contend
that increasing abstraction coincides with more knowledge
of the object, and therefore greater connectedness.
Witkinfs reading of autonomy seems based solely on
57
Werner's view of differentiation as maximizing self-object
polarity and becoming less stimulus-bound. He therefore
equates Mahler's separation-individuation model of
original processes in the beginning of life (Mahler, Pine,
& Bergman, 1975) to concepts such as self-reliance,
autonomy, and separate identity: they are "conceived as
identification by the individual of attributes, needs,
values he recognizes as his own and as distinct from
others; a developed sense of separate identity allows the
individual to function with a degree of autonomy from
others" (Witkin & Goodenough, 1981).
FDI theory offers a limited grasp of the self-nonself
distinction, confined by a strictly structuralist view of
differentiation. Within this framework, subject and
object are recognized as two categories, but that is as
far as it can go. It is not adequate for an understanding
of their internal constituents or of the vicissitudes of
this distinction throughout the life span (Blasi, 1983).
In ED, however, advance is characterized by an
increasingly rich and differentiated inner as well as
outer life, and by responses to both physical and
interpersonal aspects of the environment. Loevinger's
model presents self versus nonself as the first of many
issues to be successfully resolved before individuality,
autonomy, or identity can be achieved— accomplishments
58
which include enriched interpersonal relations. Several
additional possibilities of relationship come to light
within the ED model because it allows for a continuing
process of differentiation and integration. The social
behaviors described by Witkin seem comparable in many ways
to the interpersonal style and conscious preoccupations of
the Conformist EL in Loevinger1s model (belonging,
helping, superficial niceness, social acceptability— see
Table 1).
Differentiation and Integration
Overreliance on differentiation as a hypothesis led
to the neglect of integration as an additional and
necessary developmental feature of FDI theory. Although
Gestalt theory accounts for articulation of figures and
organismic developmental theory accounts for articulation
of parts, both of these occur always in reference to
wholes, so that parts are still related. Phenomenal
organization results in a differentiated and articulated
configuration and development results in hierarchic
integration. Hierarchic integration is in fact the
inescapable concomitant of differentiation in a normally
developing organism? to ignore its possible contribution
to either FI or FD processes is to limit the developmental
conception of both.
59
Witkin did acknowledge the existence of integration
and even viewed it as equivalent to adaptation (Witkin &
Goodenough, 1981). Unfortunately, this insight did not
lead him to question the dimensionality of his construct
or the assumptions of his measure. He proposed that the
actual effectiveness of integration, in the sense of
harmonious relations or good adjustment, was not
inherently connected to degree of differentiation,
pointing to an array of complementary pathologies at both
extremes of the FDI continuum (Witkin, 1965? Witkin et
al., 1979). In other words, an individual's personality
could be highly developed without necessarily being
accompanied by positive mental health. Witkin's
observation about integration is similar on this point to
that of Loevinger in reference to advanced ED. As an
attempt to establish the value-neutrality of FDI, the
point nevertheless fails to account for the specific
nature of the interrelationship of differentiation and
integration. From a developmental perspective, however,
differentiation can logically be considered to precede
hierarchic integration temporally, as elements are
discriminated before their recombination into new
compounds. By analogy, states Bloomberg (1967),
hierarchic integration "would appear to represent a higher
level of development than differentiation" (p. 129).
60
Various operations purportedly associated with high
GEFT scores and therefore ascribed solely to FI appear in
some studies of FD as well. The proposition of FD
interpersonal competencies suggests that FDs are active
structurers in their own right (Missler, 198 6): they are
said to exercise "selective attention" to social cues.
Factor analytic studies (Goodenough & Karp, 1961) found
distinct but correlated factors for FI and attention-
concentration, and the EFT did not load on the latter.
Therefore it appears that the FD tendency to close on the
larger Gestalt is also a "holding against distraction"
(Missler, 1986). Neither is there a relation of FD with
impulsivity as opposed to reflection, as had been supposed
because of the impulsivity-reflection construct's roots in
the analytic-synthetic distinction (Kagan et al., 1963).
In fact, there is evidence that speed (i.e. impulsivity)
in discrimination tasks is enhanced by the configural
properties of the stimulus rather than its features. A
related point is that the timed GEFT favors FI (Missler,
1986).
These findings suggest that passive registration in
the extreme cannot characterize all FDs; that certain
processes are shared by some subjects labeled FD and FI;
and that these shared processes consequently do not
pertain to analytical disembedding. The overapplication of
61
connotative associations to FI has both distorted what FI
is and overstated the case of what FD is not. The
suggestion of the current study is that integrative
processes represent the neglected phenomenon.
Cognitive Structuring and Restructuring
Witkin reasoned from differentiation to autonomous
functioning, i.e. independence from the field. Then he
associated FI with disembedding or analytical ability,
which required an active approach to an organized field in
order to "go beyond" information as given, isolate a
discrete part, and "reorganize" or "restructure." Witkin
realized that "attempts to act on stimulus material" would
also define efforts to "structure" inherently unsystematic
or unorganized fields, as well. Unfortunately, he decided
that restructuring and structuring were both complementary
abilities of one style (articulated or FI), without
considering the consequences of measuring just one
activity and treating the score as equivalent to both
activities. This last point refers back to the problem of
lumping both closures together.
On the GEFT, subjects are presented with an organized
field in a no-choice situation requiring an analytical
solution, although the "restructuring" label indicates
that Witkin included several overlapping constructs within
its scope (Missler, 1986). A more accurate and
62
comprehensive measure for the notion of style might test
for "structuring" in a free-choice format, in which a
presented stimulus could potentially be organized in a
variety of ways. Witkin*s presentation of a very cohesive
figure with directions to articulate is analogous to
playing a melody accompanied by directions to pick out a
i
note. If subjects had been given a somewhat less
differentiated field with similar articulation
instructions, the resulting range of responses from single
"notes" to "melodies" might have yielded both structuring
and so-called restructuring responses and richer data.
i
Kurtz (19 69) also suggests that time-occupying
achievements afford a better indication of character
traits or dispositions than single occurrences, as in the
case of a speeded test. These observations can be taken
as recommendations for the current attempt to locate and
measure ICS.
Theoretical Inconsistencies
Witkin applied the notion of biological
differentiation toward an understanding of problems of
perception in developmental terms. However, by mixing
traditions he also mixed metaphors in such a way that led
to questionable distinctions between FD and FI. j
Differentiation was used not as a universal principle but j
as a hypothesis, and this led to a characterization of FD
63
as undeveloped and relatively unthinking, whereas thought
was characterized as increasingly analytical in the FI
mode. Although the processes of perception and thought
are undoubtedly related to development (the Gestalt
psychologists considered human processes to be isomorphic
at all levels), the use of notions from one field may be
arguable when applied to the problems of the other. By
measuring one thing (perception) to imply another
(development), care must be taken not to confound formal
and efficient cause constructs (see Rychlak, 1981, p. 7,
i
for a review of these Aristotelian terms).
Research in cognitive psychology is inescapably
wedded to epistemological issues that function perpetually
as propositions and perspectives rather than "facts." On
a fundamental level, perception and development address
the conventionally distinct dimensions of space and time.
At issue also is the relationship of self with the object
perceived, and the relationship of parts to the whole
(reductionism versus holism). Definitions of processes
like analysis and "cognitive restructuring" are therefore
shot through with the possibility for ambiguity that even
the precision of an operational definition cannot totally
reduce. Because Witkin did not address these limitations,
his construct began to assume the character of a
"grandiose principle" instead of being guided by its
64
relations with other constructs (Cattell & Warburton,
1967, p. 108). Vernon (1972) notes the usefulness of a
construct that overlaps with other constructs, if the
construct of interest comprises a unitary, measurable
• factor. As Missler (1986) observes:
The breadth demanded of a style is to be
established by empirically linking a
construct with correlates. The generality
of a style is not to be gained by
broadening the construct itself, as was i
done with the differentiation hypothesis. |
(P. 24) j
i
Conclusions
As a result of the review of the background of FDI
and ED and the subsequent critique of FDI, procedural
evidence has accumulated to support the following
theoretical propositions as initial grounding for the ICS
hypothesis: \
1. Development and form do have a relationship, j
I
although the perception of form and the development of j
that form may be interpreted quite differently, depending
on what phase of the total process the investigator
focuses on. As has been shown, Witkin misapplied his
developmental metaphor. On the other hand, Langer (19 67)
has carefully considered the association of perception and
development, arguing "that an image of mind is that of a (
I
living process. . . . This vital appearance is exhibited,
i
of course, by living forms in nature, which— as that great ;
I
I
morphologist D'Arcy Thompson has observed— are almost all J
records of growth" (p. 152). Thus D'arcy's book is j
i
entitled Growth and Form (1942), in which he also states:
"We are here justified in thinking of form as the direct
j
resultant and consequence of growth . . . whose varying !
i
rate in one direction or another has produced . . . the !
final configuration of the whole material structure" (p. j
i
57) . I
i
Yet of still greater interest, however, is that |
Langer develops her argument by drawing some distinctions [
between the biological processes that produce living form j
and the perceptual and cognitive processes that apprehend
and produce images, through a close analysis of artistic
processes. In the following passage, she captures the i
full range of perceptual-cognitive possibilities with a
grasp that the partial developmental process of
differentiation can simply not achieve:
The principles of creation in art are not
those of generation and development in
nature; the "quality of life" in a work of
art is a virtual quality which may be
achieved in innumerable ways. Yet it is
in noting the differences between
biological exemplifications of living form j
and ways of creating its semblance in art ;
that one finds the abstractions of art
which emphasize the obscure, problematical
aspects of life that are destined to !
develop into or to underlie higher I
activities, felt as emotion or sensation |
or the spontaneous ideation that is the
intellectual matrix of human nature, the ;
mind. (p. 152)
Langer's conception will be referred to again in the !
formulation of an ICS definition in Chapter II.
2. An integrative cognitive style presents what the
FDI concept and the GEFT measure miss. Integrative
processes hold the promise of (a) encompassing the vision
described above, (b) resolving potential discrepancies !
i
caused when perceptual measures are employed to draw
developmental conclusions, (c) addressing the untenability
of an FD conclusion based on a GEFT score, and (d)
offering a theoretically superior account of the nature of !
I
higher developmental processes. This is a reasonable j
J
proposition because integration is a positive feature in j
developmental and perceptual theories central to Witkin1s
construct, and Witkin himself associated integration with
adaptation. Because it is proposed that all individuals j
I
participate in both analytic and synthetic or integrated j
processes to some degree, an overlap must be expected; the |
i
"style" aspect of a cognitive style would be indicated by
the relative strength of one process over the other.
3. Integrative processes represent a predominant
i
cognitive disposition at high ELs. Loevinger's theory, as |
i
well as the low scores of postconformist ED subjects on j
I
the GEFT in one study (MacDonald, 198 6), suggest this
possibility. The survey of the literature on integrative
processes in the next chapter will address this
67
relationship in more detail, as the definition and measure I
of ICS are explored and refined.
This cannot be an entirely straightforward task,
because the aforementioned dilemmas pertain equally to the
current endeavor, in terms of a certain philosophical
ambiguity and different usages of the concept
"integration" in different psychological fields.
Therefore, in order to delineate a construct that provides j
a reasonable degree of utility and confidence, varying j
I
trends in theoretical and empirical literature will be
surveyed for points of convergence, so that a definition
comprised of consensual agreement can be presented. Only
then can consideration be given to how this construct
might be measured. In the following chapter, integration
i
as it relates first to universal processes, then to j
narrower cognitive processes, and thereafter to creativity
will be reviewed. Common elements will be related to
literature on optimal functioning to prepare the ground
for a unified theoretical definition of integrative
cognitive style. Finally, an experimental measure will be ;
developed and the conceptual rationale for each of its
variables described in full.
I
I
CHAPTER II
The Construct and Measure of an j
Integrative Cognitive Style
Certain common terms and definitional distinctions
will be used in the following discussion. First,
integration refers to bringing together parts into a j
I
whole; this can refer to the creation of new wholes, or to I
I
the reconstitution of a whole that has been decomposed. j
Its opposing concept could be segregation, or j
I
differentiation and articulation. Most generally, these
words refer to a separation of parts. In the latter pair
of words, however, there is a sense in which the part is
distinguished from out of a more generalized amorphous
whole, as in developmental processes, rather than merely !
i
separated from the rest. Another slight contrast is ]
!
offered by the concept of analysis. which suggests a
complex whole being separated into its elements, as in J
decomposition. The counterpart to analysis, synthesis. is|
I
very similar to integration, as it also refers to the
combination of parts into a complex whole. Although it
would undoubtedly be possible to continue by elaborating
69 '
subtle distinctions between the concepts of integration
and synthesis, they would not be germane to this study, |
I
which attempts to define a cognitive style that unites j
!
more readily than it divides— that tends toward the whole j
i
1 rather than parts. As the following relevant literature i
i j
is reviewed for sources of elaboration of the integrative
style construct, the terms "synthesis" and "integration"
I
will therefore be considered equivalent. The overall |
i
purpose of this chapter is to delineate varied aspects of !
ICS and its measure by reviewing related broad and j
narrower constructs and also by comparing and contrasting I
I
these conceptual findings to the FDI and ED constructs |
I
where relevant. Two problems of particular pertinence |
i
will also be addressed in reference to the similarities |
and differences of the differentiation-integration concept
with two other bipolar constructs, concrete-abstract and
s imple-complex.
Relations to Broad Universal Processes 1
i
Phenomenology: Self and Object
At the beginnings of psychology*s development as an
i
independent science, experimental psychologists assumed
that their task was to discover general laws for human
behavior in the style of natural scientists— to look for
underlying explanatory principles at the level of basic
elements and physical energies. HUsserl (1965) regarded
70
this search as misguided, and developed a philosophy that
1
took its lead from the Kantian distinction between noumena i
(what something is in itself, outside of our sensations of
it) and phenomena (our sensory knowledge of things in the
external world). Kant claimed that we can never obtain 1
knowledge independently of the mind that frames in the j
meaning of the noumenal world through its categories of !
understanding (Kant, 1781/1961). It follows, claimed
Husserl, that the objectivity that scientists obtain by ;
j
subjective cognitions of things outside of themselves— |
knowing things from without— must necessarily be j
i
approached with a complementary method in psychology: the
study of consciousness as a phenomenological experience
from within. Consciousness is always a consciousness of
something beyond itself, in patterned relation to the
self.
i
Following in this tradition, Fuller (1983) has j
examined the essential participation and preeminence of
integrative processes in human experience in his article
i
entitled "Synthesizing the everyday world." He emphasized
that the wholes or unity of experience are achieved by
bringing its individual aspects to determinate
organization in experience. The aspects come to
constitute the whole out of themselves and the very
relativity of these aspects to one another. An experience
is an accomplishment of the meeting of this experienced j
whole or OTHER-side with the experiencing I-side. It
resembles the Gestalt concept of insicrht. defined by J
Ktihler (19 69) as "insight into relations that emerge when J
certain parts of a situation are inspected" (pp. 152-153),
or the grasping of how aspects objectively decide one
another. Insight includes the experience of being i
i
affected by the other, but also the active effort of the .
j
experiencing I-side to determine the experience of ;
i
objective relations. Experiences accomplish a continuing I
structuring and restructuring of life and one's capacitiesj
for experience. \
i
What comes to be actively synthesized in experience
depends on what a person has been concerned with in the
past, on one's current interests and intentions, and where
one wants to be later. A particular set of interests
initiates a particular encounter with a certain part of
the world and its objective relations; but there are
always alternative objective wholes that can be j
I
j legitimately actualized. Fuller calls this process !
I I
I historical selectivity and regards it to be basic to human1
l ;
life. Developmental differences reflect the extent of ;
I
such active synthesizing. Former synthesizings provide usj
familiar wholes that enter into current synthesizing in an!
unobtrusive, unconscious manner. At any given moment in
life, a person also has the capability to synthesize j
certain unfamiliar objective wholes, while certain other |
unfamiliar wholes are beyond reach. Present synthesizing
of an objective whole builds upon former expenditures of
time and effort, which cumulatively shape current
experience. |
I
I
For instance, says Fuller (1983), in viewing an |
unfamiliar terrain, we depend upon familiarity with some j
of its segments such as trees, paths, hills, and valleys,
in order to bring the terrain into organization as an
objective whole. Thereafter, our concern can be directed
toward the larger whole with which we are not familiar, |
I
and to which the recognizable segments stand as parts.
When current explorations are over, those former wholes
will contribute to the new whole, tree-path-hill-valley.
When this new whole is later recognized as familiar,
interest can then turn to other terrains. In terms of , i
i
r
Loevinger's ELs, whether or not a person notices a
particular tree on a journey, and whether the path's \
intersection with the tree is perceived in relationship,
i
is partially determinative of that person's EL.
In the phenomenological approach, we thus find an j
emphasis on integrative processes at the most fundamental !
!
level of immediate experience: subject and object are
integrated rather than segregated from one another, which
73
is impossible. Consciousness itself places us in an ]
i
!
integral relationship to the world. This orientation is !
}
of more than broad philosophical interest to the current
study, because it offers a direct contrast to Witkin's
i
characterization of a key distinction between FD and FI |
1
(e.g. Witkin et al., 1962/1974). Witkin reasoned that an
emphasis on subject-object relationships must be '
I
developmentally lower than the detachment, articulation,
and analysis of FI. For him, the enhancement of social
relations was evidence of the FD individual's diffuse
embeddedness in the field and consequent dependence on
others for information— a type of compensatory skill. j
i
In a study of perceptual processes entitled j
i
Metamorphosis. Schachtel (1959) also presented a |
classification that ran counter to the FDI distinction. !
He integrated data from diverse studies to formulate the j
autocentric and allocentric perceptual modes of
relatedness. Little or no objectification, or
separateness of the subject from the object, is present in!
the autocentric mode. Awareness is only of what the ;
subject is made to feel, not of what the object is like.
To an infant in this mode, any new stimulus is an
undesirable interruption of complete, protected !
embeddedness. Later on, stimulation may be pleasurable
and consciously enjoyed, but the feelings are still fused
I with the perceived sensory qualities of the object. In
I
the case of autocentric vision, the perceiver merely gazes j
at the total impinging field. Such descriptions of
autocentric perception document a global dependent quality
i
very similar to that of FD. !
The subsequent developmental step is unlike FI,
however. Allocentric perception is an active process of
attentive looking and centered (focal) attending j
constitutive of a unified, clearly grasped Gestalt. j
I
Several renewed approaches oscillating between passive-
i
receptive and selective-structuring-integrating are often
required to apprehend an object's totality, to consider it
from all sides and in all its aspects. A more accurate ]
and encompassing perception beyond the noticing of detail
i
results, though the object may then seem to nearly elude .
classification. |
Not only would the nature of the GEFT appear to j
disallow such an approach, but it could favor what !
Schachtel (1959) terms secondary autocentrism. The shift :
from primary autocentric to predominantly allocentric
perception is a normal one in childhood, establishing an
independently existing reality. Along with the emerging \
world of objects, however, develops an instrumental
attitude towards them; only their need-related aspects are
attended to. Interest is not in receptively looking at
the object but in looking for those partial aspects that |
serve the observer's needs and purposes and that j
correspond to preferred preconceptions— as when vigilant
scanning of the environment detects the presence or
absence of danger. Such an approach bears a resemblance
to FI.
Although the perspective of secondary autocentricity
is useful, its excess leads to stagnation and secondary
i
embeddedness in a closed pattern of life. For any true I
i
encounter with an object to take place, the embeddedness '
i
principle must yield to the transcendence principle of
openness toward the world. From the vantage point of |
t
Schachtel's model, then, higher stages of ED presuppose a i
retained or regained allocentric attitude, whereas the
apparent independence of FI is a kind of secondary
embeddedness, or perception of the object as an extension j
of one's own aims. If allocentrism were the customary 1
mode for people at high ELs, there would be no opportunity
to display it on the GEFT. Furthermore, as a speeded |
test, the GEFT does not provide much opportunity to
I
readjust to a relatively uncustomary mode of processing. !
' A low GEFT score could indicate either the autocentric !
i
I
global tendency of FD or the allocentric propensity for |
abstracting the essence of organized figures. J
Gestalt Psychology: The Whole and Its Parts i
The first area of study to be influenced by i
phenomenology was provided by Gestalt psychologists, who
acknowledged the role of phenomenological description in
their sensory investigations (Koffka, 1935; Ktthler, 1969;
i
Wertheimer, 1945). They attended to the totality of human j
experience as an alternative to reductionism, and
established empirical justification for a number of laws
of organization. Paying particular attention to the sense
of vision in their research, they regarded the field of |
vision as organized into a molar totality or Gestalt. I
Although the field may be articulated or segmented into !
distinctive suborganizations, the whole is unique and thus'
transcends reduction to its parts. The same thing applies
in the aural modality, where a melody emerges as a total
impression and cannot be referred to its individual notes
for explanation. j
These relative properties within the field obey
certain laws of organization, which include the following I
(Rychlak, 1981): (a) The law of unit formation and
segregation states that uniform stimulation in the field j
produces unity and cohesion, whereas unequal levels of j
stimulation produce articulation or division; therefore, 1
the most cohesive form is generally the one most likely to
be "seen" and remembered; (b) the law of closure refers to
cohesive pressure to close an ongoing curved line into a ;
i
complete figure of some sort; (c) the law of good shape
states that the shapes that emerge will have the best
balance and symmetry, unlike a blob whose disbalanced,
uneven edges are difficult to picture or describe; (d) the
law of good continuation observes the internal tendency of
lines or contours to continue as they are— like the demand
placed on a melody to flow in the line of best
continuation; and (e) the law of proximity and equality
I
states that items that are equal or similar will be i
grouped together in the perceptual field. Under this law,
the perceiver tends to relate closely articulated subparts|
as one total Gestalt. As the distance between these parts
is extended, different totals, rather than subparts of one
total, will eventually be seen.
These individual laws were embraced by a law of
figure-against-ground formation (Koffka, 1935). A figure
is the content on which attention is fixed, in relation to
j the background or "ground" on which it rests. Together
i
j they comprise two segregated areas of the perceptual
j field, with the background enclosing and the figure being I
*
enclosed by it. The holistic orientation of the
Gestaltists is expressed in this conception, because the !
ground always orders the figure, and never the reverse.
The contour of a figure works inwardly to differentiate
78 1
what is perceived, and the organism differentiates and
articulates a figure out of an indistinct or uniform
ground. In Hofstadter's terms, this is "downward" or
"introactive" causality, where causes are seen to flow
"inward" from the wholes to their parts (Hofstadter,
1981).
By applying a Kantian approach at the level of the
sensory receptors, Gestalt psychologists detailed a very
different account of how things are known through the j
. . I
senses than that offered by an atomistic breakdown of the |
nervous system. They further postulated that the same
i
laws ordering things at the level of sensation order j
things at the central, higher level of cognition. Rychlaki
i
(1981) makes the distinction between the sensory
i
phenomenology of the Gestaltists and logical
i
phenomenology. which emphasizes conceptual understanding j
i
over sensory organization. The latter places relatively i
i
more emphasis on constructed meanings and on final .
i
causality than the formal cause of patterned I
organizations. In both cases, however, the process of ,
organization and integration is indistinguishable from the1
I
concept of differentiation. Each organized
(differentiated and articulated) configuration of subparts
can be seen within a broader context, and subparts can
alternatively be seen as separate figures on the ground of
the total configuration. In logical terms, an integrated
conceptual scheme is a developing organization that
becomes more complex over time, and an increasingly
comprehensive frame of meaning applied at any moment.
These formulations stand in contrast to Witkin's (e.g.
Witkin et al. 1962/1974), which attends to the process of j
articulation alone. j
J
General Systems Theory: Differentiation and Integration |
The Gestaltists emphasized the perception of pattern !
rather than sums of sensations, and concluded with a
holistic view of human behavior in general— the whole i
determines the nature if its parts. Austrian biologist
i
Ludwig von Bertalanffy (1928/1961) acknowledged the |
Gestalt orientation in his developmental model of human I
!
functioning. His analysis of embryonic development |
convinced him of the necessity of regarding the living
organism as an organized system. He believed that (
biology's basic task was to discover the natural laws by
which the parts of an organism are organized for the ’
survival of the whole. Nature itself evolves from simple
i
to complex organisms with a force exceeding that of
natural selection, which he identified as the organismic-
i
systems principle of organization, or anamorphosis
(Davidson, 1983).
80 :
Mechanistic science, Bertalanffy (1928/1961) j
asserted, will never explain the ability of living I
organisms to violate the Second Law of Thermodynamics
(i.e. entropy). Physical events move from available ;
energy to unavailable energy and from improbability of j
i
order to probability of disorder (batteries wear out, sandj
I
castles and mountains erode, lakes evaporate and I
disappear). In contrast, biological entities exhibit j
I
necrentropy. "the tendency toward increasing order and I
i
complexity, the movement from formlessness to form"
(Davidson, 1983, p. 77). The thermodynamics of an open
system maintains an organism's dynamic existence through |
I
continuous interaction with the environment: matter and ^
I
energy are exchanged, as food and oxygen are imported and i
i
carbon dioxide and nitrogenous wastes are exported.
Bertalanffy believed that organisms are "charged with form!
the way batteries are charged with electricity" and ;
concurred with physicist Erwin Schrttdinger that "what an
organism feeds upon is negative entropy" (Davidson, 1983, 1
pp. 81, 8 3).
i
A second important component of general systems |
I
thought was hierarchical order. Systems in hierarchical
order, according to Koestler (1964), are J anus-faced
entities in that they are wholes when facing downward and
parts when facing upward. In biology, a hierarchy is
81
l
formed from cell to tissue to organ to organism. As life
ascends the ladder of hierarchical complexity, both
progressive integration and progressive differentiation
occur, so that parts become both more specialized and
dependent on the whole.
For Bertalanffy (1928/1961), the living organism
represented the system par excellence because the parts
and the whole exist in a state of total reciprocity. In
contrast to Witkin's adaptation of the organismic
I
I
developmental model, integration is seen as essential and j
inseparable from the process of differentiation, j
comparable to Gestalt psychology's emphasis on the [
i
organization of wholes. Bertalanffy noted the
intellectual affinity of his organismic model with the
"humanistic" approach to developmental psychology of
Piaget (1936/1952, 1970) and Werner (1940/1964), and the |
I
depiction of the self-actualizing, creative personality by*
Maslow (1962).
i
Categories and Information Integration !
I
A category exists whenever two or more I
distinguishable objects are treated equivalently (Mervis &
Rosch, 1981). A relatively new field within cognitive
psychology investigates the structure of natural
categories— the correlated attribute structure of the
perceived world. Although British empiricists used
elementary sensations, almost anything has been used as anj
attribute in this research, including parts, physical
characteristics such as color and shape, relational
concepts such as taller, and functional relationships.
What is considered a category and what are considered its
attributes depend on the level one is describing, because
the same item (red or circular) can be what is to be
explained (category) or what is referred to as part of thej
explanation (attribute). In other words, the attributes j
I
themselves can be examined as a categorization problem. j
If category formation required the rational (
abstraction of what is essential and the ignoring of what
I
is inessential (e.g. James, 1890), then all examples of |
the concept should be cognitively equivalent. However, a j
i
growing amount of empirical evidence suggests that not allj
members are equally representative of their category. [
Gradients of representativeness have been found for color '
I
and geometric shape categories as well as semantic, !
i
linguistic, and psychiatric classifications (Brown, 197 6;
i
Lakoff, 1977; Mervis, 1980; Rips, 1975). The cluster of
shared attributes that determines category membership is
not perfect, and exemplars differ in the extent to which
they share attributes with category members. Rosch and
Mervis (1975) call this variable family resemblance, with
the highest family resemblance scores assigned to those
items with the most shared attributes. These items also
share few if any attributes with items in related but
contrasting categories.
One can predict that category membership of
representative exemplars should be established at a young
age, but that membership of less representative exemplars
will vary. This also suggests that an adult's j
i
categorizing criteria will individualize not only what is |
!
regarded as a basic level category but superordinate
categories, as well— depending on how "generous" the j
interpretation or clustering of correlated attributes is. |
t
This variability would show as different cognitive styles.j
Those individuals whose family resemblance category
I
structure is restricted to the most obvious or j
I
| representative exemplars might tend more to stereotyped j
thinking than those individuals whose categories permit
more variability and change. Some research has attempted
to demonstrate that category boundaries are nondefinite
(indeterminate), especially because poorer members of
I
categories are likely to contain attributes from
correlated attribute clusters of other categories !
(McCloskey & Glucksberg, 1978; Rosch & Mervis, 1975). An
alternative view is that cognitive representation itself '
is indeterminant and can change in meaning and structure,
especially in an interactive, contextual construction of
meaning (see e.g. Anderson & Ortony, 197 5; Rabinow &
Sullivan, 1979). Again, the extent to which a person
8"4“!
allows such a change process to influence cognitive [
representations might account in part for adult individual
differences in cognitive style (Mervis & Rosch, 1981).
Virtually all accounts of categorization involve
f
abstraction, which can be regarded as the creative
determination of those elements that are essential versus
those that are irrelevant in a situation, and the creation
of higher order information about the logical relationship!
among these attributes. The resultant categorical
information is used to classify novel items (Mervis &
i
Rosch, 1981). Presently, however, the relative 1
I
contribution of analysis (decomposition) versus more j
holistic processing of categories has become the subject j
of debate. i
i
In part, this general issue has informed an |
epistemological debate within psychology since its j
inception. James (1890), countering the classical j
i
conception of an explanation as the decomposition of the
thing to be explained into its elements, emphasized the
i
unitary nature of a complex thought. Messmer (1903)
distinguished synthetic and analytic types of perception
in reading, and Kretschmer (192 5) proposed the same |
typology for dissociative and integrative characteristics
I
I
in personality.
Richardson (1985) contrasted four competing
hypotheses about the role of decomposition versus J
-Q- g-
integration in the retention of connected sentences
expressing complex ideas. Constructive and imagery
theories emphasized the integration of information,
whereas tally and fragmentation theories emphasized the
decomposition of information. In three separate
acquisition and recognition experiments, his findings
supported a constructive approach, in which integrated
representations or holistic semantic ideas are created in
memory from the propositional content of individual
sentences. Concrete ideas did not show better integration,
l ,
than abstract ideas, as would have been expected in the |
i
t
mental imagery theory. \
I ;
Hock (1973) also investigated two fundamental modes
of processing that he labeled structural. i.e. processes j
i
that organize the detailed parts of a stimulus into a j
i
well-formed whole, and analytic, i.e. processes that j
decompose the stimulus into features. Utilizing a same-
i
different comparison task, he hypothesized that "same"
responses require the comparison of all the information ini
i
each stimulus and the structuring of the stimulus into !
well-organized units or chunks, but that "different1 1
responses depend on analytic processes, or the detection
of a single differentiating feature. Hock discovered
systematic individual differences in subjects in the
"same" condition, concluding that subjects may use both
structural and analytic processes to make that judgment.
Kogan (e.g. 1971, 1973) has addressed developmental •
and individual bases for judging similarity between j
objects and events. In the prototypical task, subjects
are presented with stimulus arrays and asked to specify
what goes with what and reguested to state the reason for
their groupings. Three styles of conceptualization have
been delineated (Kagan, Moss & Sigel, 1960, 1963), two of
I
i
which are based on similarity and one which rejects ;
similarity as a basis for grouping. The first is analytic!
(common stimulus elements); the second is categorical 1
i i
(stimulus as a whole is treated as an exemplar of a ■
i
class); and the third is complementary (a specific
function or thematic relation is attributed to the j
I
stimuli). These groupings have also been interpreted as ;
reflections of preference for analytic versus non-analytic
i
cognitive processing.
Most models of categorization include decomposition,
but debate the level of abstraction at which it can be
!
said to occur. They also include a holistic component.
These components are posited to combine in a variety of
ways. There might be a holistic processing stage, or
different levels of abstraction may require more analytic
than holistic mechanisms. The capacity to match figures
holistically may be a cognitive strategy that reveals
individual differences (Cooper & Podgorny, 1976).
Perceptual processing of figures has been shown to proceed
87 j
I
from local to global under some conditions and from global'
to local under others. In other perception models, a
visual scene is seen as a hierarchical network of
subscenes, and it is claimed that higher order properties
are processed before lower order properties. In
linguistics and artificial intelligence, the controversy
is largely an issue of which linguistic level to consider <
I j
] a whole and which to consider the elements (Mervis &
Rosch, 1981). I
Lashley (1951) animated much discussion about this
general issue nearly 3 0 years ago with his influential !
critique of stimulus-response chaining as an explanation
i
! for complex behavior. He argued that sequences of skilled
; voluntary behaviors such as speaking or playing a musical
j . . i
i instrument occur too rapidly to depend on the feedback of \
one element in a series to the next. Organization of ;
i
these behaviors must be in terms of larger chunks instead j
of "one element at a time," to allow an overlapping j
initiation of each step in the sequence. j
The concept of chunking has received much subsequent !
J
attention in cognitive research, and has been extended j
into inquiries regarding self-schemas. It has been j
proposed that the cognitive structure of the self j
i
influences the categorizing or chunking of information, in
a manner analogous to the larger categories formed by
chess experts about their playing strategies (Chase &
— 8
Simon, 1973). The self is thus an expert about itself. i
In a study by Markus, Smith & Moreland (1985), it was
found that people who were given a specific content
cognitive structure about the self divided a domain
relevant film viewed during an experiment into larger
units than those without such cognitive structures.
The view of the self as an individual's own cognitive
representation or category links information processing to
personality and development. It also leads to an |
expectation of finding individual differences within the !
i
universal processes described in categorization research. |
The information-processing self-system is close to j
i
Loevinger's definition of the ego, so one might expect 1
dynamic changes in the self-system and in cognitive |
processes like categorization to vary as a function of ED.
Perceptual and cognitive attitudes revealed on SCT
protocols at high ELs seem analogous to the highly !
i
practiced and coordinated behaviors addressed by Lashley j
(1951), in which larger units become the customary
I
vehicles for functioning.
i
Abstract versus Concrete Thought j
Could the hypothesized ICS be more simply accounted
for by general processes of higher conceptual thought?
Both processes appear to be alike in many ways. Following
in the Gestalt tradition, Goldstein and Scheerer (1941)
differentiated between "concrete" and "abstract
89 |
attitudes," regarded as pervasive, cross-sectional aspectsj
j
of the total personality. The concrete attitude is a
i
realistic, more or less automatic perceptual response to
the immediate claims that arise from a stimulus. The
subject is passively dominated by a definite organization
of objects, which thrusts itself into the foreground and
upon the subject at the moment of his or her surveying j
i
activity, and who subsequently surrenders to the total j
impression. This description is very close to that of FD,
I
a correspondence that is undoubtedly due to their common !
reliance on Gestalt principles. I
►
If a person initiates by conscious will a shift of
aspect within this concrete, phenomenal realm, however, he
or she exhibits an independence from the immediate
existential reality of the objects by regarding them from !
the point of view of "the mere possible"— which
constitutes the voluntary act of abstraction. Other kinds
of abstraction shift to conceptual frames of reference, in
further detachments from reality. Among the behaviors
i
that accompany the abstractive attitude, Goldstein and j
i
Scheerer (1941) list the holding in mind of various j
I
aspects simultaneously, grasping the essentials of a givenj
i
whole, breaking up a whole into parts and synthesizing
them, and abstracting common properties reflectively with
reference to a concept, class or category. They believe
the latter contributes to the understanding of analogies
and metaphors.
Goldstein and Scheerer (1941) point out that the
abstract attitude is not simply a combination of existing
lower functions in a next higher synthesis. It is an ,
emergent, generically different functional level from the |
I
i
concrete attitude, characterized by the factors of
l
conscious will, the more complex capacity of active 1
i
synthesizing, and the ability to shift and to perform a j
greater number of shifts. The authors stress that these j
qualities may seem to be present in the concrete attitude
j as well. However, there is "a decisive difference between
active synthesis, active shifting, and a passive, global
reaction to stimulus constellations or a passively induced
change in reactions which only overtly coincide with the
!
true characteristics of synthesizing activity and (
shifting" (Goldstein & Sheerer, 1941, p. 23). In terms of
the present study, there is a difference between the
i
global FD cognitive style as described by Witkin and the !
i
hypothesized ICS? they represent contrasting approaches to
wholes. As an example, if the partial conformities of
different objects are sensorially impressive, then a
phenomenal (rather than conceptual, i.e. actively
abstracted) grouping may arise spontaneously in response
to the field. In an object sorting task, subjects may
thus sort objects by the quality "red" only because it has
- -g _1 -
come into phenomenal relief, without any conception of |
"red" as a common denominator.
One can note a split between perception and ideation
in this formulation of concrete and abstract attitudes,
akin to the different emphases of sensory and logical
phenomenology described by Rychlak (1981). This dichotomy
is identical to Witkin's description of FD and FI, with
i
one critical exception: the two theories focus on
J
different processes as significant indicators of higher |
j
level functioning. For Witkin, it is analysis; for
Goldstein and Scheerer, it is synthesis. The position of
j
the present study is that, assuming a certain number of i
i
unimpaired adults do exhibit a relatively concrete or FD |
i :
j cognitive style, then the abstract attitude of Goldstein (
and Scheerer (1941) describes a more logically consistent ;
j
endpoint than FI. The distinction between a global and an
integrative attitude is missed by the GEFT, which may
I
1 measure some sort of developmental midpoint. The next l
I
question is whether ICS is then essentially equivalent to j
abstraction. I
Abstract versus Integrative Thought I
Although theoretical formulations indicate some
correspondence between the processes of abstraction and
integration (Goldstein & Scheerer, 1941; Mervis & Rosch,
1981), ICS as conceptualized in the current study is not
simply a duplication of abstract thought. Langer (1942,
9 2 j
I
1953, 1967) labels the common understanding of abstraction I
i
a generalizing abstraction. This is the abstraction
familiar to logicians and scientists, who abstract
concepts by a sequence of widening generalizations through
a process of discursive thought. There are other kinds of
abstraction, although this is rarely recognized, each with
a different admixture of analytic and synthetic approaches
to both content and process.
Langer (19 67) has studied these "numerous, more or
less incongruous, yet often intersecting principles of j
abstraction" (p. 197) because "abstraction is involved in 1
I
all symbolic functions . . . has so many meanings . . .
i
i
[and] is alternately praised and decried, according to '
what the word means to the speaker at the moment" (p. 1
!
154). Langer identifies a whole class of abstractions not]
attainable by generalization but only by presentational
symbols, such as those that comprise a work of art, and
labels this class presentational abstraction. In a •
presentational abstraction, the artist symbolically ]
i
transforms subjectively known realities into objective |
i
semblances, which acquire vitality through "indirect and |
I
subtle orders of abstraction: isolating, metaphorical, j
I
secondary, transcending and perhaps others for which one j
could invent suggestive names" (p. 157). The most
fundamental elements of creative work in progress seem to
be tensions, which by their very occurrence immediately
9 3 i
i
engender a structure. However, a work of art needs not j
just perceptually created tensions but an objective
presentation, as well. The latter is met by a principle
i
of abstraction inherent in perception itself— the Gestalt |
principle that organizes impinging sensations into larger
units (Langer, 19 67). Thus an isolative abstraction bears
resemblance to the articulation of a figure on ground:
I
Langer describes it as "abstraction by emphasis . . . in
i
which a visual structure may be seen at once as a whole |
i
and its parts as articulations of the whole " (pp. 166, :
169). The interdependence of analysis and synthesis in 1
I
such an operation is evident even in the term itself.
This Gestalt principle has intimate relations with ;
i
the first principle of dynamic, tensive structure, and the^
interaction of these two creative processes produces the j
i
life of every design. Intellectually they are conceived !
(
only by turns, but they are heard, seen and poetically j
I
understood simultaneously; this is one of the fundamental]
I
differences of artistic structure from discursive form
(Langer, 1967). Tracing the origin of abstraction in both
presentational and generalizing form, Langer (1942)
contends that "the unconscious appreciation of forms is
the primitive root of all abstraction, which in turn is
the keynote of rationality . . . we abstract a form from
each sensory experience, and use this form to conceive the
experience as a whole, as a 'thing'" (pp. 83-84).
_ j
A claim of identity between abstractive and
integrative processes in general would therefore be
i
misleading, especially because most tests and scoring
procedures designed to assess abstraction assume a
generalizing abstraction only. The process exemplified by
ICS bears more in common with the active, holistic,
complex presentational abstraction. This type of j
i
abstractive tendency, however, is ill-suited for high j
achievement on the GEFT. i
As discussed in the preceding section, concreteness !
is more associated with the "beginnings" of a generalizing!
I
abstraction, and with sensory-perceptual experience. Thus;
a theory may contrast concreteness with an antithetical j
definition of higher thought, i.e. with a quality unlike j
whatever processes immediate perception purportedly j
entails. Subsequently, higher thought is commonly treated}
as equivalent in all cases and respects to the mode of
generalizing abstraction. Processes that deal with !
concrete factors are "perceptually-bound" or "stimulus- j
bound." A problem arises when the conception of
integrative processes is not limited to progressive
hierarchical structures. How can active, complex
manipulation of concrete factors be accounted for— factors
that other theories favorably regard as "living," "real,"
or "expressive"?
Rychlak (1981) has delineated two contrasting models j
of theoretical explanation that provide assistance in j
overcoming the unwanted dichotomy of the concrete-abstract
distinction. He used the labels "Lockean" and "Kantian" |
i
in order to function as convenient reminders of the two j
differing outlooks: British Empiricism's view of mind as
a tabula rasa, a receptacle taking order in, versus
Continental Philosophy's view of a pro forma intelligence '
bringing order to bear. The Lockean or constitutive model!
is equivalent to Langer's (1967) generalizing abstraction.
Beginning from what is assumed to be most basic— that is,
the concrete details of reality as immediately given— (
I
singularity gives way to generality in a hierarchy of
simple to increasingly complex ideas. Simple ideas ;
I
(number "1") combine to form more complex ideas (numbers
"2," "3," etc.), starting from a triangular base on up to
i
its apex. Meaning proceeds upward to the highest levels j
I
I
of abstraction, which must ultimately be broken down to j
the lowest levels or "givens" to be understood.
In Rychlak's (1981) Kantian or conceptual model, an j
increasing abstraction of terms continues to build a j
triangular form as described above. At its apex, Rychlak j
places a pair of eyeglasses to distinguish the phenomenal
realm on the inside portion of the glasses from the
noumenal realm on the outside portion. Meaning begins and
proceeds from the highest levels of abstraction downward.
- - 9 6]
I
The ultimate "given" resides in the eyeglasses— in the
point of view that frames the ideas below them into some
kind of order. The essential facts cannot be found at the
lowest level of abstraction, the "concrete" level;
complexity and abstractness are not uniform. The
complexity of an idea depends on the number of connecting
links or meaningful ties that a given numbered idea makes
with other numbered ideas. A 1-construct has only one (
meaningful connection, a 2-construct has two meaningful i
connections, and so forth. Because it is irrelevant at j
which level of abstraction these interconnections take
place, an idea can be complex without being abstract. For
i
the purpose of the present study, these connecting links
can be thought to represent the process of integration.
i
In this conception, figures do not just subsume one |
another into ever larger figure-ground configurations, but|
also form links to other, separate figures in the field.
Whether one sees a web of relationships (i.e. an open
system) or a vertical organization (i.e. hierarchical
order) depends upon frame of reference— a cross-section ofj
a moment in perceptual space versus a longitudinal section]
i
of a segment in (developmental) time.
i
Relations to Narrower Cognitive Constructs !
In the study of perceptual disabilities and reading
processes, a distinction has been made between whole-
perceivers and part-perceivers. Whole-perceivers note
— — .................
objects in their entirety, whereas part-perceivers focus i
I
on details and miss the concept of the whole (Scruggs & I
Mastropieri, 1984). The distinction has been considered
an integral component of classroom learning. If beginning!
i
readers become overreliant on part-perception and fail to i
l
integrate perception of the letters with the word as a
whole, they may encounter reading difficulties (Mercer,
i
1979). Lerner (1976) found that part-perceivers tended to
color each arm of a single figure differently. [
Kirkcaldy (1987) has examined the differences between
introverts and extroverts on a visual retention test, j
which involved the memorization and reproduction of a i
i
display of geometrical figures of varying complexity. He i
I
found that the trait of neuroticism interacted with group I
membership, so that unstable (i.e. neurotic) introverts ;
i
were the least accurate of all the subgroups, and that thej
I
most accurate subgroup was comprised of stable introvert
subjects. Kirkcaldy concluded that personality variables
do appear to relate to form perception and to be
responsible for the formation of response styles.
Barron (1954) studied the complexitv-simplicitv
dimension, a bipolar factor which opposes a preference for
perceiving and dealing with complexity to a preference for
perceiving and dealing with simplicity. His attitude
scale deals largely with aesthetic preference, but his
i
work has suggested correlates in self-conceptions,
susceptibility to social influence, and sociopolitical
attitudes. In a study which compared the construct of
levelincr-sharpenincr to the aforementioned complexity-
simplicity dimension (Berkowitz, 1957), it was postulated
that individuals with strong leveling tendencies would
prefer simple phenomenal experiences. Holzman (1954)
first differentiated between levelers and sharpeners on j
the basis of their tendency to accentuate differences. In
leveling, an individual changes or forgets relatively morej
I
discriminating details of the original stimulus complex
(e.g. a story) over time than would occur in sharpening.
Berkowitz did find that leveling in a design and story
i
recall task was significantly associated with a preference!
for simple experiences. Notice that sharpening and '
complexity are therefore also significantly associated.
In other words, the converse of a leveling preference is
preference for yet another type of totality— one that is
not smoothed out but highly profiled.
The last several studies mentioned reveal a positive
valuation of the process of integrating details of a
stimulus, and a corresponding implied limitation of
analytical or part perception, especially when performed
apart from integrative processes. Higher scores and/or
more effective performance were often determined by
complexity and organization of parts. Yet the alternative
modeled by Berkowitz (1957), in which articulation of
■ 9 9:
i
detail is conceived as accompanying a configurational ■
approach to the whole, did not capture Witkin's attention, j
The global pole of his construct appears to share much in !
common with leveling-simplicity, in which details are I
i
assimilated and undifferentiated. On the other hand, the !
progression oh the FDI is away from the whole toward an j
increasing differentiation of its parts, reflected in the ■
i
, increasing analytical ability to disembed or isolate
! detail while ignoring the whole on the GEFT. Different
|
j value judgments seem to adhere to wholes and parts,
j influencing the research question and the scoring
procedure, i.e. what is measured and what is missed.
In order to investigate metaphoric thinking as a !
continuation of research in individual judgments of I
i
similarity, Kogan and his colleagues developed a test of
"visual metaphor" called the Metaphoric Triads Task
(Kogan, Connor, Gross, & Fava, 1980). The test assesses
"metaphoric sensitivity" by presenting the subject with a |
I
set of three pictures that can be paired in three |
i
different ways, only one of which presents a metaphoric
similarity. Metaphor refers most fundamentally to '
l
I
similarity in the midst of difference, but it refers to a
special kind of similarity, one that overrides
conventional category boundaries and brings together
objects or events that normally belong to different
domains.
■ rocrj
Langer (1942) has referred to metaphor as a kind of ;
"abstractive seeing." Metaphor involves connotative
meaning and therefore implies throwing a wider
categorical, or cross-categorical, net to form acceptable
groupings other than those typical groupings based on
denotative meaning. An appreciation of metaphor requires
'
j a form of reasoning that is not necessarily tapped by the |
j
j object-grouping and concept-formation tasks of
i
investigators whose interests lie in superordinate, j
logically-based categories and the development of formal ,
I I
| reasoning (e.g., Inhelder & Piaget, 1964; Vygotsky, 1962).;
I
However, this link between broadly-based metaphoric
| thinking and uncommon conceptualization suggests that the !
l
field of creativity may have produced findings relevant to:
f
ICS. |
Relations to Creative Thought I
i
Interest in the apprehension of similarity has I
characterized much research in the field of creativity, J
and the previous definitional summary of metaphor bears a j
resemblance to two types of divergent-thinking measures j
commonly employed in creativity assessment— i.e. fluency
and uniqueness. In the first, the subject is asked to
generate a multitude of possibilities appropriate to the
i
presented stimulus. Fluency is indexed by the number of ;
responses, which is enhanced by a greater tolerance for
marginally appropriate instances (Wallach, 197 0).
101]
Correlated highly with fluency is the uniqueness— i.e. !
statistical infrequency— of the responses. These two j
processes also relate conceptually to previously discussed|
t
markers of higher EL and dimensions of individual j
differences along the course of ED. 1
i
Complexity and Synthesis |
Various theories regarding creative activity have
i
j posited a special capacity for combining two or more i
I
different ideas, images, or objects. Mednick (1962) |
i i
I . . '
j proposed that creativity involves the combination of
| elements that are remotely associated to each other, and
i
; Koestler (1964) specifies the combination of two self-
i j
| consistent but habitually incompatible frames of referencej
as a creative act of bisociation. In Arieti's magic |
synthesis (Arieti, 1976), the creative individual combinesj
i
primary and secondary process into a tertiary process.
Barron (1957) stated that "the ability to synthesize . . .
complexity is the mark of the creative individual" (p.
130). Consider also that Albert Einstein told a j
creativity investigator that he was aware of "combinatory j
play" (Hadamard, 1945). j
Rothenberg (1986) examined the concept of artistic
synthesis by means of a process termed homospatial
thinking. a creative combination of separate elements
requiring their transformation into a new organization or
integration. To measure this construct, Rothenberg
102
prepared one set of superimposed (homospatial) slides and
another set in which a figural element from one slide was
pasted to the second in a distinct figure-ground
relationship. Visual art students in the experimental
group produced superior sketches after viewing the j
j superimposed slides as rated by a panel of expert judges, j
i
Rothenberg concluded that complex rather than additive |
combining leads to creative products. |
Raychaudhuri (19 65) studied the relationship of j
i
creativity to the complexity-simplicity dimension by \
comparing visual and musical artists with nonartists on
the Revised Art Scale of the Welsh Figure Preference Test j
(Barron & Welsh, 1952), a series of white and black !
1
figures ranging from simple geometric form to complex
pattern and free-hand designs. In agreement with factor-
analytic as well as other psychometric studies,
Raychaudhuri found that visual artists rated significantly
higher in their preference for complexity than the musical
artists, and all artists rated higher than non-artists. 1
I
Field Dependence-Independence
I
Bloomberg (19 67) addressed Witkin's omission of
integrative processes (e.g. Witkin et al., 1962/1974) by
noting that FDI tests failed to tap the capacity for
hierarchic integration, and he turned to research on
creativity for support of the hypothesis that, although
FDs are probably similar due to limited differentiation,
I — — .
I
FIs may show marked contrasts. Taking his cue from an I
! aspect of creativity research that emphasizes regression !
in the service of the ego (Kris, 1952), Bloomberg proposed
that some FIs can adopt either an analytic or a global
approach. His reasoning thereby equated adaptive j
regression with hierarchic integration. In other words, j
certain FIs are mainly analytical, and the rest are both 1
✓
analytical and integrative. Note that integrative
processes must necessarily be located only in FIs in this
conception, given that integration is defined by the I
incorporation of a global attitude into the FI repertoire.
Under these conditions, speculation about the possible
i heterogeneity of FDs is unnecessary. !
i
The contention that guides the current study is that -
I
Bloomberg (19 67, 1971) missed the very subject he had j
i
begun to pursue— integration— and arrived back at Witkin*s
bipolarity, in which the only two explanatory choices are
global and analytic functioning. He did not develop the
implications of his central thesis, that FDI tests cannot
measure integrative processes and that the only conclusion
possible from them is that a subject either did or did not
behave analytically.
The confusion may arise from those aspects of i
creativity theory that, although not antithetical to the
concept of hierarchic integration, exercise a regressive
pull, i.e. novelty or originality, and accessibility to
— — r o - 4 -,
primary process or fantasy. The value of so-called more j
primitive, global, and diffuse cognitive processes seems
further enhanced when they occur in a creative person who
can "relax . . . take time out to enjoy some passing
impression, or daydream . . . .1 1 (p. 136) .
\
Hierarchic integration, then, was equated with j
i
mobility as opposed to fixity, Werner*s original !
t
explanation for adaptive developmental regression (Werner,i
I
1940/1964). Bloomberg (1971) also hypothesized the j
presence of mobility, i.e. the ability to shift from ;
i
global to analytic functioning, for creative individuals. *
i Integration and creativity were therefore also regarded as;
I I
similar, and Bloomberg hoped that a mobility measure would
flush out integrative processes in FIs. Given the
conceptual assumptions involved, however, it is not
surprising that no significant relationship was found j
between creativity and FDI. Bloomberg (1971) concluded
that the operational definition of mobility failed to j
i
i
capture the essential aspects of integration.
Other similarly inconclusive findings regarding the
t
relationship between FDI and creativity may be traced to
conceptual and methodological problems (e.g. Bieri,
Bradburn, and Galinsky, 1958? Getzels & Jackson, 1962).
For instance, Noppe and Gallagher (1977) measured
creativity with the Remote Associates Test (Mednick &
Mednick, 1967), without explaining how this criterion was
105. ,
validly related to analytic perceptual ability. Neither
j
did they provide a rationale for aligning aspects of j
creativity such as mobility and originality to FI. j
I
Hemispheric Specialization *
i
Brain research on hemispheric lateralization (see <
e.g., Ornstein, 1977) distinguishes a sequential, analytic
»
mode of information processing from a more simultaneous, j
holistic mode. In neurologically intact individuals, most;
complex tasks involve an integration of these two modes, |
i
providing reciprocal control of thought and affect, and a ,
sense of control and balance to normal cognitive and
i
emotional life. Within this normal framework, individual j
differences are many. Some people appear to suffer a kind!
of cross-callosal inhibition, whereas other people may
possess an overly agile interhemispheric line of
1
communication. Variations in interhemispheric interaction|
have been offered to account for impulsivity, !
compulsivity, mood swings, and creativity (Gabel, 1988;
Hoppe, 1989; Miller, 1988). j
I
Bogen and Bogen (19 69) emphasized a role for the \
\
corpus callosum in creativity. They proposed that both !
hemispheres initially develop their own processes without j
too much interference, and that the presentational symbols)
I
and imagery of the right hemisphere are then related j
i
i
through the left hemisphere, where they are transformed
through verbalization. Hoppe (198 6) termed this process
1
106
transcallosal svmbollexia. He believed the creative
moment to consist of synthesizing the two hemispheric
planes— applying the code of the left hemisphere to the
overwhelming number of possibilities of the right.
Accordingly, he called creativity "hemispheric
i
bisociation," after Koestler's previously noted concept of
creativity (Koestler, 1964). The conception of creative j
thought as a more highly integrated, transcallosal 1
i
process, rather than isomorphic with the right hemisphere |
alone, avoids the "neuromythologizing" (Hoppe, 1989) of i
right-brained (FD?) versus left-brained (FI?) ]
personalities.
Disordered Thought
i
Several studies support the similarity of test
performance on certain cognitive measures by highly
creative normal subjects and schizophrenics (Barron, 1972;
Dykes & McGhie, 1976; Woody & Claridge, 1977). For
instance, psychoticism as measured by the Eyesenck
Personality Inventory (i.e. the P scale) was strongly
related to divergent thinking in a group of 100 university
students (Woody & Claridge, 1977). In another study, a
principal components analysis of a wide range of
psychophysiological parameters known to discriminate
psychotics from normal controls were also significantly
related to divergent thinking tests (Claridge, 1972). It
has been argued that, although schizophrenia is strongly
hereditary, incomplete penetrance of the gene is common.
Thus, for every person who develops the disease, as many
as ten carriers remain undetected, comprising from 10% to j
2 0% of the general population (Dobzhansky, 19 64).
Researchers in creativity have therefore proposed that a
i
single thinking style may lead either to a controlled ]
usefulness or to an uncontrollable impairment. An unusualj
cognitive condition could manifest itself as originality
J
of thought or creativity instead of thought disorder, if j
certain ameliorative factors or moderator variables such
as intelligence or high ego strength were present (Barron \
i
& Harrington, 1981; Woody & Claridge, 1977).
Dykes and McGhie (197 6) noted that such studies often!
construe the observed similarities to be a reflection of
adaptive regression, i.e. creative individuals are simply
less restrictive and more spontaneous. Alternatively,
they proposed that one of the most frequently observed
characteristics of creative individuals bears close
comparison with one of the best substantiated theories of
schizophrenia: both populations deploy an attentional
strategy that samples an unusually wide range of available)
i
environmental stimuli. Dellas and Gaier (197 0) found that
creative individuals deployed their attention more widely I
I
and retained more prior stimulus experience in usable
form, tending not to screen out the irrelevant. This is
close to the description of a widening of attention in
non-paranoid schizophrenics (Lehmann, 19 66):
[such patients are impacted by a] higher
number of discrete sensory stimuli per
time unit of experience than most other
pathological or non-pathological
individuals. . . . If [the patient] is I
capable of coping with this greater than
average influx . . . he might perform at a
better than average level; but when the
extraordinary sensitivity of his receptive
apparatus is not matched by an equally 1
extraordinary performance of his central
processing apparatus then his integration j
breaks down and he may become psychotic. \
(Dykes & McGhie, 1976, pp. 51-52) |
Dykes and McGhie (1976) compared attentio’ nal
I
strategies of high and low creative groups with a 1
schizophrenic group. On the Lovibond Object Sorting Task,I
i
scaled for degree of conventionality, the high creative
i
and schizophrenic groups produced an equal proportion of
unusual sortings, which differed significantly from the
low creative group1s scores. The similar performances
were ascribed to loose associations. On the Chapman Card
Sorting Test, requiring a more convergent type of thinking
i
to arrive at the correct answer, both high and low
creative groups could adapt to the demand characteristics
of the task, whereas the schizophrenics made a higher
number of associative errors.
Mednick's theory of creativity (Mednick, 1962),
essentially equivalent to his theory of schizophrenia,
states that the flattening of associative hierarchies
results in unusual associations, and lends additional
' 1091
support to the preceding conclusions. However, it is not j
just the creative subjects* willingness to be j
I
unconventional that accounts for their high scores, but
the manner in which they organize the data. Thus, in a
t
multiple regression analysis of the relationship between j
the P-scale and ten creativity indices, Woody and Claridge j
I
(1977) found that 70% of the variance in P was accounted
for by the Similarities and Pattern Meanings tests from j
the Wallach and Kogan (1965) battery. Wallach (197 0) !
concurs that breadth of attention and the resultant
1 I
| ability to make wide associations are features of j
creativity, but proposes at least two other independent 1
I
factors: a preference for complexity and ideational
fluency.
Ideational fluency and a broad attentional sampling
strategy in turn share a resemblance to a process first
termed overinclusive thinking by Cameron (1939), who
performed pioneer studies in the field of schizophrenia.
His definition referred to the inability to preserve
conceptual boundaries, so that distantly associated and j
even irrelevant ideas come to be regarded as essential
parts of the concept. Overinclusive thinking can lead to j
overlapping concepts which would normally be mutually !
I
exclusive, so that the thought-disordered person
| entertains two mutually incompatible ideas at the same
i
time, without recognizing the contradiction.
Payne and Friedlander (1962) developed a three-test
battery to measure overinclusive thinking, which included
the Goldstein-Scheerer Object Sorting Test or OST (1941).
The test was originally designed to assess the presence of j
concrete and abstract attitudes, constructs that have been 1
discussed above. In the handing-over procedure of the j
i
test, a subject begins with one object as a point of ■
departure and is asked to hand over all the other objects J
that belong with it. Payne and Friedlander reasoned that 1
overinclusive subjects would tend to build up a more
I
inclusive concept around the starting object, thus |
selecting more objects, and so the total object count
constituted the overinclusion score. This score had the
second highest factor saturation on the factor of
overinclusion in Payne and Hewlett's 1960 study of thought
disorder, and a very low loading on the factor of
intelligence.
Investigators have subsequently used the OST to
address questions concerning thought disorder as well as
creative processes. When 50 schizophrenics were given 10
of Guilford's divergent thinking tests and an |
overinclusion measure as a traditional indicator of j
i
thought disorder, a significant negative correlation was
found (Al-issa, 1972). Andreason and Powers (1975) used
an elaborated system for scoring the OST (Harrow,
Himmelhoch, Tucker, Hersh, & Quinlan, 1972; Harrow,
! Tucker, & Adler, 1972; Himmelhoch, Harrow, Tucker, &
i
Hersh, 1973) in their examination of overinclusion as a |
conceptual style in creativity and psychosis. This |
scoring system provides for a qualitative evaluation of
overinclusive sorting behavior, including idiosyncratic
thinking and richness. The results showed highly creative!
t
writers to be overinclusive on the OST and to resemble a
I
manic comparison group more than a schizophrenic
comparison group. This finding is not surprising, j
according to Hasenfus and Magaro (1976), who suggest that ;
i
ideational fluency and overinclusion may be the same
i
phenomenon. Ideational fluency tests resemble the sorting
and classification tasks intended to tap schizophrenic
deficits, and are similarly scored (the more creative
individual gives the greater number of responses)— but
with different theoretical conclusions.
Overinclusion does resemble other constructs that
have been associated with normal and creative functioning.
Pettigrew (1958) has proposed a dimension of broad versus |
narrow categorizers, measured by a category-width task, in|
i
which each item offers a central tendency value, such as j
i
the average width of a window. The subject must choose |
the value of the upper and the lower boundary of the !
i
i
category from multiple-choice alternatives. According to |
Pettigrew, broad categorizers prefer to risk errors of j
overinclusion rather than overexclusion. Overinclusion isj
also close to what McConaghy and Clancy (1968) termed |
allusive thinking, which characterized the performance on j
i
object sorting tasks not only of schizophrenics but also, |
in less exaggerated form, of a sizeable number of a sample^
j
of normal students and their parents, interpreted as a
I
possible indicator of highly creative thinking. j
Additional evidence comes from Barron (1957), who has [
i
shown that the effort to include many aspects of the
i
Rorschach ink blot into a single, whole, synthesizing
image correlates highly with a composite originality j
score, whose constituent parts include numerous divergent
thinking tests.
Personality Correlates
i
i
Barron*s (1957) finding of the relationship between
the Rorschach response and an originality score emerged
from a comprehensive study of originality and its
correlation with a number of related constructs. He
defined originality as the ability to respond to stimulus
situations in a manner that is both adaptive and unusual,
independent of an intelligence factor. One group of |
variables that associated significantly with the ]
I
originality composite was interpreted to represent a i
"disposition towards integration of diverse stimuli" (p.
735), including the Rorschach response variable referred
to above. According to Barron, an integrative disposition
suggests an openness to a variety of phenomena, "combined
“1T3 !
i
with a strong need to organize these phenomena into some j
I
coherent pattern1 1 (p. 737, emphasis added). He elaborated|
this disposition as displaying resistance to premature j
closure, in combination with persistent effort to achieve
closure in an elegant fashion. Everything that can be 1
perceived must be taken cognizance of before a
I
configuration is selected as a final one.
Barron and Harrington (1981) note that creative
people are often perceived and rated by associates as more
intelligent than less creative people, even in samples
where no corresponding correlations between tested \
*
intelligence and creativity obtain. Beyond a possible
halo effect, this relationship may reflect personality
characteristics that facilitate the translation of raw
talent into more or less effective and socially impressive
behavior. Some of these characteristics resemble those of
high ED: autonomy, broad interests, attraction to
complexity, independence of judgment, and the ability to
j
resolve antinomies— to accommodate apparently opposite or
i
conflicting traits in one's self-concept.
Barron (1963b) described the creative individual's I
"exceptionally strong need to find order where none [
appears" (p. 160). The creative person willingly admits ^
complex and disordered perceptions in order to make order i
out of apparent chaos, through the use of current
abilities and past experiences. This creative attitude
overlaps with notions of high ED. Could it be that the j
i open and structuring approach of creativity is the force j
J
that precludes permanent equilibration at lower ELs? Such
a force may impel an individual through relatively
unstable transitions toward new levels of development. i
Optimal Functioning, Creativity, and Cognitive Style
Wexler (1974) viewed self-actualization as the degreej
to which a person characteristically engages in a mode of ;
i
information processing that both differentiates and j
integrates meaning. He found that the degree to which j
people distinguish and synthesize information is related j
to the degree of self-actualization, and suggested that
self-actualization may involve heightened complexity,
either of the construct system or of the rules for
transforming information.
Landau and Maoz (1978) found a positive relation of
creativity and self-actualization to positive coping
attitudes toward the process of aging and death. They j
I
equated the highly self-actualized person to the creative |
!
personality, who needs to experience itself in a changing, !
dynamic process and to get away from the details,
integrating the entirety of its situation into a whole.
Seeman (1959, 1983) has developed a theory of
personality integration, defined as an organismic
integration that maintains intersubsystem communications:
"Such a process has the very fundamental effect of
maximizing the amount of information available to the
person as a basis for decision and action" (Seeman, 1989, j
p. 1101). The emotionally mature person makes available ai
I
maximum amount of information and synthesizes the
I
information effectively (Seeman, 1959). Behrends (1986) |
reviewed several different strands of theory and research
pertaining to the hypothesis of maximal utilization of \
information. One idea that emerged is that high-
I
integration individuals may simply take in more |
information from the environment. People who exhibited ;
more conceptual elaboration were able to attend to j
significantly more information in a selective attention !
i
task (Thomas & Seeman, 1972). Another possibility is that|
the high-integration individual might hold more
I
information for processing in larger unitary
representations or chunks (e.g. Wexler, 1974). An
increased organization of information would also be
expected to increase its future accessibility.
Kegan (1983) has proposed a continuing process of
differentiation and integration as the underpinning of all;
movement from one ego level to the next. In contrast to
the common attribution of growth and development to
differentiation, and of dependency and immaturity to
integration, Kegan asserts that the evolutionary cycle
emphasizes each process in turns, engendering a continual
emergence from embeddedness of a new self. Each emergence
rr6'
i
! is accompanied by separation anxiety, but it also
i |
represents a victory over isolation, in its concurrent !
I
i
creation of a more extensive object to be "related to"
rather than "embedded in." Self and object therefore
compose a lifetime rhythm, punctuated by the
differentiation of the first and the expansion of the j
I
second.
Otto Rank ( 1 9 3 2 ) discussed the creativity of everydayj
i
life, arguing that individuals act on their life ■
i
conditions in integrative, creative ways in order to cope 1
i
with constant change and the challenge of shifting j
patterns of contingency. He considered emotional well
being to result from a creative synthesis of autonomy and
intimacy, reserving the term "artist" for the
i
psychologically healthy (Strickland, 1 9 8 9 ) . Creativity, j
1
suggests Strickland ( 1 9 8 9 ) , means "finding connections j
between seemingly unrelated issues— sometimes bringing |
order out of chaos" (p. 7 ) . I
I
A Unified Definition of Integrative Cognitive Style
The link of high-level functioning to creativity and
i
integration brings the review of the literature full
circle, inviting a final definition of ICS and its ^
i
relationship to postconformist ED (Loevinger, 1 9 7 6 ) . Key |
i
related concepts examined toward this end have included j
I
categorization and information processing, abstraction,
complexity, wholes, and metaphor, as well as creativity
” ii7 ;
and its connection to complexity, homospatial thinking,
I !
! remote associations, transcallosal symbollexia, and
!
overinclusion. The optimally functioning personality has j
been described as analogous to the creative personality, |
whose cognitive style appears to embrace aspects of all j
the above concepts. The overlapping research traditions i
i
represented in these various studies are supported by
i philosophical-theoretical convergence within j
t
phenomenology, Gestalt psychology, and general systems
1
theory regarding the significance of integrative
processes. !
The predominant correspondence among the various j
constructs and theoretical bases may be termed conceptual j
depth and breadth, to account for vertical-hierarchical as
well as "horizontal" integrations of disparate data that
are perceived to be related. Thus, an individual may
integrate data to build taxonomies of class inclusion and !
may also form cross-classifications. Such an orientation i
I
is necessary to encompass each of the reviewed concepts in;
I
a common definition. For instance, metaphors and remote
associates imply the far reach across conventional
boundaries, whereas certain kinds of categorization, |
i
abstraction, and the synchronization of subsystems imply a!
hierarchical organization. The perceptual grasp of parts
as a total configuration may alternatively resemble either
vertical or horizontal integrative processes, or both.
Another similarity among these associated concepts is;
that they invariably indicate the presence of some kind of'
i
deviation from a norm, whether it be in the form of ;
disordered thought or of creative thought. The latter has
|
been differentiated from the former most fundamentally by j
i
aspects of coherence and adaptiveness— thus creativity canj
be considered a resource instead of a deficit. Deviance j
i
in the creativity literature appears as divergent
thinking, originality, unusualness, and remote I
associations. All of these concepts in turn refer in some
I
way to the scope and relatively broad range of the j
individual's thought. Therefore, ICS would sometimes !
i
include statistically rarer elements because of its
tendency to grasp more comprehensive wholes. It is
holistic in the sense of bringing together separate,
i
differentiated parts into one organized relationship. In
summary, then, ICS can be defined as a decided propensity
to synthesize elements that are discrete and often
i
relatively disparate into complex wholes, resulting in j
j
comprehensive concepts of relative richness, depth, and
breadth.
i
Postconformist ED individuals share several !
j
characteristics of integrative and creative functioning as!
i
described earlier, strengthening the suggestion that their
information processing is also characterized by ICS. One
immediately apparent connection is that postconformist ED
' 1T9i
is intrinsically associated with nonconventional, j
i
unstereotyped thinking. Also, the construct of cognitive j
i
i
complexity contributes to the larger construct of ED, and ;
I
individuals at higher ELs provide more sentence |
i
completions that manifest complexity, integration of
i
disparate elements, and an ability to create novel J
i j
solutions to difficult problems. The qualitative shifts j
i
from one EL to the next described in Chapter I can be
j
viewed as the cumulative growth of integrative processes. |
!
For instance, recall that the Individualistic person !
combines thoughts that lower EL individuals choose as one :
i
alternative or the other, perceiving paradox rather than
forced choice. At Autonomous and Integrated ELs, some \
form of cognitive complexity appears consistently with I
richness, variety, and uniqueness; specific and general,
inner and outer, multiple possibilities and
contradictions, and opportunities for reconciliation are
all considered. In general, the increasing cognitive
complexity manifested as ELs advance testifies to the j
increasing importance of the ability to integrate !
experience into an ever-more-encompassing "thing.” j
Conceptual Dimensions of the ICS Construct
A clear conceptual scheme is needed to guide the
development of a measurement scheme of the ICS construct.
Problems associated with identifying the conceptual
extreme point of a dimension have been discussed earlier.
! The opposing pole of integration or synthesis is often
i
j considered to be differentiation or analysis. However,
]
Witkin (e.g. Witkin et al. , 1962/1974) opposed
differentiation to a diffuse, passive, "global1 1 perceptual*
process, eliminating the presence of integration from his 1
bipolar construct in effect if not in principle. |
I i
Depending on the point of view, another distinction j
to integration may be reasoned on the basis of its
categorical organizational properties, which could simply :
i '
be labeled "no organization" or "no integration." This
position has several advantages. First, it allows
!
integration to be regarded unambiguously as a unitary,
unipolar measurable dimension amenable to operational i
I
definition. It avoids the possibility of a faulty
conclusion based on low scores, i.e. the concept and
measure would not presume analysis to be present in its j
I
absence, as in the default conclusion of FD for low GEFT j
I
scores. i
I
The guestion of the specific dimensional relationship!
i
of analysis to synthesis is beyond the scope of the |
i
current study. It could presumably be explored by future j
investigations of factor structure: they may be opposite ;
ends of a single bipolar dimension or negatively J
correlated dimensions. However, previous factor analytic
studies related to this guestion have not been definitive.
Messick and French (1975) found the closures to merge at
I ' 121j
the second order, and other studies found them to load on
\
i
I a single factor, with no indication of bipolarity (Adcock
*
& Martin, 1971; Busse, 1968; Wilson, 1954). Katz (1972) ;
I
reviewed Quine's conclusion that there is no absolute |
dichotomy between analysis and synthesis, and that the
j problem may be one of empiricist thinking, which j
I . . . . '
bifurcates actualities for purposes of objective ,
: theorizing.
j In any case, even if analysis and synthesis (or j
l i
j . . . 1
| differentiation and integration) were to comprise a single|
continuum, and a person were considered as closer to one j
i
pole or another at any one time, the salient issue should j
be one of the propensity of style, not developmental (
regression. In other words, they are not considered to
replicate the entire developmental continuum. As
Loevinger (1976) has shown, any "degenerate typology"
i
representing two types cannot capture the
multidimensionality of the developmental narrative. Many
traits show an ascendancy at certain ELs and then recede j
as they lose their saliency at later ELs. The continuous ■
measurable surface of most traits does not coincide with
the fundamental developmental continuity. FDI may be
among the more limited "milestone" dimensions.
Characterizing the endpoint of ED, however, presents
a special case. Although some new qualities appear, a few
other traits must be considered a culmination of constant
‘12 2";
progression from the earliest ELs. Only a few measurable ,
i
traits are polar aspects of the underlying dimension of
ED, states Loevinger (1976), among them being the growth
of autonomy and its interpersonal context, inner
i
consciousness, realistic thinking, and cognitive
complexity. It is posited here that integrative cognitive)
i
i
processing is also a polar aspect of ED (though not a 1
J replica of the developmental continuum per se), in terms
of increasing integration. Reasoning forward, integration:
I
increases steadily along with cognitive development, as
transitions are negotiated through a "sequential process
I
of progressive decentration" (Missler, 1986, p.25; see
I
also Kohlberg, 1964). Reasoning back to early ELs, the !
absence of integration is well represented in the immature
extreme of the unreflective, chaotic character of the
i
Impulsive EL. The view of holistic perception as j
undeliberate and developmentally immature may come from
this and other early ELs (e.g. the stereotypy Conformist
EL), and parallels the FD account. Therefore, the :
integration dimension has the added advantage of actually :
representing FD in its low end (as defined by Witkin) j
rather than assuming it by default. At its high end, the j
dimension portrays a highly organized degree of holistic
' cognitive-perceptual processing. [
123 ;
Operationalizing the Integrative Cognitive Style Construct
Various measures referred to in the literature were (
examined and eliminated before selecting empirical j
j
variables that were congruent with the ICS definition and <
i
, I
capable of assessing a distinct integrated style. No
i
measures of memory or partial, related constructs such as j
r
uniqueness, remoteness, metaphor, or homospatial thinking :
were able to tap all of the particular aspects or the !
!
focus of the ICS conception. Related single bipolar
i
dimensions such as leveling-sharpening and simplicity-
complexity were likewise considered imperfect parallels to
integrative processing in general. Although promising in 1
some respects, creativity measures tend to assess
performance that may be the consequence of ICS, such as 1
originality or productivity.
As can be seen in the Chapter I critique of FDI, an
effective measure of ICS should present the subject with a
free-response, untimed "structuring" or organizing task ini
I
i
which typical performance is accessible to measurement.
Moreover, for purposes of alignment with the perceptual
orientation of FDI and its measure, an ICS measure should
also present the subject with a visual field, accompanied
by instructions to manipulate it in some way. Most ■
psychometric tests, observes Rychlak (1981), are based on
the assumption that a subject's regular behavior patterns
are being sampled, but only a few tests consider freely
1 " 124,
t I
i I
chosen alternatives to be the pool from which samplings
are drawn. The latter provision, however, would enhance j
the assessment of a cognitive style construct. Two tests |
i
were considered to meet all of the criteria described j
I
above and were therefore adapted to serve as complementary
I
measures of the ICS construct: the Goldstein-Scheerer O S T j
j
(1941) and the Rorschach. j
i
I Object-sorting tasks have the capacity to address key'
j I
elements of the integrative style definition, such as \
i
category breadth and the relation of parts to wholes.
Object-sorting tasks have been used across several
investigative fields relevant to the definition of ICS,
such as cognitive development, concept formation and
I
categorization, creativity, and thought disorder. If
commonplace objects rather than abstract ones (e.g.
i
geometric shapes) are used, subjects can be given an
opportunity to apply associations and experience to the
task, which may then more effectively elicit a subject's |
characteristic style.
1 The Goldstein-Scheerer OST
, - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
| As described earlier, the Goldstein-Scheerer OST was j
i i
originally designed to differentiate between "concrete"
and "abstract" attitudes, and was adapted later as a
measure of overinclusion. Overinclusion was demonstrated
to have relations with creativity and higher development
when not associated with the wandering, uncontrolled
12 5'
i
disorientation of thought disorder. As such, it is
i
j congruent with the construct of ICS in its similar j
disposition toward larger, integrated Gestalten.
To take the OST, subjects begin with one object and
are asked to continue sorting all other objects that I
I
belong with it. They must compare objects with others |
i
within a realm, decide on a group, and determine the |
membership and nonmembership of individual objects. The
conceptual groups can be rigid or fluid, narrow or loose j
(Rapaport, Gill, & Schafer, 1968). The subject is
confronted with numerous alternative objects that demand
i
the building and delimiting of categories from many
associative possibilities. Goldstein and Scheerer (1941)
observed that, without other evidence, a sorting could
i
indicate the presence of either the concrete or the |
abstract attitude. Because the objects are familiar, they
can almost always be sorted on the basis of passive
registration (akin to a FD style) and stereotyped,
customary concepts. The "safe" answer may serve to |
I
obscure a disorganization in the process of concept !
formation; likewise, the demand can lead to deviant
responses from psychotic patients (Harrow, 1987; Rapaport
et al., 1968).
For instance, sorting objects because they are red
could indicate a quality that has merely come into
phenomenal relief for the subject, who actually lacks any
— _ 12 6
conception of the common denominator shared among the
objects (i.e. the abstract attitude). Thus, the authors
stipulated that the procedure for sorting the objects be
observed and queried. Himmelhoch et al. (197 3) concurred
that a subject could conceivably arrive at a high number
of sorted objects by various paths. Low intelligence, a
propensity for concreteness, or bizarre or idiosyncratic
motives should be considered along with overinclusiveness
as possible contributors to the formation of vague,
inconsistent categories. On the other hand, a high number
of sorted objects may stem from "the ability to select
original, well-structured sortings through [intelligence
and richness of experience], and its resultant
associational material" (Himmelhoch et al., 1973, p. 3),
rather than from overinclusion in its narrow sense.
McConaghy and Clancy*s (1968) description of allusive
thinking in highly adaptive and functioning college
students and their parents lends support to this
possibility.
Payne & Friedlander (1962) advocated a quantitative
approach to the overinclusion construct by stipulating the
total number of items included in a sorting as its
measure. In accordance with Goldstein and Scheerer
(1941), Harrow, Himmelhoch, Tucker, Hersh, & Quinlan
(1972) stressed the value of a qualitative analysis that
would be geared to all levels of thinking, including
12 7 ;
I
features basic to both ordered and disordered thinking andi
i
perception. They therefore devised five scales to ,
i
supplement the object count (Behavioral overinclusion)
!
score, described in the Manual for Assessment of Selected j
Aspects of Thinking; Object Sorting Test (Himmelhoch et
al., 1973): (a) Conceptual overinclusion, (b) Bizarre or !
i
idiosyncratic thinking, (c) Richness of associations, (e) I
f
Concrete thinking, and (f) Underinclusive thinking. Based
on the descriptive criteria provided for each scale, threej
scales were chosen as appropriate for the purposes of this
study: Behavioral overinclusion (i.e. count of total
objects sorted), Conceptual overinclusion, and Richness of!
associations. The latter two scales are capable of
differentiating a vague, inconsistent, "hit-and-miss"
category breadth from a well-integrated, broad, and
consistent category formation (though the latter may at
times be unusually based). When a subject selects an
!
unusual and creative dimension that is also somewhat i
overgeneralized, or that requires the force-fitting of one
or more objects, the selection can be scored for both rich
associations and conceptual overinclusion. j
The Rorschach Test
The Rorschach is a procedure in which a subject is j
presented with 10 inkblots one at a time and asked, "What [
I
might this be?" Originally introduced into America by I
David Levy in the early 1920s, the Rorschach was welcomed
by personality psychologists who were struggling during
• . . . [
that time to conceptualize personality m a holistic j
manner, rather than the prevailing atomistic approach of
trait psychology. Attention was directed toward
tendencies to perceive either the parts of the stimulus or
i
the whole, to respond to certain perceptual determinants I
of the stimuli (e.g. form, movement, color, and shading), 1
I ;
! to perceive the stimuli accurately or inaccurately, and to*
be interested in particular content, such as human or
animal forms. In the ensuing decades, the refinement of
sophisticated empirical methods has offered greater j
precision and specificity in personality assessment with I
the Rorschach (Blatt, 1986). Additionally, interest has '
begun to center on constructive and representational j
I
cognitive processes and how cognitive schemes influence j
I
individual behavior. Correspondingly, the Rorschach has
i
been considered not just a "perceptual test," but an j
experimental procedure that systematically presents !
i
individuals with ambiguity and studies how they construct ;
i
meaning from it. The connection between perceptual and
; cognitive processes that can be derived from Rorschach
I scores thus makes it a promising instrument for the |
| i
j current study. In addition, by presenting the subject |
i I
I
with stimuli that are similar to those in the GEFT— i.e. i
I
whole forms— but that are accompanied by open-ended
I
response requirements, the Rorschach can provide a
I 129:
i
t
i comparative measure of the subject’s cognitive style and
i
an enhanced opportunity to elicit it. .
The empirically derived scoring and interpretive j
Comprehensive System for the Rorschach developed by Exner ^
(1985, 1986) provided the source for and the definition of'
I
specific Rorschach variables used in this study. In this j
system, responses are coded in eight categories. In I
i
addition to single scores, interpretable ratios, j
I
percentages, and numerical derivations are also provided, j
as well as variable clusters (Alumni Newsletter, 1989).
. . . . . I
In order to select positive indices of ICS, attention was ;
i :
focused on the following cognitive processing variables:
1. Location: W:D (Economy Index). The majority of ,
responses in most records are to common detail (D) areas
of the blot. It seems probable that many more D answers [
are available than are answers that utilize the whole blot,
(W). Examples of these two types of answers (that can be j
i
apparent to the reader without reference to the blots) j
include the following responses to Card II: "The whole
i
thing looks like a face" (W) and "this part looks like a I
candle" (D). The presence of several discrete detail
j segments in each blot affords an easier formulation of
multiple potential answers, whereas the subject who gives
multiple W answers to the same figure must retranslate the
I
entire blot each time. Whenever W is elevated, it signals
the investment of extra effort in the task. The ratio
______
130,
i
that would normally be expected is 1:1.5 or 1:2, though
i
ICS may exhibit a higher proportion of W.
2. Developmental quality: DO+and v/+ (DOsvn); W+.
Individuals select and organize blot locations
I
differently, manifesting what Rorschach recognized as j
different apperceptive approaches or Erfassungstypen ]
1
(Exner, 1986). Based on Werner1s (1940/1964) theory of j
!
cognitive development, Friedman (1953) developed a 6-
category coding system to distinguish among these levels
of cognitive processing, more generally described by
various Rorschach investigators as unorganized,
combinatory, and superior.
Through an extensive data analysis, Exner (198 6)
later adapted Friedman1s six categories to four. Vague
(v) answers are concrete and not well organized, requiring
no specification of features— such as "clouds" or "dirt.1 1 j
These undifferentiated responses identify the stimulus j
field in a way that avoids the need to be specific. j
Ordinary (o) responses are higher but still somewhat (
i
j
simplistic and economical, identifying a single object
that requires a specific outline or specific structural '
features (i.e. an object with a specific "form demand")-- ,
such as "a bat" or "a vase." A much higher level of j
cognitive activity is required to produce a synthesized
response, wherein two or more objects, either with a form
demand for at least one object (+) or without a form
. _ 131
! demand for either object (v/+) are described as separate
i
| but related. Examples would be "two people picking up a
boulder" ( + ), or "some storm clouds coming together" ;
j
(v/+). Conversely, the response "some icicles" would not 1
i
posit the objects in any meaningful relationship and would'
be coded o. J
In particular, the W+ response has been described as j
I
"an expression of abstractive and integrative strivings" |
by Blatt and Allison (1963, p. 269), who reported a \
j
significant positive relationship of more highly developedj
Ws with problem-solving ability. The combination of W and
DO+ is of particular interest to the present study because!
i
of the cognitive operations required to formulate such a \
i
response. The whole blot must be reported in the i
response, yet the response must consist of at least two
j
separate objects in relationship. Thus, the blot as a 1
unitary stimulus must be perceptually articulated and then
reintegrated into a well-differentiated unifying whole.
i
In other words, both analysis and synthesis occur, but in
i
i
the service of a preferred final integrative response.
The Card X response, "A lot of things that you see in the !
i i
water like fishes and crabs" is coded Wo; "a lot of j
underwater creatures like fishes and crabs swimming around|
these rocks" is coded W+. Because the conceptual
parallels between DQ and the ICS construct are especially
strong, ICS individuals could be expected to produce more
DQsyn and W+ responses.
j
3. Determinants; M . All human movement (M) answers!
must be projected onto the stimulus field, because the I
1
i
blots do not move. They can be analyzed according to the j
I
types of movement and figures involved, but all Ms are |
[
generally associated to intellectual operations. M has
been positively correlated with abstract thinking and withi
I
I
creativity, as well, although the varying criterion |
i
measures for creativity make a general interpretation of ;
i
research results difficult (Dana, 1968). Several !
i
investigators have related M to delaying operations, which!
I
J
may enable the subject to refrain from spontaneous I
responses and to sort the stimulus field and possible !
i
responses to it more extensively (see Exner, 198 6). The j
J
response to Card III, "It's a cheerful party, yes, here's ]
two people and the decorations," has no human movement, in!
I
contrast to "two ladies moving furniture" or "a waiter !
practicing his bow in the mirror." To the extent that an
M response represents active and deliberate reasoning,
imagination, and higher conceptualization processes, it j
i
p
can be considered to be an indicator of ICS. j
i
4. Form quality: FOo%. The use of form in a
response is considered an ego or thinking operation, in
which attention and judgment regarding environmental !
standards are salient— akin to reality testing operations.
133 |
Rorschach (1921/1942) noted that almost all responses
include some contour features, and postulated that the
i
varying quality of such responses would represent the
subject's ability to perceive things accurately and
conventionally. Exner's (198 6) evaluation of F£) is based
both on the statistical frequency of all items reported in j
J
7.500 protocols (including 162,427 responses) and on some j
distinction of good versus poor form. Items listed are
j
identified as ordinary (o) when they involve readily j
j
apparent blot contours reported by at least 2% of the
i
7.500 subjects; as unusual (u) when they involve readily !
apparent blot contours reported by fewer than 2% of the !
7 500 subjects; and as minus (-) when they occur with low j
i
frequencies and seem to be imposed on the blot, violating
its contours or creating contours that do not exist.
i
Items may also be scored superior-overelaborated (+) if j
the form has been precisely articulated and specified in j
the response. For example, one of the most popular j
responses to Card I is "a butterfly,” coded o. A + :
I
response would identify many more form details of the
| t
\ butterfly than is commonplace (e.g. "exotic— irregular !
i
I
edges on the wings--triangular white markings— very small ,
i
i
antennae— round nobs at the head"). "Bees" and "beetles" j
J
are designated as u, whereas "baboon" is coded -.
Exner (1986) regards the percentage of FOo and +
responses (labeled X+%) to indicate conventional,
commonplace form, and the percentage of F0- responses (X- ,
%) to indicate either (a) degree of perceptual-mediational
distortion or unmodulated affective experience, or (b) j
degree of commitment to individuality. An accompanying i
high frequency of FOu, Exner suggests, would indicate the
latter. ICS individuals might tend not only toward FOu or
+ responses, but also toward the imposition of closure on
a broken contour (FQ-), thus integrating the whole in a
more meaningful manner. An FQ index of ICS should
I
therefore predict that all statistically infrequent
i
responses would occur relatively more often, i.e. to j
predict that FOo would appear relatively less often. I
5. Organizational activity: Zf and Zd. When !
organizing activity occurs, relationships are established
between elements of the stimulus field, and a Z score is
assigned to each response that contains evidence of such
activity. Z. values are weighted according to the j
i
complexity and/or effort required, in reference both to j
the card number and to whether the response indicates (a) 1
!
a W with a D£) of + , v/+, or o; (b) two or more separate
objects in meaningful relation, either as adjacent or as
distant Ds (these responses would also be assigned DQ
codes of + or v/+); or (c) white space integrated with !
other areas of the blot.
A low Z. frequency (Zf) likely represents a reluctance
to tackle the complexity of the stimulus field, and thus a
1351
I
higher frequency would be expected in individuals with ,
ICS. Zd is the difference between the actual sum of Z !
scores assigned (ZSum) and the best weighted ZSum
prediction when Zf is known (Zest). Exner (198 6)
postulated that Zd scores represent organizational
efficiency; 70% of the normative sample score between +3.0
and -3.0. Individuals scoring >+3.0 are thus exerting
greater effort toward complexity on each organizational
t
attempt, reflecting an uncommonly thorough attentiveness j
to and perceptual scanning of stimuli that Exner labeled j
overincorporation. In contrast, underincorporators have a|
tendency to be hasty and negligent in processing i
information, reflected in <-3.0 Zd scores. Despite its j
mild negative connotation, overincorporation was not j
considered to present as much of a clinical problem as
underincorporation. Overincorporation bears
characteristics of overinclusion, and Zd should therefore 1
be another appropriate empirical indicator of ICS.
I
6. Special Scores: INCOM and FABCOM (COM). There
I
| are currently 12 Special Scores in the Comprehensive
System that signal the presence of an unusual processing
or content characteristic in the response. Six concern
unusual verbalizations, an important element in the study
of cognitive processing in that a momentary or more
extended occurrence of cognitive slippage will often
manifest verbally. They are evidenced in Rorschach
13 6 i
i
responses in three categories: (a) Deviant Verbalizations,!
I
I (b) Inappropriate Combinations, and (c) Inappropriate |
i i
Logic (Exner, 1986) . Two of the three combinative I
i
responses, INCOM and FABCOM, are of interest to the !
i
i
present study; the one that was eliminated was considered 1
i
a more bizarre response. INCOM condenses or merges blot
details into a single incongruous object, such as a "woman
with the head of a chicken" or "red bears." FABCOM
--------------------------- i
involves an implausible relationship between two or more I
I
j
separately identified objects, such as "a little baby j
driving a car." !
Although the thought disorder approach to these data :
I
interprets them as inappropriate and unrealistic, it also |
seems possible to formulate such responses from the j
standpoint of ICS, exactly as in the case of j
i
overinclusion, with an awareness and willingness to ;
! combine disparate elements. In fact, these responses
i
could be a likely result of an integrative or
"overinclusive" approach to the blot. Exner (Alumni j
]
Newsletter, 1987) later addressed the issue of differing
degrees of severity by distinguishing Level I and Level IIi
coding levels, based on the magnitude of faulty thinking
or language use.
It is proposed here that Level I responses, coded forj
mild or moderate instances of illogical, fluid, or
peculiar thinking, would predominate in persons utilizing
- — —
ICS. Viglione and McMillan (1990) have further refined j
research criteria for Level I INCOM and FABCOM responses,
which "are relatively clearly communicated in a benign,
often playful, parenthetical or misinformed manner" (p.l),
and which suggest the presence of the following ]
characteristics: (a) adaptive regression ("two dancing
bears"), (b) meaningful imagery that is well represented
in awareness ("a horse with a horn"), (c) equivalent
i
substitutions ("a mouse, here are the hands and here are 1
i
the feet"), (d) an obsessive concern with form accuracy
1
and details that produces an inconsistent or idiosyncratic^
combination ('two women with bird heads . . . the head j
part looks more like a bird . . . the nose is more like a j
beak, and they don't really have any hair"), and (e) j
i
conventional or meaningfully communicated symbolism ("the |
three ages of man . . . an infant's face, and young man's ;
I
face, and an old man's face all connected"). The authors i
view the last response type as an intellectualized attemptj
that may at times include the narcissistic aim to impress
i
the examiner. If they were evaluated solely from a
perceptual point of view, some of the responses could
appear quite disturbed.
Theoretical Summary and Overview of Empirical Aspects
The theoretical and philosophical heritage that has
brought the constructs of ED and FDI together has been
impressive, and yet evidence for their monotonic
r3 8'j
i
association has been weak (MacDonald, 1986). It is shown j
in the preceding discussion that Witkin's neglect of
integrative processing, although it figures predominantly
in the intellectual roots of his construct, may have been
responsible for the nonsignificant results, as well as his
assumption that the FDI continuum replicates the
developmental continuum. Additionally, his construct is
considered to be misconceived in that it assumes the
character and presence of FD by default, bifurcates
sensory from logical processing, and misconstrues |
indications of relatedness and holism solely as signs of j
developmental immaturity, a conclusion that is |
inconsistent with other indicators of high ED. |
I
The proposition of an integrative cognitive style hasj
been theoretically developed and translated operationally I
I
into corresponding measures that show promise for j
investigatory purposes, adapted from the OST and the !
Rorschach. These operationally defined, theoretically !
derived variables are further hypothesized to be criterion1
measures of postconformist ED, substantiated by
relationships implied in Loevinger's theory and also by
the fit of ED theory with other theoretical treatments of j
such common aspects as integration, complexity, and |
j
creativity. It is also suggested that Loevinger's account1
of development from Impulsive to Integrated ELs parallels
a conception of the ICS dimension as moving from global
i (FD) to integrative processing, so that the ICS dimension
i !
may be a more probable polar aspect of ED than FDI is.
The current study therefore addressed two issues of
construct validity. The first concerns the cognitive
processing style of individuals at postconformist ELs as
defined by Loevinger (1976), and a comparison by stages
design was followed. Care was taken to obtain two groups
i
of adult women that comprised strong representatives of
high and low ELs. To evaluate the validity of the ICS
j
construct that was developed in relation to the first j
research aim, and also of its exploratory empirical j
variables, an attempt was made to establish convergent and;
discriminant validity among the criterion measures.
The research hypotheses were as follows:
I
1. Features indicative of a high level of ICS will '
be more prominent among high ED subjects than among low ED1
I
subjects, as measured by the theoretically derived and 1
adapted ICS measures. A related expectation is that high !
ED subjects will exhibit a greater overall number of the
i
ICS features measured by Rorschach variables than will low
ED subjects, when the variables are tallied as positive
for the criterion of ICS and summated into a Composite ICS
Index (Cl).
i
2. The ICS measurement variables will serve as
effective discriminating criteria for the identification
of high ED.
[_ - _ _ -14_ 0_
I 3. Relative to the low ED group, the high ED group
i will show no significant difference in analytic cognitive j
I 1
I i
| processing (FI) as measured by the GEFT. Neither group is
expected to obtain especially high scores, because
i analysis may reach its ascendancy on average in the j
dropped, Conscientious EL. Consistent with the j
possibility that analytic and integrative cognitive |
processing represent two separate dimensions, high and low
scores for some subjects on the ICS indices and the GEFT j
i
I may variously overlap, but no specific predictions '
i *
regarding the relationship are tested. j
' 4. The OST and Rorschach variables will exhibit a !
i
I
significant degree of association with each other, because!
i 1
they are purported to measure the same ICS construct, in
distinction to a lesser degree of relationship with the
i
GEFT, which is considered to tap the FDI continuum and to
indicate by higher scores the presence of an analytical
i
(FI) cognitive style. Although general support for the
convergent and discriminant validity of these measures is j
thus expected, an unspecified degree of association
between the GEFT and ICS variables is also anticipated, !
for two reasons. First, both integrative and analytic |
i
cognitive processes appear to rely on each other in a
reciprocal if subsidiary manner when one or the other is
predominant. Second, low ICS scores have been suggested
as indicators of a global cognitive style, whereas low
141 j
scores on the GEFT likely represent a global cognitive
style for some subjects, as well. To reiterate the
I
position of the present study, however, the GEFT cannot
differentiate alternative cognitive styles that may have
precluded the attainment of a high score on the GEFT.
The major restriction placed on the scope of this
study was that the possibility of directly testing the
developmental continuity or discontinuity of either ICS or
F D I , as in a multiple regression analysis with E D as a
continuous variable, was not addressed. Neither did the
J
design incorporate a factor analysis of the I C S j
construct's dimensional properties in relationship with j
i
F D I . However, support for construct validity in terms of ;
i
mean group differences, discriminant analysis, and test j
intercorrelations should yield evidence pertinent to these!
related questions.
142
CHAPTER III
Empirical Methodology
In this chapter, procedures are outlined by which the
preceding research hypotheses were empirically tested.
Research design and control procedures, the selection and
classification of subjects, and also the content,
administration procedures, scoring properties, and
measurement characteristics of the test instruments are
described. The chapter ends with a description of the
data analysis procedures that were selected to test each
hypothesis.
Research Design and Control Procedures
The research design was determined by the aim and
procedures of construct validation. Logical
considerations of ED theory formed a basis for predicting
group differences in the extent of integrative processing
between high and low ELs, and thus two samples of subjects
were formed to represent both extremes of the ED
continuum, skipping the Conscientious EL. This EL is
usually considered to be the first clear indication of
postconformist functioning, and it is followed at later
levels by more of the complexities of thought that suggest
143 j
I
ICS. Because the goal was to sharpen a particular j
dimension that may characterize the highest levels of ED, |
i
the formation of an uneguivocally high target group was a |
necessary feature of the research design. The omission of
the Conscientious EL also established a clear boundary
between the high group and the low comparison group. The
low group included levels through the Self-aware EL. This
EL is considered to be the transitional level preceding
postconformity, and Loevinger's extensive data base
suggests that it is probably the modal EL for the larger
population. It is at least as difficult to find adults
who function below the Self-aware EL as it is to find
adults above the Conscientious EL, and it was necessary to
include Self-aware subjects in the low group. The study
was restricted to women in order to contribute a focused
account of the nature of women*s functioning at high ELs.
Another construct validity issue concerns the
viability of the ICS construct and its measure, as well as
its distinction from FI. Thus, the construct was
operationalized by using multiple variables from two wide
band tests, in order to avoid a mono-operation bias and to
render any irrelevancies heterogeneous. This combined ICS
measure was then administered with the GEFT. Several
methods were also used to avoid mono-method bias, in which
the influence of the method cannot be separated from the
144
i
t
influence of the construct. Subjects produce
manipulations and verbal descriptions for the ICS
measures, whereas the GEFT is a paper-and-pencil test. In
addition, experimenter bias was controlled for in several
ways. Because the Rorschach is sensitive to response
cues, two experienced Rorschach examiners who were blind
to the study assisted the investigator in test
administration. They examined a majority of the subjects,
and they also coded every response on all of the
protocols. In addition, 20% of the protocols were double
I
I
scored by a fourth experienced rater who was also blind t o '
the study, resulting in negotiated compromised scores for i
i
these eight Rorschach records. All four raters had been I
I
trained in the Comprehensive System of scoring and ]
interpretation (Exner, 1985, 1986). Every SCT and OST
protocol was also double scored, and disagreed ratings
were discussed and compromised by the raters involved.
With two constructs (i.e. analytic and integrative
cognitive processing) and three measures (i.e. OST,
Rorschach, and GEFT), conditions were met to test for
convergent and discriminant validity of the ICS.
Selection and Classification of Subjects
The subjects were 4 0 women drawn from a larger sample
of 150 women volunteers, all of whom were recruited by
mail, telephone, or through an introductory presentation
145
i
at several job and school sites in the greater Los Angeles |
area. The only stipulations were that they be at least 18
years old and that they be available for follow-up testing
if contacted following the completion of the initial task,
i.e. the SCT (Loevinger, 1985; Loevinger & Hy, 1989;
Loevinger & Wessler, 1970; Loevinger et al., 1970).
Eighteen women who had participated in previous studies of
ED (MacDonald, 198 6; Urbaitis, 1988) agreed to participate
i
in the follow-up testing procedure. The decision to use 1
i
i
subjects previously rated for ego level was based on the j
theoretical assumption of an invariant, sequential order j
of ego developmental stages (see Loevinger, 1979a, 197 9b, j
j
197 9c), in which any change would be progressive rather
i
than regressive. Regression under stress probably takes j
I
place, but does not contradict the conception of an i
individual's ego structure as the level at which one is
capable of functioning consistently (Loevinger, 197 6) .
Recruitment continued until 22 additional subjects were
found who qualified either for a high ED group or a low ED
group, out of a total of 132 returned protocols. Twelve
protocols were eliminated because of skipped items. Six
protocols were eliminated because the responses appeared
to be formed by non-native English speakers of limited
fluency, which could confound the accuracy of the scoring.
On the basis of their SCT total protocol ratings (TPRs),
146
2 0 women were placed in the low ED group (1 at Self-
protective, 7 at Conformist, and 12 at Self-aware Els),
and 2 0 women were placed in the high ED group (12 at
Individualistic and 8 at Autonomous ELs).
After the 4 0 subjects were selected on the basis of
their Final TPRs for the high and low ED groups, they were
contacted by telephone, and an appointment for individual
follow-up testing was arranged. Subjects met with one of
three experienced Rorschach examiners, including the
present investigator, who also administered the OST and
the GEFT in the same testing session when possible. If
time schedules did not permit this arrangement, subjects
were administered the latter two tests in a separate
i
i
session. They were generally tested in a separate office !
i
or conference room where task orientation and privacy
could be maintained. A few subjects had to be tested in
their homes, where effort was directed to establish these
conditions, as well.
The mean age of the low ED sample was 38.6 years
(range = 18.6-58.4 years, mdn = 38.2). Educational levels
were computed by grade level through college (senior year
= 16). Succeeding educational levels were represented as
follows: (a) pre-masters = 17, (b) masters degree = 18,
(c) pre-doctoral = 19, and (d) doctoral degree = 20. Low
ED subjects averaged a 13.1 level of formal education
147
(range = 8-18, mdn = 12), and 60% were currently enrolled
in some kind of educational program at a community
college, adult evening school, or skills center. Sixty
percent were employed in occupations ranging from teaching
I
and management positions to various types of assistant
positions (e.g. instructional, office, retail, and
clerical). The mean age of the high ED sample was 54.1
years (range = 26.8-71.7 years, mdn = 54.6). They
averaged a 17.5 level of formal education (range = 12-2 0,
mdn = 18). Ten percent were currently enrolled in an
educational program. Seventy-five percent were employed j
in occupations that ranged from teaching, counseling, j
administrative and secretarial positions to positions in
accounting and advertising. Three subjects (15%) in the
high ED group had retired.
Clear, statistically significant differences between
high and low ED groups appear in both age and educational
levels (p < .001 in each instance). The oldest subjects
had all participated in career counseling at a local
i
i
university, and they all scored in the postconformist j
range on the SCT. The youngest subjects selected for the !
study were enrolled in adult education and vocational !
t
classes, and scored at the lower end of ego development. i
i
The disparity between groups is likely attributable to
more than the limitations and consequences of these
148 l
I
particular sampling methods. Several findings (see I
Hauser, 197 6) suggest age trends in ego development, in
the sense that age may be a necessary but not sufficient
condition for reaching certain ego levels. In addition to
the influence of chronological age, Loevinger*s model also
posits that the developmental sequence is influenced by
various kinds of environmental stimulation, of which
amount of formal education may be considered a significant!
component. I
Instrumentation I
----------------------------------------------------------------------------- i
Sentence Completion Test. On the SCT, subjects are j
asked to fill out a sentence completion form by finishing |
i
a series of incomplete sentence stems covering a variety
of topics in a person's life. Sixteen of the subjects
from previous studies were rated on the SCT Form 11-68 for
women, the first revised form to provide alternative
versions for both women and men. The original SCT was
devised for use with women, and this focus of interest
influenced inclusion of some sentence stems (e.g. "A
pregnant woman— ") as well as the content of the original
scoring manual, which was derived solely from female adult
and adolescent responses (Loevinger et al., 1970). Forms
11/68 resulted from the development of a men's form and
the replacement of four of the least interesting items on
the women's form, and were included in the manual as
149
recommendations shortly before its publication (Loevinger
& Wessler, 1970).
At the time of the SCT's development, use of women as
subjects in psychological research was infrequent, and
interest in using the test for both men and women and in
the study of gender differences was not anticipated. The
more recent Form 81 and Supplementary Manual (Loevinger,
1985; Loevinger & Hy, 1989) have updated Forms 11/68 by
making the men's and women's forms more closely
comparable, and by integrating some items originally
composed for men and for which the scoring manual was
derived from male samples. Loevinger addresses the
question of whether the stems, the manual, and even the
whole conception have been slanted toward women— much as
she contends Kohlberg's conception is male-oriented
(Loevinger & Knoll, 1983)— by noting that her ED construct
was originally borrowed from studies of delinquent young
men by Sullivan et al. (1957). Additional improvements on
Form 81 include optimal ordering of items and the
equivalence of first and second halves to use as short
forms of 18 items each. Therefore, the remaining 13 2
volunteers were screened with the first page of Form 81,
on which all stems are identical for men and women. All
previous and current items are regarded as equally valid
and reliable indicators of the ED construct. The test
150
format on Form 81 and the wording of all retained stems
remains the same as for previous forms.
Reliability studies have addressed the test itself as
well as the scoring system, particularly the critical
issues of interrater reliability. Loevinger and Wessler
(1970) report extensive evaluative studies using 543
protocols. Five raters personally trained by Loevinger
and her group obtained a median interrater agreement on
TPRs of 61%, and a median interrater correlation of .86.
They were compared to two self-trained raters who used the
manual's series of self-training exercises, answer keys,
and explanatory notes (Loevinger & Wessler, 1970,
Appendices C and D). The median interrater correlations
ranged between .89 and .92. The median percentage
agreement on individual item ratings for personally
trained and self-trained raters was 78%. The factor
structure of the test was also examined, and it was
concluded that the first components are essentially
identical irrespective of the type of training of the
raters. In terms of factor validity, the items taken
together additionally appear to measure a unitary
dimension that can be considered to be a developmental
trait.
Redmore and Waldman (1975) discussed studies of the
reliability properties of the SCT itself, using test-
151
retest, split-half, and internal consistency (alpha)
indices. These authors found reliabilities ranging from
.68 to .91, with values varying dependent upon situational
factors and "motivational sets." Other studies have
reported an internal consistency alpha coefficient of
approximately .77 (Cummings & Murray, 1989; Holt, 1980?
Waugh, 1981).
Many studies, summarized by Hauser (197 6) and
Loevinger (1979a), document the substantial construct
validity of the instrument in relation to such topics as
stage sequences and the behaviors and attitudes
characteristic of each stage. Correlations with other
developmental stage tests (conceptual and moral) and
behavioral measures have been reported by Loevinger
(1979a), although she notes that construct validation is
complicated by the cognitive developmental orientation of
ED and the milestone sequence model, which is not always
amenable to correlational investigations (Loevinger,
1984). Recent studies pertaining to discriminant,
predictive, and construct validity have examined ED in
relation to perceptions of educational issues (Cummings &
Murray, 1989), object relations in middle childhood (Avery
& Ryan, 1988), various facets of personal constructs
(Deitsch & Jones, 1983), bulimics' conceptualizations of
their disorder (Teusch, 1988), in-session cognitions of
152
counseling supervisees (Borders, 1989), Black identity
I
(Looney, 1988), and self-monitoring and the imaginary
audience (Lapsley, Jackson, & Rice, 1988). A recent 1
i
critical review of the theoretical and empirical backing 1
I
for the SCT has been provided by Broughton (1989). |
Group Embedded Figures Test. In the GEFT, the j
subject is asked to locate a simple figure within a j
complex geometric pattern. Light sectional shading servesj
to emphasize large organized Gestalten that embed the
simple forms. Each of the 25 items printed on the test
booklet pages contains one of the eight possible simple
figures displayed on the back cover. The test is divided
I
into one 2-minute practice section of seven very simple !
I
items, and two succeeding 5-minute sections of nine more j
difficult items each. Witkin et al. (1971) considered j
that an appropriate method of estimating reliability with \
a speed test was the correlation between parallel forms
with identical time limits. Correlations between the
First Section and Second Section scores, corrected by the
Spearman-Brown formula, produced a reliability estimate of
1
.82 for both males (N = 80) and females (N = 97). A more j
J
recent reliability study examined test-retest reliability j
and found some suggestive discrepancies (Kepner & Weimark,
<
1984). Unlike subjects who consistently scored in the FD i
i
or FI range, one subgroup of latent FI scorers obtained
153
retest scores that moved them from the first, FD quarter
to the fourth, FI quarter, regardless of whether the ;
I
retest was immediate or delayed. The researchers j
i
characterized this latency as a FI style that is not j
i
spontaneously deployed.
In the manual, Witkin (1971) presented validity
coefficients for the GEFT and three criterion measures:
the EFT, the RFT, and an assessment of human figure
drawings for degree of articulation, ranging from -.3 9
to -.82 (negative because of reverse scoring). Subsequent|
I
research linking FDI to other constructs has proliferated, j
i
often with conflicting findings and interpretations, and
it is toward the issue of construct validity that the
discussions in Chapters I and II were in part directed. j
I
Studies have begun to question whether FDI is a style or
an ability (McKenna, 1984? Shade, 1984). Moran (1985)
recommended that the nature of cognitive restructuring
receive closer scrutiny from researchers, and that
programs of construct validation be conducted on the EFT
and RFT.
Object Sorting Test. The objects used for this study
were those that Goldstein and Scheerer (1941) recommended
as suitable for male subjects. "Male" objects were
selected because they were considered to be more neutral
and less gender-typed by current standards, in contrast to
the set of "female" objects, which included items such as
darning blocks and knitting needles. The objects used are
commonly obtainable and included tools, eating utensils,
smoking utensils, edibles, and toys; they were of
different shapes, colors, and materials, and thus provided
many potential bases for sorting (see Appendix A). They
were placed on a checkered tablecloth as suggested to
facilitate their placement, in an arrangement pictured in
the figure accompanying the study.
The validity of the OST as a measure of overinclusionJ
has been supported in several studies that found j
i
significant relationships to other related measures of
I
overinclusion, and also to indicators of other related ]
i
cognitive processes (e.g. Andreasen and Powers, 1975; |
Harrow, Harkavy, Bromet, & Tucker, 1973; Harrow,
Himmelhoch, Tucker, Hersh, & Quinlan, 1972; Harrow,
Rattenbury, Marengo & King, 19 90; Harrow, Tucker, & Adler,
1972; Payne & Friedlander, 1962; Quinlan, Schultz, Davies,
& Harrow, 1978). In general, the OST has long been
employed to obtain indices of thought disorder (e.g. Payne
& Hewlett, 1960; Rapaport et al., 1968).
Following the recommendation of Harrow (Harrow,
Himmelhoch, Tucker, Hersh, & Quinlan, 1972) to evaluate
sortings on several different scales, two of his
qualitative scales for overinclusion were selected for
155
this study, in addition to the quantitative scale of !
"behavioral" overinclusion. The authors report interrater
reliabilities, using Pearson product moment coefficients,
of .89 for the Conceptual overinclusion scale, and .79 for
the Richness of associations scale.
Rorschach. Rorschach believed that a subject
responds to the presentation of an inkblot by integrating j
the stimulus sensation with existing engrains or memory j
traces in a conscious associational process. Exner (1986)!
i
describes the Rorschach as a problem-solving situation in I
which the subject is forced to convert the blot into
something it is not, eliciting a range of possible
responses and requiring decision processes to select a
final answer from among them. As such, both theorists
believe that the response process provokes a complex of
psychological operations that results in a broad array of
answers. These propositions are clearly compatible with
conceptions of ego processes and personality style.
Exner (1986; Alumni Newsletter, 1989) has provided
several tables of normative data and comparative data from
three adult psychiatric groups to help in the
understanding and utilization of the test and the
interpretation of the wide variability that it elicits.
The updated normative sample consisted of 3 50 females and
350 males, drawn randomly from a total sample of 1300
156
nonpatients and stratified by SES and geographic
distribution (Alumni Newsletter, 1989). Exner (1986) ]
cautions against the noncritical use of means and standard j
i
deviations as reference points, especially because the j
distributions of many Rorschach variables fall on a J- !
I
curve, i.e. most of the values fall on one, two, or three
data points with very few deviations. Therefore, he has
added frequency, range, mode, skewness, and kurtosis
descriptive statistics to his normative and comparative
data, and recommends a consideration of all seven measures
before a decision is made as to whether a certain variable
value is "average" or "deviant." He also observes that itj
t
j
is erroneous to assume that norms reflect normality, or I
i
i
that deviation reflects liability. Deviations do indicate i
, , I
uniqueness, but this can be an asset, and the individual
with many scores that fall outside the average range may J
i
be well-adjusted. !
Validity studies span the history of the test itself,!
i
and the Rorschach has been used in a vast array of |
i
criterion-related research, especially in studies of the j
personality organization of specific groups, and even more;
I
particularly of psychiatric populations. In terms of *
reliability, Exner postulates that the selection process !
inherent in the Rorschach task calls upon the relative 1
consistency of personality characteristics and thus |
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ i
157
contributes to the strength of the test's reliability over
time. He criticizes earlier efforts to establish the
test's internal consistency with a split-half technique,
because the Rorschach blots are not equivalent stimuli.
Exner has devoted much of his work to the study of
temporal consistency of classes of response, postulating
i
i
that people's preferred response styles should be |
consistently evident in repeated testings. He has j
reported more than 3 0 test-retest studies completed by j
1983 (Exner, 1986). Two studies with lengthy retest
intervals (12-14 months and 3 6-39 months) show that most
variable correlations fall between .80 and .89; only five
variables, related to state rather than trait features,
fall below .70.
Procedure
Administering and scoring the SCT. After obtaining
the cooperation of potential subjects by mail, telephone,
or personal contact, number-coded SCTs were either mailed
or distributed at the site (e.g. school or place of
employment), accompanied by a brief cover sheet requesting!
i
i
demographic information for statistical and follow-up |
i
purposes and an introductory letter explaining the generalj
intent and procedures of the study. The protocols were j
I
either completed and collected at the site or returned to ;
the researcher in stamped, addressed envelopes that were |
158
provided for that purpose. The SCT forms were separated
from the identifying information on the cover sheet, and
the item responses were then rated stem-by-stem, according
to the procedures outlined in Loevinger & Wessler (197 0).
Rating was performed by three experienced raters: the
current investigator, a male professor of psychology, and
a female postdoctoral psychological assistant. All raters
had been trained using the exercises in Appendices C and D
of the original scoring manual (Loevinger & Wessler,
19 7 0), and had scored several other samples for previous
research studies. There were two raters per item, with
the current investigator always being one of the two.
i
They scored each item independently by using the original j
i
manual (Loevinger et al., 1970) and the supplementary !
manual (Loevinger & Hy, 1989). Each response was assigned!
its own EL by matching it to the numerous response
categories in the manual that are provided to guide their
placement. Essential features of each category are
highlighted by listing a combination of actual responses,
composite responses, or abstractions of the category's
common elements.
Discrepancies between raters were resolved through
discussion, shared judgment, and compromise. Additional
consultation in reference to subjects rated for a previous
study (MacDonald, 1986) was also solicited from a group of
159
experienced raters at the Washington University Social
Science Institute. For purposes of computing interrater j
reliability, the original scores before discussion were
used. With nine levels and strong theoretical support for
the underlying continuity of the developmental
progression, ED ratings were regarded as comprising a
continuous scale for the purposes of selecting a
reliability coefficient. An ANOVA intraclass correlation
coefficient (ICC) was used, which assesses variance within
a data set for a given subject as well as between
subjects, and which can be tested for significance with
the F statistic (Bartko & Carpenter, 1976). As shown in
i
Table 2, ICCs calculated for exact agreement of item |
|
ratings on 45 protocols ranged from .56 to .82 with a I
i
i
median of .72. The corresponding reliability of the TPR
was .89. When agreement of item ratings between raters
was defined as plus or minus one EL, values ranged from
.68 to .92 with a median of .81. The corresponding
reliability of the TPR (plus or minus one level) was .89.
Loevinger and Wessler (1970) proffer the following
reasoning for computing this second reliability index:
Considering each [level] as one point on a
scale, we have a [nine-point] scale. If
we think of ego development as a
continuum, then disagreements [of one
level] are inevitable for intermediate
cases. Hence they are in this sense not
errors, or not necessarily errors. (p.45)
160
Table 2
Indices of Interrater Aareement bv Items for the Sentence
Comoletion Test (SCT), Six Rorschach Codincr Cateaories,
and Two Obiect Sortincr Test (OST) Scales
SCT agreement
a
Rorschach exact agreement1 3
(45 cases, 18 items per case (8 cases = 171 Responses [R], 6 coding
[810 scores]) categories per R [1,026 scores])
Reliability
+/~
Coding
(ICC)*** Exact 1 EL Category N
O , *
.56-.60 1
.61-.65 1 DO 125 73.0
.66-.70 5 1 Det 12 7 74 . 0
.71-.75 5 1 m 121
71.0
.76-.80 5 7 Zf 152 89 . 0
.81-.85 1 5 ZSum 141 83 . 0
.86-.90 3 Special 132 77 . 0
.91-.95 1
Total items 18 18 Total agreed 7 98 78 . 0
ICC range .56-.82 . 68-.92 % range 71.0-89.0
Median ICC .72 .805 Median % 75 . 5
TPR ICC .89 .93
OST scales0
O-in agreement Rich agreement
(36 cases, 5-point scale (36 cases, 5-point scale
[36 scores]) [36 scores])
Statistic Exact
+/“
Exact
+/~
1 point 1 point
Po *44
.86 . 56 . 67
Pc .26 .26 . 19 . 20
k .24 .81 .46 .58
z 2.36* 10.38*** 4.60*** 5.90***
aICC = intraclass correlation coefficient. df = 44, 45. EL = ego level.
TPR = SCT total protocol rating. All F values associated with ICCs (F range
= 3.49-24.29 for exact and +/- 1 EL agreement) were significant at p <.001.
bN = number of responses in a coding category with perfect agreement (total
possible = 171 responses). Det = total determinants coded for a particular
response. Multiple determinants in one response are possible. Special = total
Special Scores coded for a particular response. Multiple Special Scores in one
response are possible. DQ and FQ categories require one of four possible
codes. FQ. is occasionally absent from a response. Zf, ZSum. and Special
Scores are also frequently absent. Interrater agreement was defined by the
perfect match of every element in a response coding category, including the
absence of codes.
c0-in = Overinclusion scale. Rich = Richness of associations scale. P =
proportion of agreement observed between raters. P^. = proportion expected by
chance. k = Cohen's kappa. z = associated z score.
*p <.05. ***p <.001.
161
After all item responses were scored, the protocol was
I
evaluated as a whole in order to assign an impressionistic!
i
TPR. Thereafter, each set of varied item responses was <
reduced to a single EL, using the cumulative frequency 1
i
distribution to construct an ogive TPR. Automatic ogive
rules or cutting scores had originally been developed for
the 3 6-item test in order to determine which EL to assign
as the ogive TPR, but an adapted set of TPR rules for the
short form of Form 81 has only recently been developed (Hy
& Bobbitt, 1990; see Appendix B). Following the
assignment of an ogive TPR, the final step in rating was
to combine both the ogive and impressionistic TPRs into a
Final TPR rating. Thus, the Final TPR begins with the
empirical rating but may be improved on the basis of
intuitive theoretical understanding (K. H. Bobbitt,
personal communication, 12/9/89).
To evaluate the new ogive cutoff rules and a set of
alternative item sum rules, the investigators calculated
percent agreement between the automatic TPRs derived from
them and the previously assigned Final TPRs for 1,000
protocols, obtaining a range of .46 to .80 (K. H. Bobbitt,
personal communication, February 9, 1990). The ogive
scoring method offers several advantages over alternative
scoring paradigms. It retains valuable information by
allowing for a mixture of responses and a variety of
162
themes, and can therefore include the presence of a few
i
very low responses on high ego level protocols. Using a !
configuration of responses is more like a clinician's
diagnostic thinking, and is compatible with the formal
conception of ego level as core functioning. In contrast,
for example, using the mode as the TPR of an item-ratings
distribution implies that an individual's characteristic j
ego level is the one where he or she most frequently !
operates. j
Some disadvantages accompany the item sum or mean TPR j
scoring method, whereby an interval scale is assumed for j
t
the Els. In this case, higher ratings would be awarded to;
subjects with, say, more conformist answers than to those
with a mixture of conformist and other lower level
answers. Access to many developmental layers, however,
may itself be an indication of high ego level (Loevinger &
Wessler, 197 0). The automatic ogive scoring rules are
therefore both convenient and particularly suitable for
the ego development construct.
Administering and scoring the GEFT. Manual
directions were followed to administer and score the GEFT
(Witkin et al., 1971). The manual includes a scoring key
in which the simple form is outlined over each complex
figure. The score is the total number of simple forms
163
correctly traced in the combined second and third
sections.
There has been some criticism of the scoring manual
for its lack of clarity on score interpretation (Hall,
1978; LaVoie, 1984). Norms are provided only for college
students, and only score values corresponding to the four
quarters of that distribution are reported. Di Nuovo
(1984) has examined the administration times of the GEFT
and the significantly different norms for men and for
women that result, noting that in an unlimited time
condition, sex differences disappeared. Another problem
stems from uncertainty regarding the use of the reported
norms (e.g. Hall, 1978): "Surprisingly, in spite of the
tremendous number of research studies conducted by Witkin
and others, no interpretative standards are provided for
the score scale" (p. 572). Researchers therefore commonly
derive cut-off scores for the classification of FD and FI
individuals from the first and fourth quarters of college
student norms, in accordance with Witkin's assumption that
such classification is meaningful only at the dimension's
extremes (Witkin et al., 1962/1974).
Administering and scoring the OST. Goldstein and
Scheerer (1941) described five experimental sorting
conditions in their research with the OST, but the first,
"handing over" experiment was used in subsequent
164
investigations of overinclusion (Harrow et al. , 1973;
Himmelhoch et al., 1973; Payne & Friedlander, 1962), and
was therefore used in the present study, as well.
Instructions provided by Goldstein and Scheerer were
followed. The subject was asked to select any item he or
she liked as a point of departure, to hand it over to the
examiner, and to continue handing over all the objects
that the subject thought could be grouped together with
the starting object (SO). Then the subject was asked to
explain why the objects were sorted together. The
experiment was repeated three more times, according to
Payne and Friedlander's (19 62) procedure. The examiner
recorded on a scoring sheet the number of objects sorted
with each SO, the order in which they were sorted, and the
verbal explanation given for each sorting.
Scoring of results was based on the three scales
selected from Himmelhoch et al. (1973). They were
considered capable of measuring varied levels and
organizational characteristics of perceptual-cognitive
processes pertinent to the hypothesized ICS, as discussed
in Chapter II. Although the 5-point scoring guidelines
provided for each scale were generally followed, some key
adaptations were made to establish consistency with the
conceptual logic of the current study. Features of the
165
three scales are described below (refer to Table 3 for
scoring details):
1. Behavioral overinclusion scale: sum of sorted
objects. Payne and Friedlander (19 62) found positive
evidence for the presence of overinclusion in
schizophrenia, using a count of total objects sorted as
the scored index. Harrow, Himmelhoch, Tucker, Hersh, &
i
Quinlan (1972) reported that a number of studies have I
. . i
supported Payne and Friedlander*s findings, although j
i
others have found negative evidence in attempted 1
i
i
replications. They suggested that confusion may arise ■
i
from confounding behavioral overinclusion, conceptual !
I
overinclusion, and stimulus overinclusion (i.e. a j
difficulty with selective attention and distraction). |
j
Harrow and his colleagues developed qualitative scales for!
conceptual overinclusion, but also retained the behavioral
inclusion index as part of the overall assessment of
1
overinclusive thinking. Given the partial evidence in the
literature of its empirical value, the behavioral
overinclusion scale was included in the present study.
2. Conceptual overinclusion scale: weighted 1-5
according to degree of conceptual overinclusion displayed.
i
In the manual (Himmelhoch et al., 1973), four types of j
i
conceptually overinclusive responses are described, along j
i
with guidelines for weighting them according to whether a ,
Table 3
Description and Scoring of Three Ratincr Scales
for the Object Sorting Test TOST)
Scale
step
Scale
score
Sorting
behavior
Behavioral Overinclusion
Sum
obj ects
Number of objects sorted
Conceptual Overinclusion
1 1 or 2 Uses several dimensions of
SO without recognizing discreteness:
1 = wandering response; 2 = simple break in set
2 1 or 2 Arbitrarily changes starting points:
1 = greater frequency or intensity; 2 = lesser
3 3 Normal, logically delimited sortings
4 4 or 5 Force-fits objects into chosen dimension:
4 = harder to see; 5 = easier
5 4 or 5 Uses distantly related category for sorting:
4 = more vague and loose; 5 = less
Richness of Associations
1 0 No richness of associations
2 1 Uses a clearly different dimension as
a sorting principle for each SO
3 2 Clearly recognizes many categorizing
principles for SO; acts on recognition
4 3 or 4 Uses original dimensions of the SO:
3 = mild instance; 4 = strong
5 3 or 4 Makes an unusual and appropriate inclusion
of an object in the selected category:
3 = mild instance; 4 = strong
Note. SO = Starting object.
167
relatively milder or relatively more extreme form of this
behavior is displayed. The behaviors are weighted
according to their theoretical severity, although the
different behaviors appear to maintain a nominal or
categorical relationship to one another. For the purposes j
of the current study, the manual scale was modified in j
order to portray a postulated underlying ordinal continuum!
i
of categorizing behavior, and to remove the manual's null j
code of "no overinclusion" (which was actually assigned a j
"1"). The modified scale is exhaustive, in that every |
!
subject is assigned a score corresponding to some kind of
sorting behavior. The weights reflect a dimension that
moves from relatively more disorganized, undisciplined,
and wandering responses toward those categories that tend
toward expansion while still preserving an opportunity for
structuring. Thus, a lower score represents behavior that
is relatively dissimilar to the ICS conception, and a
higher score indexes behavior that more closely resembles
ICS. The manual also equates the behavior of "no
conceptual overinclusion" (absence) with the presence of
"normal, logical groupings," assigning it a score of "1."
In the current study, such behavior is reconceptualized to
be the midpoint between the random, loosely constructed
categories at one end of the continuum and the more
I
strategically expanded categories at the other end. |
i
i
t
i
168
Because one rating must be assigned for the subject's
performance on all four sortings, it is possible that one
type of sorting behavior on the scale could co-occur with
another one. For such a possibility, the following rule
of precedence was defined: whichever instance of sorting
behavior could be judged as more compelling, according
either to the frequency or to the intensity of its
appearance over the four sortings, was the behavior that
should be assigned a code. This rule also made the set of
conceptual overinclusion codes mutually exclusive (see
Bakeman & Gottman, 1986, for a detailed discussion of
I
t
coding). j
The ordered response types and corresponding scores |
I
are overviewed in Table 3. The two highest responses I
i
differ from the two lowest responses in that the subject j
seems to be consciously articulating and maintaining a
categorical definition in the former case. Thus, the
subject finds a way to include items as category members !
that are judged to be far-fetched (in the fourth response)
or to form categories based on "overgeneralizations" (in
the fifth response)— a type of behavior originally labeled
"overabstract" by Cameron (1939). Himmelhoch et al.
(197 3) observed that such elaborations may become so vague
that almost any possible item could be sorted.
Nevertheless, these two behaviors were most highly
169
weighted as potential indicators of ICS, given the
conscious assertion of a category in these responses and
the likelihood of using integrative processes to formulate
the sortings.
3. Richness of associations: weighted 0-4 according
to degree of behavior evidencing uncommonly seen but
appropriate dimensions in the SO, and/or richness of
associations. Score points were arranged to match
increasingly prominent indications of this behavior,
according to the weightings suggested in the manual |
j
(Himmelhoch et al., 1973). The same rule of precedence |
i
described above for co-occurring behaviors throughout the |
t
four sortings was defined. However, not all subjects
necessarily manifest richness or originality in their
sortings, and thus a null code designating "no richness of
associations" was set as the low point of the scale. This
scale enhances the discriminability of the Conceptual
overinclusion scale by increasing the opportunity to
distinguish high-level integrative processes from
inappropriate, undisciplined meandering.
One score for each of three OST scales was assigned
to every subject's test performance across the four object
sortings. All tests were scored by two independent
raters, a female postdoctoral psychological assistant and
the investigator, by following the guidelines summarized
170
in Table 3. When Cohen's kappa was computed to assess the
reliability of the paired ratings for all subjects on the j
I
Conceptual overinclusion and Richness of associations
i
i
scales (see Table 2), agreement was found to be above I
chance on each scale (Overinclusion k = .24, z. = 2.36,
p <.05, and Richness k =.56, z. = 4.60, p <.001).
Although the results were significant, the kappas
i
were a cause for concern because they were lower than j
values generally considered an indication of good (.60- ]
.75) or excellent (>.75) agreement (Bakeman & Gottman, I
J
198 6). Following the reasoning of Loevinger and Redmore !
(197 0) quoted above, the graded responses of the OST
i
i
scales, especially of the Conceptual overinclusion scale, |
i
would lead to inevitable discrepant ratings for |
I
intermediate cases. Although the scales did not have
enough gradations (or a strong underlying logic of
continuity for the Richness scale) to justify computing
ICCs, there was justification for alternatively defining
interrater agreement as plus or minus one rating. When
the raters later came together to arrive at compromised j
scores for the tests, it was discovered that the vast j
i
majority of disagreements involved the top score on each j
of the two scales? one rater had mostly avoided it as an J
option while scoring initially, but expressed a greater !
understanding of and willingness to use the entire range
171
after the experience provided by the process of
compromising scores. The raters had trained by studying
the manual and by analyzing four of the tests together,
but this preparation was not comparable to the type of j
extensive training and practice that was obtained by all j
i
i
raters before scoring the SCTs and the Rorschachs. It is '
I
expected that experience with a greater number of training\
i
protocols would greatly improve interrater agreement, but j
i
given the current limitations, the plus-or-minus-one-point
agreement appears to be a fair judge of the adequacy of j
interrater agreement for the study. As Table 2 shows,
both coefficients of reliability improved for this
definition of agreement, most dramatically for the
Conceptual overinclusion scale, k = .81, z = 10.38,
p <.001, and with a more modest elevation of the Richness
coefficient to just below the "good" range, k = .58, z =
5.90, p <.001. The compromised scores benefitted from the
process of negotiation and increased understanding, and
were therefore submitted for data analysis with confidence
in their measurement of OST performance.
Administering and scoring the Rorschach. The
Rorschach was administered by the investigator, a female
marriage and family therapist, and a female postdoctoral
psychological assistant. The assistants also coded the
I
Sequence of Scores according to standard procedures
172
explicated by Exner (1985, 1986). In the initial Response
or Association phase, the subject is presented with 10
successive inkblots and asked, "What might this be?"
Responses are written down verbatim with a minimum of
comment by the examiner. In the following Inguiry phase,
the objective is to gain whatever additional information
is necessary to score the response accurately. After the
protocol has been collected, the Sequence of Scores is
listed, which actually consists of response codes that are
designed to record, shorthand-style, the various
components that exist in a response.
To establish interrater reliability, a fourth
experienced rater (a licensed male psychologist) scored a
randomly selected group of eight protocols, some of which
were scored by each of the other three raters. Interrater
agreement was defined by the perfect match of every
element in a response coding category (including agreement
on the absence of any codes), and was calculated by
percentage of agreement in the following categories:
Developmental Quality, Determinants, Form Quality, Z-
scores, and Special Scores. Although this procedure
ignores chance agreements, there are approximately 8 6
different codes within the above five categories;
moreover, separate determinants can appear in all types of
combinations, and the frequency of Z scores varies by
173
protocol. These conditions, in addition to the stringent
definition of complete agreement within a category,
reduced the distortion due to chance. Interrater
agreement for the Rorschach is an example of a problem
that arises in clinical research, in which important but
infrequent signs of some behavior are rated as present or
absent. There may be agreement on an item's absence,
although the reliability for that item cannot be
calculated. Accordingly, a percent agreement descriptive
statistic is probably the best recourse (Bartko &
Carpenter, 1976). As shown in Table 2, percentage of
agreement among the five categories ranged from 71% to 89%
with a median of 75.5%.
After the codes for all responses were listed in the
Sequence of Scores for each protocol, they were entered
into the RSP computer program provided by Exner (see
Alumni Newsletter, 1988), which calculates the Structural
Summary from these data. It represents the composite of
code frequencies and the numerous percentages, ratios, and
derived scores that are calculated from them, and on which
many interpretive strategies are based (Exner, 198 6). The
following variables were selected to comprise the
Rorschach measure of the ICS construct (see Table 4): W:D
(Economy Index), W+ (integrated wholistic percepts), DO+
and v/+ (meaningful synthesis), M (human movement),
174
Table 4
Hypothesized Rorschach
Integrative Cognitive Style (ICS) Index
Function Positive Variable Code and Cutoff Score
Cognitive effort
1. "High" Economy Index
(i.e. more effort, less economy)
W:D ratio > 1:1.5
Integrative strivings
2. High integrated wholistic percepts
W+ > 3
Higher Conceptualization, Imagination
3. High human movement percepts
M > 4
Perceptual Conventionality
4. Low commonplace form features
FOo/R (F£o%) < 1:1.35
Cognitive Initiative and Complexity
5. High meaningful synthesis
D0+ and v/+ (DOsvn) > 7
6. High organizational activity
Zf > 12
7. "Overincorporative" information process
ing (i.e. thorough perceptual scanning)
Zd > 3.0
Unusual Combinative Ideation
8. High Level I merged percepts
and “loose associations"
INCOM and FABCOM (COM) > 1
Note. R = total number of responses per protocol.
175
F0o% (percentage of commonplace features) , Z_f
(organizational activity), Zd (overincorporation), and
Special Scores INCOM and FABCOM (merged percepts and
J
"loose associations"). As discussed in full in Chapter
II, these variables were considered to be congruent with
key aspects of the ICS construct, such as the organization
of wholes and category breadth. (Because of the necessary
reference to a large number of abbreviations and Rorschach
codes in the present study, a summary table of
abbreviations and codes has been provided in Appendix C.)
Table 4 groups the variables by function and
summarizes empirical expectations in reference to the
magnitude and directionality of each variable's value by
means of cutoff scores. These expectations were based on
an examination of various descriptive statistics of
normative data supplied by Exner (Alumni Newsletter, 1989)
as well as pertinent research findings regarding other
special groups (reviewed in Exner, 1986). To determine
critical cutoff scores, the shape of the normative
distribution for each Rorschach variable was examined. If
it approached a normal curve, then the mean was selected
as the most likely locus of the split score. If either
skewness or kurtosis was elevated, the mode became a
likely cutoff (the derivation of cutoff scores from
normative data is detailed in the following chapter).
176
Protocols were scored for these Rorschach ICS
variables in three ways. First, simple frequency tallies
were recorded for all eight variables of interest on each
protocol. They were treated as continuous variables for
the initial comparisons between high and low ED groups,
and for a comparison of these groups with the normative
sample. Second, the variables were treated as dichotomous
in order to focus the Rorschach ICS variables into an
experimental ICS Index. The variables were recoded for
each protocol as either above or below their corresponding
critical cutoffs. Directionality of each cutoff in terms
of identifying ICS is reported on Table 4. Finally,
tallies were made of the number of variables for each
protocol that met the cutoff criteria stipulated for ICS.
The sum of variables that were positive for criterion
comprised a Composite ICS Index (Cl) for each protocol.
Data Analysis
To test the null form of Hypothesis 1, that there
would be no difference between the high and low ED groups
in integrative cognitive processing as measured by the
theoretically and empirically defined ICS variables, a
preliminary examination of descriptive statistics for
Rorschach variables selected from those in standard use in
the Comprehensive System (Exner, 1986) was conducted, as
well as for variables derived for the specific purposes of
177
the current study. ED group means on each individual
variable were compared with independent t-tests; they were
also evaluated in reference to Exner*s normative sample
(Alumni Workshop, 1989). The normative data provided the
baseline for the critical cutoff scores of the
experimental ICS index and for subsequent data analysis
and interpretation of findings. Differences between groups
on the Cl variable were also evaluated by a t-test. Chi-
square tests were then conducted with the frequency data
of each Rorschach study variable based on the critical
cut-off scores of the ICS Index. To test the difference
between ED groups on the OST scales, independent t-tests
of group means were conducted.
To test the null form of Hypothesis 2, that the ICS
criterion variables did not discriminate between high and
low ED groups, a Multiple Discriminant Analysis was
conducted to assess the frequency of predicted group
membership as compared to actual group membership. Prior
to the evaluation of overall classificatory effectiveness,
the discriminant function and its associated weights and
loadings were examined to enhance understanding of group
differences, and if possible to help explain which of the
ICS variables best differentiated the groups. Although
the findings from a discriminant analysis of small groups
should be considered with caution, the value of
178
simultaneous examination of variables, and the aptness of
this multivariate approach for the purposes of the current
study, warranted its inclusion in the research design.
To test Hypothesis 3, whose research form was
equivalent to the null prediction that high and low ED
groups would be comparable in terms of the degree to which
an analytic cognitive processing style (FI) was present,
independent t-tests were used to compare raw score means.
Frequency distributions for each group were also compared
to the combined group quarters with a chi-square test to
indicate whether either group had a disproportionate
loading of highest or lowest scores. Finally, GEFT
frequency distributions were computed for the two groups
in comparison to the GEFT normative sample*s quartile
points. If each ED group were no different from the
norms, one would expect approximately 2 5% of the subjects
in each group to fall into each score category produced by
these quartile points, and a chi-square test was used to
test this hypothesis.
To test the null form of Hypothesis 4, that any
correlations among the ICS measures, or between ICS
measures and the GEFT, were attributable to chance alone,
a correlation matrix of the ICS and FI variables was
constructed. The ED mean score for each group was also
included as an adjunct representation of the relation
179
between cognitive style and ego development. The ED
correlations were not intended to support a presumption of
linearity for cognitive style throughout the developmental
sequence. As discussed in Chapter 1, discontinuities of
many character traits produce intermediate curvilinear
relationships as development progresses, becoming more and
then less predominant. Interest in the correlations was
confined to the strength of the relationship that could be
inferred between ICS and the highest levels of ED.
As support for convergent and discriminant validity
of the research criterion measures, the entries in the
matrix should ideally have the following properties:
reasonably high and statistically significant correlations
between two or more different measures of the same trait—
in this case the multiple measures for ICS— and low
correlations between measures of unrelated traits— i.e.
between ICS and FI, or analytical cognitive style.
180
CHAPTER IV
Analysis and Interpretation of Results
In overview, the findings of this study are generally
consistent with the predictions of the four research
hypotheses. First, high and low ED groups are dissimilar
in relation to ICS, as measured by the theoretically
derived ICS variables. Second, these theoretically
derived variables function as effective discriminator
variables for the purposes of classification of high
versus low EL. Conversely, analytic cognitive style (FI)
as measured by the GEFT does not significantly
discriminate between groups, and neither group received
high scores on the GEFT. Finally, there is some support
for the prediction of both convergent validity among the
ICS measures and divergent validity between the ICS and
GEFT measures. Data analyses performed to test the
hypotheses are presented below, supplemented in each case
by discussion regarding the specific interpretation of
each of the separate findings. In order to follow the
findings as reported, the reader may wish to review the
description of the variables and their abbreviations in
181
the preceding chapters, or to refer to the Summary of
Abbreviations and Codes (Appendix C).
Ego Level and Integrative Cognitive Style
(Hypotheses 1 and 2)
Hypothesis 1
Hypothesis 1 stated that the high ED group would
exhibit more prominent ICS characteristics relative to the
low ED group, and an initial analysis of descriptive data
(see Table 5) revealed that the high ED group gave
significantly more mean responses for four of the eight
Rorschach ICS Index variables than did the low ED group,
namely (a) integrated wholistic percepts or W+, t(38) =
-3.26, p<.01 (M = 4.05 vs. 1.50, respectively); (b)
meaningful synthesis or DOsyn. t(38) = -4.54, p<.001 (M =
9.25 vs. 4.85); (c) human movement or M, t(38) = -3.52,
P < . 0 1 (M = 5.40 vs. 2.85); and (d) organizational activity
or Zf, t (30.7) = -2.30, p<.05 (M = 16.20 vs. 11.45).
A more detailed examination of the Rorschach
descriptive data for the high (h) and low (1) ED groups in
relation to the normative sample (n) is informative. Two
preliminary comments will provide an orientation for this
inspection. First, Table 5*s column 1 variables comprised
the basic elements for the eight experimental Rorschach
ICS index variables. Three column 1 variables were used
directly as ICS indices without any tailoring or
comparisons— namely M, .Zf, and Zd. The remaining column 1
182
Table 5
Descriptive Statistics of Selected Rorschach Variables
for Hiah and Low Ego Development (ED) Groups
Compared to Normative Data
Vara Min Max Mean SD Med Mode Skew Kurt Varb Mean SD ICSindex'
R Cl
h 14 80 26.75 16. 09 22.50 16 2.07 5.55 h 5. 30 2 . 08
1 14 37 21. 50 6.86 19.50 14 0.93 -0.16 1 3.15** 1. 57
n 14 38 22 . 67 4.23 23 23 0.54 1.37
w W:D
h 2 39 10 . 95 8. 70 8 8 2 . 28 5.57 h 2 .15 3 . 08
1 3 19 8 . 25 4. 17 8 4 0.91 0.76 1 1.82ns 2. 60
n 3 20 8. 55 1. 94 9 9 1. 28 8.87 W:D > .66
D W+
h 0 32 11.40 8.51 9.50 5 0.86 0.32 h 4.05 2 . 86
1 0 19 8.25 4 . 67 7 11 0.38 0.03 1 1.50** 2 . 01
n 0 22 12 . 89 3.54 13 14 -0.38 1.02 W+ > 3
D£+ DOsyn
h 5 16 8 .85 2.37 8 7 1.41 3.40 h 9.25 2.57
1 0 12 4.80 3.47 4 2 0.70 -0.52 1 4.85*** 3 . 48
n 2 13 7.31 2.16 7 6 0 . 28 -0.43 DQsyn > 7
D0V/ +
h 0 3 0.40 0.82 0 0 2.26 4.90
1 0 1 0. 05 0.22 0 0 4 . 47 20
n 0 2 0.41 0.66 0 0 1. 35 0.53
M
h 1 9 5.40 2.30 5.50 3 -0.03 -1. 03
1 0 10 2.85** 2.28 2 2 1. 63 4 . 12
n 1 9 4.30 1.92 4 3 0.52 -0 . 73 M > 3
FOo F0o%
h 3 29 13.55 6.60 12 10 1. 06 1.14 h 0.54 0. 19
1 8 23 12.20 3 . 79 11 11 1. 60 2 . 53 1 0.58ns 0. 11
n 7 29 16.99 3.34 17 17 0 . 04 0.52 F0o% < .74
Zf
h 8 41 16 .20 7.95 14 13 2.07 4 . 52
1 5 22 11.45* 4.67 11 10 0.75 0.37
n 5 23 11.80 2.59 12 12 0.26 1. 45 Zf > 12
Zd
h -22.50 19 1.30 8.52 3.25 3.50 -1. 02 3 .10
1 -9.50 7 . 50 -1.25ns 5.48 1.25 -5.50 -0.23 -1. 33
n -6. 50 9.50 0.72 3.06 0.50 -1.00 0.48 0. 09 Zd > +3
INCOM COM
h 0 8 1. 95 1.99 1 1 1.69 3.48 h 2.85 2.50
1 0 5 1. 55 1. 67 1 0 0.96 -0.11 1 1.7 0ns 1.72
n 0 4 0.52 0.65 0 0 1.39 3.56 COM > 1
FABCOM
h 0 4 0 . 90 1.17 0. 50 0 1. 32 1.27
1 0 1 0 .15 0.37 0 0 2.12 2 .78
n 0 2 0 .17 0.41 0 0 2.27 4 .43
Note: High and low ED group means on ICS index variables were compared using Student's t
test. df = 3 0.7 for F0o% and Zf; df = 38 for remaining six variables. h = high ED group
(n = 20); 1 = low ED group (n = 20J ; n = normative sample (N = 700) provided by Exner
(Alumni Newsletter, 1989).
aVar = standard Rorschach codes, summed as frequency tallies for each protocol. R = total
number of responses per protocol.
bVar = ICS variables derived from combinations of Rorschach variables listed in column one.
Cl = Composite ICS Index (sum of eight Rorschach ICS variables scored positive at critical
cutoffs) . DOsvn = sum of synthesized DQ (DO+ and v/+) . F0o% = FOo/R. COM = sum of INCOM
and FABCOM.
'Critical cutoffs for ICS Index (used in subsequent analyses) were derived from
distributions of normative sample.
*p <.05. **p <.01. ***E <.001.
183
variables were combined in several ways to constitute the
other five hypothesized ICS indices, listed across from
them in column 10.
Second, normative data for column 1 variables were
useful not only to enhance a preliminary understanding of
the obtained research results, but to guide the a priori
establishment of critical cutoff scores (shown in the far
right column of Table 5) for a more definitive analysis of
ED group performance (discussed later in this section).
To examine differential high and low ED group performance
in relation to normative data, it was useful to view the
relevant criterion variables first as continuous, and then
as dichotomous. As continuous variables, they could be
compared to the whole composite of descriptive normative
data, facilitating the discernment of directional trends
and degrees of difference among the three samples. The
independent t-tests of mean group differences for each
variable, of course, were also based on an assumption of
continuity.
However, t-tests cannot ascertain the salience of the
obtained values for each ED group— that is, they do not
address the issue of their substantive significance in
relation either to the normative distribution or to
specific scores derived from past research findings. By
defining cutoff scores based on these criteria, a
184
meaningful ICS index of dichotomous Rorschach variables
was created whose values were not relative to the
distribution of ICS characteristics for these particular
subjects. In the following analysis of findings in
relation to the norm, two types of information are
therefore included: high and low ED groups are inspected
in reference to a variety of normative descriptive data,
and the normative data are separately examined in order to
explain the rationale for the cutting scores.
For whole responses (W), the low ED group's mean was
similar to the normative mean (M = 8.25 vs. 8.55,
respectively), whereas the high ED group's mean (M =
10.95) was more than 2 points above the norm. A
comparison of the modes for W reveals that the most
frequently occurring score among low ED subjects was below
the mode of the normative sample (md = 4 vs. 9,
respectively) while the high ED group's mode (md = 8) was
closer to the norm.
Overall, the high ED group appears to compare
favorably to the normed data and to exceed the low group
on the W variable, but the profile is somewhat altered
when the number of whole responses is compared to the
number of detail responses (D). The normative sample gave
an average of approximately 1.5 more Ds (M = 12.89) than
Ws. However, both high and low ED group mean Ds (M =
185
11.40 and 8.25, respectively) more closely matched the
means of their corresponding W responses.
For the ICS Index W:D critical cutoff, the 1:1.5
normative ratio of means was selected, a choice further
supported by the similarity of the ratio of the modes for
the W and D normative distributions, as well (i.e. 9:14,
or 1:1.6). However, there was no significant mean
difference between ED groups based on this cutoff.
The relationship of W to D is more accurately
evaluated within the context of R— the total number of
responses per protocol. The more or less equal W:D ratio
appears to have been influenced by the unusually large R
range (range = 14-80). The possibilities for different
whole responses diminishes as R increases, and is likely
to be exhausted when R exceeds 3 2 (Exner, 198 6), resulting
in a proportionally higher number of D responses.
The mean number of W+ responses for high and low ED
subjects (M = 4.04 and 1.50, respectively) are not only
significantly different from each other, but appear to be
well differentiated from the normed cutoff score of 3, as
well. The cutoff score for W+ was derived from a
comparison of the percentages of the three possible
location answers and the percentages of the four possible
DO answers in the normative sample. Thirty-eight percent
of R were whole responses and 34% of R were D0+ or y/+.
186
Applying the percentage of one variable to the raw score
average of the other variable produces an average combined
W+ response of 2.9, or a cutoff score of 3.
The most apparent difference between ED groups in
relation to the normative data for the D0+ and v/+
variables is that low ED subjects scored well below the
norm, whereas high ED subjects scored as high as or higher
than the norm (regardless of whether the mean, median or
mode was selected as the appropriate indicator of central
tendency; see Table 5). For the DOsvn cutoff score, a
value of 7 was first chosen for D0+. after comparing the
mean, median and mode. By examining the shape of the
DOv/+ distribution, and considering that 69 percent of the
normative sample had Dav/+ values of 0 (Alumni Newsletter,
1989), a value of 0 for this variable was chosen, summing
to a total of 7 for the DOsyn cutoff score.
In reference to the normative data for M, the high
and low ED present a profile that is similar to that of
the D£) variables. That is, low ED scored below the norm,
whereas high ED subjects scored as high or higher than the
norm (in comparison to means, medians, and modes; see
Table 5). The cutoff score of 3 for M is based on a
normative mean split, representing a compromise between
the influence of the standard deviation on the mean (M =
4.30, SD = 1.92) and the lower mode of 3.
187
The cutoff score for F0o% (.74) was obtained by
dividing 17 (i.e. the equivalent value of the normative
distribution's mean, median, and mode for FOo) by the mean
R of 23. Because a subject who tended toward ICS
information processing might exhibit proportionally more
overelaborated, unusual, and minus answers than ordinary
FO answers, the positive criterion for ICS was set at F0o%
<.74. In comparison to the normative FOo data, both high
and low ED groups showed a lower absolute frequency of
FOo. An adjustment for R (FOo%) did not result in a
significant difference between the groups (M = .54 vs. M =
.58, respectively).
The similar magnitude of mean, median, and mode in
the normative sample also determined the cutoff score of
12 for Zf. A comparison of high and low ED groups with
the normative data reveals that the low ED subjects score
somewhat below the normative mean (M = 11.4 5 vs. 11.80,
respectively), with a lower median and mode for the group
as a whole. High ED subjects give substantially more Zf
responses (M = 16.20) than either of the other two groups,
with a greater proportion of answers clustering around
their group mode of 13, and an extremely broad range of Zf
responses as a group (range = 8-41).
The Zd cutoff score of 3 was derived from a body of
literature that has investigated the characteristics of
188
subjects whose scores deviate from the range of -3.0 to
+3.0, where 7 0% of the normative sample fall (see Exner,
1986). Subjects who obtain >+3.0 are labeled
"overincorporators," and subjects who obtain <-3.0 are
labeled "underincorporators." An examination of the Zd
modal values obtained by the high and low ED groups does
reveal a striking disparity between them, on either side
of the +/- 3.0 range (3.50 vs. -5.50, respectively).
However, the disparity between group means is diminished,
although the direction of the difference remains the same
(M = 1.3 0 vs. M = -1.25, respectively, in contrast to M =
.72 for the normative sample). It is likely that the
anomalously low scores at the end of each ED group's score
range, especially for the high ED group, account for the
depressed mean values. In general, the high ED group
obtained higher scores and the low ED group obtained lower
scores overall in comparison to the normative sample.
The final cutting score to be determined was for COM,
the combination of INCOM AND FABCOM. Their normative data
indicate J-curves for both variables, with the most
frequently occurring value for either variable being 0.
In fact, 54% of the normative sample gave no INCOM
responses, and 84% gave no FABCOM responses. Given these
circumstances, a cutoff score of 1 was considered
conservative. In a comparison of both high and low ED
189
groups with the normative sample's distribution of INCOM
and FABCOM data, high ED subjects tended to give the
highest number of these two responses in comparison both
to normative sample subjects (INCOM M = 8 vs. 4, FABCOM M
= 4 vs. 2, respectively) and to low ED subjects (INCOM M =
5, FABCOM M = 1). Perhaps because the frequency of INCOM
answers in the low ED group was also elevated in relation
to the normative data, however, difference between high
and low ED group means was not enough to achieve
significance on the combined COM variable (M = 2.85 and
1.70, respectively).
A related expectation of Hypothesis 1 was that ED
groups would differ when compared by the number of
Rorschach variables scored positive for the ICS criterion,
defined by the cutoff scores established for each variable
as described above. As shown on Table 5, the groups
differed significantly on this Composite Index (Cl),
averaging a tally of 5.30 variables for high ED subjects
and 3.15 for low ED subjects, p <.01.
To analyze the frequency of high versus low responses
on these eight Rorschach ICS Index variables ("high" in
relation to the cutting scores described above defined to
be in the predicted direction for ICS), separate chi-
square analyses were performed. As shown in Table 6, high
ED subjects exhibited a significantly greater percentage
190
Table 6
Chi-sauare and t-test Analyses of Hypothesized ICS Indices
for High and Low Ego Development (ED) Groups (N = 40)
High ED Low ED
ICS Variable (n = 20) (n = 20)
N % N % X2(l)
Rorschach8
W: D > . 66 13 65 13 65 0.00ns
W+ > 3 12 60 3 15 8.64**
DQsyn > 4 15 75 5 25 10.00**
M > 4 11 55 4 20 5.23*
FQo% < . 74 17 85 17 85 0.00ns
Zf > 12 15 75 8 40 5. 01*
Zd > + 3 10 50 4 20 4 . 00*
COM > 1 13 65 9 45 1.62ns
0STb
Mean SD Mean SD t(38)
Beh 41.05 16.28 28.35 16.30 -2.47*
0-in 4.20 0.95 2.55 1.28 -4.63***
Rich 2.50 1.57 0.80 1.06 -4.01***
aCritical cutoffs derived from distributions of normative
sample (N = 700) provided by Exner (Alumni Newsletter,
1989). DOsvn = sum of synthesized DQ (D0+ and v/+).
FQo% = FOo/R. COM = sum of INCOM and FABCOM.
b0ST = Object Sorting Test. Beh = Behavioral
overinclusion scale (total objects sorted). 0-in =
Conceptual overinclusion scale (1-5): 1-2 indicate
random, undisciplined categories; 3 indicates logically
delimited categories; 4-5 indicate broader but relatively
consistent, integrated categories. Rich = Richness of
associations scale (1-4): scores indicate increasingly
strong recognition of multiple, original category
dimensions and item inclusions.
*p <.05. **]□ <.01. ***£ <.001.
191 i
of high responses than did low ED subjects on five of the I
eight ICS indices. High ED subjects produced more high-
quality whole responses (W+ >3), X2(l, N = 20) = 8.64, p
<.01, and their total responses reflected a higher, more
i
• P I
complex developmental quality overall (DOsvn >4), X‘(l, N
I
= 20) = 10.00, p <.01. They also used more human movement
responses (M >4), X2(l, N = 20) = 5.23, p <.05, more
frequently engaged in organizational activity (ZJ: >12), j
2£2(1/ N = 20) = 5.01, p <.05, and exhibited a more marked ;
i
tendency to approach a new stimulus field with a thorough,
i
overincorporative processing style (Zd >3, X2(l, N = 20) =|
4.00, p <.05.
The three separate scales of the OST comprise the
remaining ICS indices. Consistent with predictions, the \
I
results of independent t-tests (shown in Table 6) revealed!
significant differences between high and low ED scores: j
high ED subjects sorted a greater number of objects |
i
(Behavioral overinclusion scale M = 41.05) than did low ED^
subjects (M = 28.35), t(38) = -2.47, p <.05. High ED !
subjects manifested a distinct tendency to form broader
but relatively consistent, integrated categories |
i
(Overinclusive scale M = 4.20), in contrast to lower ED j
i
i
subjects' tendency toward vaguer, more undisciplined
categories (M = 2.55), t(38) = -4.63, p <.001. Finally,
high ED subjects showed a more pronounced, richer use of
*
192
multiple and original category dimensions and item
inclusions than did low ED subjects (Richness of
associations scale M = 2.50 and .80, respectively), t(38) !
= -4.01, p <.001. |
Hypothesis 2 i
j
While research hypothesis 1 focused the inguiry j
toward the prediction of degree of integrative cognitive j
processing as a function of ego developmental level, !
hypothesis 2 stipulated that further examination of an ED-
ICS relationship focus attention on the power of ICS j
i
variables to function as predictors of EL. Hypothesis 2 '
F
predicted that the ICS measurement variables would serve |
as effective discriminators of high versus low EL, and a
i
Multiple Discriminant Analysis of ICS variables supported I
I
this expectation in a number of ways. !
1
The obtained statistic of initial importance in a ;
!
discriminant analysis is the discriminant function— the I
I I
i linear equation that maximizes the difference between i
i
weighted group means (i.e. the centroids) while minimizing,
within group differences. As in multiple regression
| analysis, variables on which the ED groups differed were i
I
therefore weighted more heavily than those variables on
which they were more similar, resulting in the best linear
composite of ICS predictor scores. Values pertinent to
the interpretation of the discriminant function appear on
193
Table 7. The eigenvalue of 1.38 represents the ratio of
between groups to within groups variance (sum of squares),
so that its magnitude indicates a good function. The
canonical correlation (Rc) for this two-group study was
equivalent to a point-biserial correlation between the
continuously distributed, weighted discriminant scores for
individual subjects and group membership. The
significance of the discriminant function was tested by
using Wilk's lambda statistic, which is the ratio of
within-groups variance to total variance. It therefore
reflects the percentage of variance in discriminant scores
not explained by group membership: as lambda moves closer
to zero, more of the total variability can be attributed
to between-groups differences. As shown in Table 7, the
value of Wilk's lambda for the function was .42,
distributed as a chi-square (11, N = 40) = 28.12, p <.01,
thereby supporting a rejection of the null hypothesis of
equality of group means.
Since the overall function was statistically
significant, the weights of the individual variables (the
standardized discriminant function coefficients), were
subsequently evaluated for significance. From an
examination of the weights listed by type of variable
(i.e. Rorschach or OST) on Table 7, it is evident that
smaller Wilk's lambda values correspond to those four
194
Table 7
Results of Multiple Discriminant Analysis of
Integrative Cognitive Style (ICS) Differences
Between Low and High Ego Development Groups fN = 40)
Eigenvalue
Sc
Discriminant
Wilk's lambda
Function3
X2(ll)
1. 38 . 76 .42 28.12**
ICS Discriminator Variables6
By type By rank
SDFC Wilk's lambda 1(1/3) CVC
Rorschachc
W: D -.21 .99 . 13ns
W+ .20 .78 10.65**
DQsyn . 61 .65 20.65***
M . 20 .75 12.40*** O-in . 64
FQo% -.09 .99 . 56ns DOsvn . 63
Zf -.50 .88 5.31* Rich . 56
Zd -.02 .97 1.27ns M . 49
COM -.09 .93 2.88ns W+ .45
Beh . 34
Zf . 32
0STd COM . 24
Beh .34 .86 6.08* Zd . 16
0-in . 55 .64 21.49*** F0o% -.11
Rich .27 .70 16.10*** W: D . 05
aR = canonical
—c
correlation.
bSDFC = standardized discriminant function coefficient— i.e. discriminant
weight. CVC = canonical variate correlation— i.e. discriminant loading (the
correlation of each discriminator variable with the function). In interpreting
the direction of the weights, lower scores on F0o% were defined as indicators
of ICS. For remaining variables, higher scores were defined as indicators of
ICS.
cD0syn = sum of synthesized DQ (D0+ and v/+). F0o% = FOo/R. COM = sum of INCOM
and FABCOM.
d0ST = Object Sorting Test. Beh = Behavioral overinclusion scale. 0-in =
Conceptual overinclusion scale. Rich = Richness of associations scale. Group
centroids were -1.14 and +1.14 for low and high ED groups, respectively.
*2 <.05. **£ <.01. ***£ <.001.
195
variables for which the F is statistically significant.
Specifically, the weights corresponding to W+, DOsyn. M,
and Zf indicate that these variables best differentiate
groups in terms of ICS and make significant contributions
to the prediction of group membership. These same
variables were significantly different in preliminary
independent t-tests of group means (shown on Table 5).
Since discriminant function coefficients are
comparable to partial regression coefficients, the
obtained weights reflected the statistically unique
contribution of each ICS variable to the composite
function, over and above any variable not yet considered.
While the weights would be useful in determining which ICS
variables to retain in the set of ICS discriminators, they
would be ambiguous in terms of the substantive nature of
the composite, given that intercorrelations among the
variables could result in one predictor receiving most of
the weight while another predictor received little.
Therefore, canonical variate correlations (CVCs) or
discriminant loadings were computed, reflecting the shared
variance of each ICS variable and the underlying
composite. These correlations represent the total
contribution of any ICS variable to the linear composite
without considering its relation or redundancy to other
ICS variables, and are thus similar to factor loadings.
196
They can be similarly used to interpret the dimensionality
of group differences and to formulate a parsimonious
explanation to account for them. The loadings are listed
by rank in Table 7.
The five variables having the highest CVCs were
conceptual overinclusion (O-in), meaningful synthesis
(DOsvn), richness of associations (Rich), human movement
(M), and abstractive and integrative strivings (W+), CVC =
.64, .63, .56, .49, and .45, respectively. A modest
association was indicated for behavioral overinclusion
(Beh) and organizational activity (Zf), CVC = .34 and .32,
respectively. This ordering of importance of variables
suggests that the ICS dimension is principally defined by
variables that reflect a comprehensive yet cohesive
breadth of cognition (O-in, DOsvn. and W+), marked by
imaginative and creative influences (Rich and M). The
combination of higher level DQ responses overall with a
higher number of W+ answers points to a more sophisticated
form of organizing ambiguous situations; the combination
of this kind of synthesis with ideation indicates more
sophisticated operations overall.
The Beh and Z_f variables are relatively more indirect
measures of integrative strivings. They reflect the
quantity of effort rather than the quality of the result,
in that behavioral overinclusion consists of the number of
197
objects sorted and .Zf consists of the number of organized
Rorschach responses.
In corroboration of preceding data analyses of W:D
and F0o% variables, the discriminant loadings indicated a
negligible contribution from these variables to the ICS
dimension or to the differentiation between high and low
ED groups (CVCs = .05 and -.11, respectively). The means
of the two ED groups for these two variables were quite
close (see Table 5), and chi-square analyses using the
theoretically derived cutoffs did not reveal a
discriminating pattern of scores between groups, X2(l, N =
40) = 0.00 for each variable. These results appear to
indicate that the relative frequencies of whole responses
to detail responses, and of conventional form quality
responses to total unusual, elaborated, or minus form
quality responses, are not salient features of the
definition of integrative cognitive processes. Neither
did the sum of incongruous and fabulized combinative
responses (COM), or the extent to which an incorporative
information processing tactic was used (Zd), make a
substantial contribution to the overall discrimination of
ED groups.
In addition to clarifying the nature of group
differences, the discriminant function allows the
prediction of group membership for each subject in the
198
sample. Discriminant scores for each subject and weighted
group means (centroids) for each group are computed with
the discriminant function coefficients. An individual's
score is compared to the centroids, and group membership
is predicted for whichever group represents the closest
match between the two. This technique was utilized for
confirmatory purposes in this study by comparing the
percentage of correct predictions based on the function to
actual ED group membership. Results concerning the
accuracy of the discriminant function in classifying low
ED subjects and high ED subjects are presented by the
crosstabulation shown in Table 8. Use of the function
resulted in an overall hit rate of 90%; 17 of the 2 0 high
ED subjects (85%) and 19 of the 20 low ED subjects (95%)
were correctly identified. Compared to an a priori
probability of 50% correct classification of subjects, the
9 0% successful prediction represented a highly significant
improvement (using the z-test for the difference between
proportions, z. = 5.06, p <.0001).
Figure 2 provides a graphic representation of the
differentiation of the high and low ED groups by plotting
the distribution of scores around the group centroids
(= 1.14 and -1.14, respectively). There is a single,
extreme outlier in the low ED group: her discriminant
score of 2.12 represented the fifth highest score among
Table 8
Hit Rates Using a Discriminant Function to Predict
Level of Ego Development
Actual group
Predicted
High ED
group
Low ED Total
High ED
n 17 3 20
% 85 15 50
Low ED
n 1 19 20
% 5 95 50
Total
N 18 22 40
Q.
"O 45 55 100
Note. Hit values (in italics) on the diagonal
total 36, or 90% of the cases. The 4 misses thus
account for 10% of the cases. The overall hit rate
significantly greater than chance, z. = 5.06, £<.0001
200
Figure 2
Plot of Individual Discriminant Scores
for High and Low Ego Development (ED) Subjects Around
their Own Group Centroids (D)
5
Low ED High ED
4
Dled = - 1 . 1 4
o 3
c
CD
Z3
cr
CD
O- 2
1
+2 +4 +1 +3 0 4 • 3 • 2 1
Centroids
(The horizontal axis represents both standardized
discriminant scores for individual subjects and group
means of discriminant scores, i.e. centroids, for the low
ED group [Dled] and the high ED group [Dhed] . The vertical
axis depicts the frequency of subjects attaining
particular discriminant scores. The frequency polygons
for the high and low ED groups (n = 20 per group) have
been smoothed out into curves to provide a clear geometric
representation of each group.)
201
subjects of both groups (four high ED subjects achieved
higher discriminant scores of 2.16, 2.40, 2.64, and 3.03),
and thus strongly indicates the presence of an ICS.
Otherwise, the next highest score for a low ED subject was
-.14, and the score distribution for the low ED group
continues to descend to a -2.4 scored by 2 subjects. The
scores for the high ED group produced no outliers, but
rather a more gradual descending of discriminant score
values below the group centroid. The three false
negatives (i.e. those high ED subjects predicted to be low
ED group members) obtained scores of -.01, -.16, and -.23.
Follow-up Analyses Applicable to Hypotheses 1 and 2
When the four lowest variables ranked by magnitude of
CVC values were reconsidered, the conceptual links of ICS
with COM and Zd variables appeared to hold up well under
scrutiny, and possibly to be stronger than the connection
of the ICS construct to the W:D and F0o% variables.
Accordingly, COM and Zd were submitted to post hoc
analyses in an effort to elucidate their relatively weak
contribution to the total discriminant function. Next,
since F0o% was insignificant, alternative methods of
assessing the quality of perceptual accuracy and
conventionality (i.e., FQ) in relation to EL were
attempted. Finally, an unhypothesized Rorschach variable,
202
EA, emerged during post hoc analysis as another possible
discriminator between high and low ED.
COM. COM is a composite variable totaling the sum of
INCOM and FABCOM responses. Examination of frequency
tallies for each of these responses separately by group
revealed that the high and low ED groups gave a similar
number of INCOM responses— that is, responses which
condensed blot details into a single incongruous object
(39 and 31 responses, respectively). In contrast, the
groups differed more in total FABCOM responses— that is,
responses that posit an implausible relationship between
separate objects identified in the blot (18 responses from
10 of the high ED subjects versus 3 responses from 3 of
the low ED subjects).
In the Comprehensive System of Rorschach scoring
(Exner, 1986), INCOM and FABCOM are among six Special
Scores for unusual cognitive processing that are
differentially weighted; INCOM is assigned a value of 2
and FABCOM is assigned a value of 4. In the records of
Exner's normative sample, INCOM responses are the most
common of the six Special Scores, and one or two INCOMs in
the record are generally not considered notable. FABCOM
responses are both rarer and presumably stronger instances
of cognitive slippage or loose associations.
Specifically, 58% of the 561 normative records that
203
contain any of the six Special Scores include an INCOM
response, but only 2 0% of these records contain a FABCOM.
Therefore, the COM variable was recalculated for all
subjects by doubling the obtained frequency of FABCOM
responses before adding it to the INCOM frequency. The
new variable produced significant mean differences between
high and low ED groups (M = 3.75 and 1.85, SD = 1.84 and
3.35, respectively), t(38) = -2.22, p<.05. However,
neither a critical cutoff of 2 points nor a critical
cutoff of 3 points produced any appreciable group
differences (chi-square = .001, p>.10 in both instances),
which suggested that the INCOM variable was masking the
effect of FABCOM responses. When FABCOM was analyzed as a
separate variable, however, differential group patterns
emerged more clearly. Given the normative data, a cutoff
score of >0 was set, and the high ED group gave
significantly more FABCOM responses above the cutoff than
did the low ED group, chi-square (1, N =40) = 5.58, p<.05.
A t-test of means also produced a significant difference
between groups (M = .90 and .15, SD = 1.17 and .37,
respectively), t(38) = -2.75, pc.Ol.
Zd. As reported above, the chi-square analysis of Zd
revealed that a significantly greater proportion of high
ED subjects (50%) had a Zd score above +3.0 in comparison
to low ED subjects (20%), indicating a greater tendency
toward overincorporation among high ED subjects. On the
other hand, the results for Zd seemed less impressive whenj
the scores were treated as continuous data; the differencej
i
between means was nonsignificant, and the discriminant !
f
i
analysis indicated that Zd was of lesser importance than )
most other ICS variables. However, an examination of the I
' i
total Zd score distribution to locate any sample-specific
I
l
anomalies revealed that two high ED subjects had ;
1
remarkably low scores (-14.00 and -22.5), far lower than
i
the next lowest score in their group (-4.5). It may be j
recalled that 70% of the normative sample have Zd scores !
I
that fall within the range of -3.0 to +3.0; the lowest Zd |
i
score in the sample is -6.5 (see Table 5). Recall also j
that Zd is the difference between the actual sum of Z I
i
scores assigned (ZSum) and the best weighted ZSum !
prediction when Zf is known (Zest). Organized Z responses
are weighted according to the type of organization and the1
complexity of the stimuli involved. The two high ED
outliers produced an unusually high number of responses (r !
!
= 80 and 46), and their records show that a sizeable
proportion of these responses were formed by retranslating^
the same whole stimulus several times over into various
single objects for Cards I, IV, V, and VI. However, whole
responses to these cards require the least organizational
effort and are thus easier to deliver, because the cards
205
are the most solid of the 10 figures. They require
minimal synthesizing to be perceived as whole objects, in
contrast to Cards III and X, which are the most broken
figures. Therefore, Z scores assigned to whole responses
depicting a single discrete object for Cards I, IV, V, and
VI are also among the lowest Z scores. The retranslated
stimuli formulated by the two subjects (e.g. "a bat— a
vagina— a chest x-ray— a brain scan— a fossil" for Card I;
"a dead animal— a mountain range— a forest fire— a dancer-
-a bonsai tree" for Card IV) may be quite varied in
content, but they unfortunately produced both a high
frequency of Z and a relatively low corresponding Zsum,
much lower than Zest. The likelihood of this occurrence
would increase if a subject perceived the production of as
many answers as possible to be a demand characteristic of
the testing situation.
Therefore, Zd was retested by dropping out the two
lowest scores from both the high and low ED groups (to
keep the adjustment comparable for both groups). The
means were then significantly different (M = 3.47 and
-.33, respectively), t(38) = -2.21, p <.05, and more
congruent with other distributional data for the
unadjusted high and low groups (e.g. mdn = 3.25 and 1.25,
respectively; see Table 5). An unexpected additional
finding that resulted from this follow-up analysis was
206
that the low ED group also contained significantly more
individuals whose Zd scores were <-3.0 (45% versus 15% of
the high ED group), X2(l, N = 40) = 4.29, p <.05, a result
postulated by Exner (1986) to reflect an
underincorporative cognitive strategy.
FO. Exner (1986) regarded form quality (FQ) to be
one of the most important elements among the structural
data of the Rorschach, measuring it in several different
ways. Because the method of this investigation of
assessing possible FQ differences between ED groups (i.e.
F0o%) uncovered no differences, the data were reexamined
first with two Comprehensive System (Exner, 1986) codes:
Conventional Form, or percentage of the sum of FQo and +
responses (X+%), and Distorted Form, or percentage of F0-
responses (X~%)• Group means were nearly identical and t-
tests were nonsignificant. When the mean percentages of
F0+ and u summed together were compared for high and low
ED groups, the high ED group had a higher absolute
percentage of these scores (24% versus 19%, respectively),
but not enough to achieve significance. Regardless of the
method of assessment, subjects at high and low levels of
ED do not appear to be differentiated by the extent of
their commonplace, reality-oriented use of the form
features of the blots.
207
EB and EA. The Erlebnistypus (EB) was proposed by
Rorschach (1921/1942) to reflect an underlying
preferential response style, depending on whether
individuals are more prone to use inner life or
interpersonal interactions as resources for the
satisfactions of important needs (introversive vs.
extratensive, respectively). People who have neither one
style nor the other (ambitent) were originally thought to
be the most flexible and adaptive, but more recent
research (see Exner, 1986) has indicated that ambitent
subjects have a less consistent and reliable repertoire of
coping behaviors, and are thus more vulnerable to
difficulty in coping situations. An examination of ED
group data pertinent to EB revealed that 50% of the low ED
group and 75% of the high ED group were either
introversive (9 in each group) or extratensive (1 in the
low ED group and 6 in the high ED group). Half of the low
ED group were therefore ambitent, twice as many as the
number of ambitents in the high ED group, but the
difference was nonsignificant.
The data that are used to compute EB styles are the
number of human movement (M) responses in proportion to
the weighted values for chromatic color responses. Beck
(1960) postulated that large numbers of M and SumC
represent the development of both more inner life and more
208
affective experiences, constituting a general broadening
of available resources. The sum of these variables is
termed the Experience Actual (EA). High and low ED groups
differed significantly on the EA variable (M = 8.73 and
4.55, respectively, t(38) = -4.54, p<.0001. When a cutoff
score of 9 was set (based on central tendency measures
from the normative sample distribution), only 1 low ED
subject scored above the cutoff in contrast to half of the
high ED sample, X2(l/ N = 40) = 10.16, p <.001.
Ego Level and Analytic Cognitive Style (Hypothesis 3)
Hypothesis 3 stated that there would be no
significant differences between high and low ED groups in
analytic cognitive processing as measured by the GEFT. A
further expectation was that the scores for both of these
groups would be relatively low in comparison to the
normative distribution. Table 9 presents the results of
two chi-square analyses and one t-test that are uniformly
consistent with both aspects of this prediction. Both
high and low ED subjects numbered disproportionately in
the lowest quarter of the normative distribution— 5 5% of
each ED group. Neither group had 25% of its subjects with
scores high enough to place in the highest quarter of the
normative distribution: only 15% of the high group and 10%
of the low group were so placed. Both high and low ED
group score distributions were significantly different
209
Table 9
Chi-square and t-test Analyses
of the Group Embedded Figures Test (GEFT^ Scores
for High and Low Ego Development fED) Groups (N = 40)
GEFT scores
High
(n =
ED
20)
Low ED
(n = 20)
n % X2(3) n % X2 (3 )
Normative Quarters3
First ( o - 8)
11 55 11 55
Second ( 9 - 11) 3 15 5 25
Third (12 - 14) 3 15 2 10
Fourth (15 - 18) 3 15 9 . 60* 2 10 10.80**
Combined ED Group Quarters6
First ( o -
1)
1 05 7 35
Second ( 2 - 7) 6 30 4 20
Third ( 8 - 11) 7 35 5 25
Fourth (12 - 16) 6 30 4 20 5.63ns
M SD M SD t ( 3 8 )
8 .90 4 . 06 6. 00 6.00 -1.79ns
Normative data for women (N = 242) supplied by Witkin,
Oltman, Raskin, & Karp (1971). Interpretation of scores
is frequently limited to first and fourth quarters (as
signs of field dependence and field independence,
respectively).
Quarters computed from actual score frequencies of all
subjects (N = 40).
* P <.05. <.01.
_
210
from the normative score distribution, X2(3, N = 20) =
9.60, £ <.05 and X2(3, N = 20) = 10.80, respectively, both
P <.01.
When comparison of ED groups on the GEFT was made in
reference to their own data, score distribution was not
significantly different between groups, X2(l, N = 40) =
5.63, ns. Neither was there a detectably significant
difference between high and low ED group mean scores (M =
8.90 and 6.00, respectively). These results support the
prediction of no difference between groups in analytic
cognitive processing (FI).
Convergent and Discriminant Validity (Hypothesis 4^
Hypothesis 4 stated that the OST and Rorschach
measures would correlate more highly with each other
overall than with the GEFT, although some intercorrelation
with the GEFT was expected. The reason for this
expectation was that the two proposed dimensions of
analytic and integrative cognitive style were not
postulated to be completely mutually exclusive, and it was
additionally suggested that low GEFT scores communicated
ambiguous and possibly contradictory information about any
particular subject. A correlation matrix was constructed
in order to examine the data related to these
expectations. The strongest summary data in support of
these predictions is found in the correlations of various
211
individual variables and the Composite ICS Index (Cl) of
Rorschach variables, shown in the first row of the
correlation matrix presented in Table 10. Of the ICS
variables, only Zd and the Behavioral overinclusion scale
fail to correlate significantly with the Cl. The
Rorschach variables that yielded the strongest group
differences in preceding analyses tended also to correlate
significantly with each other (W+, DOsvn. M, and Zf).
When correlations of Rorschach ICS variables and OST
ICS variables were examined, results were both favorable
and occasionally unexpected. The Conceptual overinclusion
and Richness of associations scales were highly
intercorrelated (r = .73, p <.001), and these scales also
correlated significantly with Table 10 the Cl and with
individual variables W+, DOsvn. and Zf. Conversely, there
was a relatively weaker association of the Behavioral
overinclusion scale with other variables, even with the
other two OST scales (r = .21, ns for O-in and r = .32,
p <.05 for Rich). Another unexpected finding was that
neither of the latter OST scales was significantly
associated with M (r = .32 and .22, respectively, both
ns), but rather with COM (r - .47 and .45, respectively,
both p <.01).
The field-independent (FI) analytic cognitive style
variable correlated significantly with only 2 of the 11
212
Table 10
Intercorrelations Among Integrative
Cognitive Style (ICS) , Field Independence (FI) ,
and Eao Development fED) Variables
W: D
Var 2
W+ DOsvn
3 4
M
5
FQo%
6
Zf
7
Zd
8
COM
9
Beh
10
O-in
11
Rich
12
FI
13
ED
14
1. Cl 09
6 6*** 7 9*** 69*** 38* 63*** 25 40** 24
4 9*** 41** 29
4 4**
2 . W:D — 4 5** -03 -17 12 -02 30 01 12 23 31* 16 11
3. W+ — 63*** 3 9**
-29
4 9*** 41**
21 27 41** 42** 36* 42**
4. DQsyn — 78*** -14
7 4***
10 22 23 46** 38** 15
4 4**
5. M — -13 56*** 00 26 24 32 22 10 39**
6 . F0o% — -23 06 -05 -17 01 -02 23 -17
7. Zf — -24 22 25 37* 30* 03 21
8 . Zd — 25 10 12 18 47** 20
9. COM — 02 47** 45** 18 28
1 0 .Beh — 21 32* 15 35*
1 1 .O-in — 73*** 29 58***
1 2 .Rich — 12
4 9* **
13. FI
— 32*
14. ED —
Note. 1. All decimal points have been omitted.
2. Variables 1-9 are Rorschach ICS variables. Cl = Composite ICS Index of variables 2-9 scored
positive at critical cutoffs; DQsyn = sum of synthesized DQ CDO+ and v/+); FOo% = FOo/R; COM = sum of
INCOM and FABCOM. Variables 10-12 are Object Sorting Test scales postulated to measure ICS.
Beh = Behavioral overinclusion scale. O-in = Conceptual overinclusion scale. Rich = Richness of
associations scale. FI = Group Embedded Figures Test score. ED = Sentence Completion Test score.
* p <.05. ** E <.01. *** E <.001.
213
individual ICS variables, W+ (r = .36, p <,05), and Zd (r
= .47, p <.01). The Zd itself correlated only with FI and
W+. This suggests that the thorough, scanning approach to
the field reflected by a high Zd score might be a
concomitant aspect of both analytic and integrative
processes.
The correlation of FI and ED was .32, p <.05. The
impact of this significant association was modulated by
the previous analysis of GEFT scores, which showed that
subjects in both ED groups scored disproportionately in
the lowest quarter of the normed distribution. In
comparison to the relationship of analytic cognitive
processing and higher ED, associations between ED and
several of the ICS variables were stronger, indicating
that they were related to a greater differentiation
between high and low ELs. These ICS variables included W+
(r = .42), DOsvn (r = .44), M (r = .39), and the total
Rorschach Cl (r = .44), all at p <.01. All three OST
scales were significantly associated with EL, as well (Beh
r = .35, p <.05; 0-in r = .58 and Rich r = .49, both
p <.001).
In summary, the intercorrelation matrix displays
multiple data points of confirmation of the convergence of
ICS measures and their relative divergence from the FI
(analytic) measure. It also communicates some information
214
about the correlation of ED with each of these criterion
measures. However, the correlations were regarded as
subsidiary to the more specifically designed analyses
(i.e. the previously presented chi-square, t-test, and
discriminant analyses) of the ED-cognitive style
relationship.
215
CHAPTER V
General Discussion, Conclusions, and Recommendations
The objectives of the current study were
substantially met on both theoretical and methodological
grounds. They originated from dual concerns: (a) the
relative lack of positive construct validation for signs
of advanced ego development (Loevinger, 1966, 1969, 1973,
1076, 1979b, 1979c), and (b) the hypothesis that a more
differentiated ego would evidence an articulated (FI)
cognitive style instead of the global (FD) cognitive style
purported to be more characteristic of developmentally
earlier behavior (Witkin et al., 1962/1974; Witkin et al.,
1954/1972).
In a prior study (MacDonald, 1986) of the
relationship between ED and FDI, the data suggested that
cognitive style becomes increasingly FI only up to a
certain point in ED close to the Conscientious EL,
decreasing thereafter at higher levels of ED. The
consequent problem identified by those findings was that
the presence of FD cognitive style would then have to be
attributed to the high ED subjects, given the structural
216
explanation of FDI theory, although such attribution would
be inconsistent with the notion of higher ED.
The findings of the current study constitute a
disconfirmation of the FDI differentiation hypothesis, and
an initial confirmation of an alternative integrative
cognitive style (ICS) hypothesis, as the more plausible
account of cognitive operations at advanced ELs. Four
specific objectives were met during the course of the
investigation in order to substantiate the assertion that
Witkin omitted integrative processing from his cognitive
style theory and his GEFT measure, although it is the
style typically utilized by high ED individuals. First,
the adequacy of the FDI construct was tested by means of
procedural evidence, and second, the categorical
proposition of the existence of ICS as distinct from FDI
was tested by the same means. FDI theory was found to be
internally inconsistent at several points, whereas the ICS
proposition was shown to be supported from a wide array of
theoretical and empirical research backgrounds (e.g.
Barron, 1957; von Bertalanffy, 1928/19 61; Brown, 197 6;
Cooper & Podgorny, 1976; Dellas and Gaier, 1970; Dykes and
McGhie, 1976; Fuller, 1983; Goldstein and Scheerer, 1941;
Harrow, Himmelhoch, Tucker, Hersh, & Quinlan, 1972;
Hasenfus and Magaro, 1976; Hock, 197 3; Hoppe, 198 6; KeganJ
217
1983; Ktthler, 1969; Langer, 1942; Mednick, 1962; Mervis &
Rosch, 1981; Seeman, 1959, 1989? Wallach, 1970).
Following the conceptual method of testing the ICS
and FDI constructs, an empirical measure of ICS was
constructed. The research design examined the responses
of high and low ED subjects on 11 separate variables and 1
composite variable derived from the Rorschach and the OST.
The variables were postulated to indicate the presence of
integrative processing, defined as a "decided propensity
to synthesize elements that are discrete and often
relatively disparate into complex wholes, resulting in
comprehensive concepts of relative richness, depth, and
breadth" (see Chapter II).
If analytic cognitive processing is a concomitant of
a more differentiated ego as Witkin claimed (e.g. Witkin
et al., 1962/1974), one would expect that subjects at the
high end of ED would achieve relatively high scores on the
GEFT? increasing autonomy is a key feature of the
developmental progression proposed by Loevinger (1976).
Furthermore, the same subjects would likely receive
relatively low scores on the ICS variables, due to the
fact that the variables are specifically weighted to
indicate a less focused separation of details— that is, a
more comprehensive integration of separate details.
However, in sharp contrast to normative data, more than
218
half of the high ED subjects were in the bottom quarter of
the normative GEFT distribution, typically interpreted as
the FD score range. Only 15% were in the normative top
quarter score range commonly attributed to FI.
Failure of high ED subjects to exhibit an analytic
cognitive style cannot be attributed to FD, however,
because the subjects also achieved high overall scores on
measures of ICS ("high" being defined for each separate
variable according to prior conceptual and/or empirical
findings). These ICS measures presume a type of
sophisticated organizational activity quite unlike the
undisciplined, undirected quality of the postulated global
cognitive style. The finding of relatively low FI and
high ICS for subjects at an advanced level of ED is
entirely consistent with the suggestion that ICS may
represent a style that has supplanted FI in high ED
subjects, in contrast to the conclusion of an FI deficit
and a relative lack of autonomy for these subjects.
This possibility receives further support from an
additionally indispensible aspect of the research design,
that of the inclusion of a low ED comparison group. In
addition to a comparison of the scores of high ED subjects
on cognitive style measures to the scores predicted to be
normally expectable, a comparison of these scores to the
ones attained by a group of subjects representing a
219
systematic difference on the ED variable was therefore
possible, as well. The finding of low FI and low ICS
results for the low ED group corroborates the postulated
difference in quality of cognitive processing for high and
low ED, while also supporting the suggestion that two
different cognitive styles may be responsible for low GEFT
(i.e. FI) scores of different subjects.
The study can be regarded as a contribution to the
construct validation of Loevinger1s (1976) theory of ED,
in specific reference to cognitive features at the high
end of the developmental progression. Logical
considerations derived from the already-existing base of
extensive theoretical and empirical confirmation of the ED
construct supported the deduction of expected differences
in the average performance of high ED and low ED subjects j
in integrative cognitive processing. Because the two ED
groups did differ on a majority of the separate ICS
variables in the predicted directions, the ED construct is
supported.
The ICS construct has also received reciprocal
validation from the findings of the present study. The
hypothesized psychological process of ICS was initially
postulated from a consideration of high ED functioning,
and was elaborated by more precise postulates drawn from a
broad range of literature relevant to information
220
processing styles. The translation of the ICS construct
into measurable operations represents yet another instance
of the derivation of empirical hypotheses from theoretical
statements, in the form of test items. Thus, the
differential performance on these ICS measures by high and
low ED groups comprises positive evidence for the latter
hypotheses, as well.
Research hypotheses 1 and 3 specifically addressed
the construct validity of the ED theoretical account of
cognitive processing at advanced ELs. They predicted (a)
a higher level of ICS at high ELs than at low ELs, but (b)
no indication of marked analytic processing at either high
or low ELs. In the section immediately following, the
positive findings from this study in support of these
hypotheses are related to selected pertinent findings in
the literature. Research hypotheses 2 and 4 more
specifically addressed the validity of the ICS construct
and its measure, predicting that ICS variables would have
substantial discriminating power (hypothesis 2),
convergent validity among themselves, and divergent
validity from the GEFT assesment of FI (hypothesis 3).
Literature relevant to the positive evidence for these
hypotheses will also be presented in a subsequent section.
221
Ego Development, and Cognitive Style
The meaning of differentiation was construed by
Witkin to mean independence and increasing separateness,
so that he was required by his own reasoning to oppose
interpersonal skills and interests to the differentiated
personality in general and to differentiated or analytical
thinking in particular. He considered signs of
relatedness as signs of a lack of differentiation (Witkin
& Goodenough, 1981). The obtained evidence for an
association of high ED with features of comprehensive
integrative cognitive strategies (ICS) is incongruent with
Witkin's position that interpersonal competencies and
limited autonomy are related in the same way as are self-
reliance and an impersonal orientation. Much research
with the ED construct reveals that increasing autonomy and
increased appreciation of others coincide at high ELs
(e.g. Loevinger, 1976; Loevinger & Wessler, 1970;
Loevinger et al., 1970), and this ability to span the gap
between self and other may be facilitated by the habits of
mind characterized here as ICS.
Consider the alternative portrayal of development
proposed by Kegan (1982), in which there is perpetual
subject-object balancing. Every developmental stage is an
evolutionary truce between the extent of an individual's
differentiation from his or her environment and the degree
222
of embeddedness, representing a temporary solution to the
perpetual tension between desires for inclusion and for
separateness. The image of a spiral, posits Kegan, can
better portray the constant return to old issues but at
new levels of complexity. Subject and object are
constantly being recomposed; and as the subject continues
to differentiate itself from the world, its relationship
to the world (in striking contrast to Witkin*s view) is
increasingly organized, through its increasing
integrations of the environment. Kegan (19 82) states:
"With each move, [one makes] a better guarantee to the
world of its distinct integrity, qualitatively reducing
each time a fusion [of oneself] with the world, thereby
creating a wider and wider community in which to
participate, to which to be connected" (p. 71). His view
is reminiscent of various theoretical and philosophical
influences introduced in Chapter I (i.e. phenomenology,
Gestalt psychology, and general systems theory).
The results of this study are consistent with
literature from narrower domains that have dealt more
specifically with thought processes. Mervis and Rosch
(1981) have reviewed data indicating that gradients of
representativeness in adult categorizing criteria exist
for color and geometric shape categories as well as for
semantic, linguistic, and psychiatric classifications
223
(Brown, 1976; Lakoff, 1977; Mervis, 1980; Rips, 1975).
Interpretations of "family resemblance" differ according
to the degree of variability and change permitted. The
extent to which a person allows such a change process to
influence cognitive representations may in part account
for adult individual differences in cognitive style, with
more restricted category structures tending to more
stereotyped thinking.
Present findings are also consistent with Goldstein
and Scheerer's (1941) original formulation of the concrete
and abstract attitudes, in which the description of the
two poles that anchor their proposed cognitive dimension
is comparable to descriptions of global and integrative
approaches to stimuli. High ED subjects, like the higher
functioning subjects in Goldstein and Scheerer's
investigatons, showed superior, consciously-directed
organizing tendencies, in contrast to the posited FDI
progression away from the whole.
Research in creative thought processes has frequently
addressed the combination of elements that are remotely
associated to each other (Mednick, 1962); high ED subjects
in the current study reflected this propensity in a higher
than usual number of FABCOM responses (see Viglione &
McMillan, 1990). Fluency and uniqueness are often
associated with creative cognitive processes, as well
224
(Wallach, 1970), and high ED subjects tended to expand
their OST sorting categories and their Rorschach responses
by including more marginally appropriate and/or
statistically infrequent observations than did low ED
subjects. High ED subjects averaged a higher overall
total of Rorschach responses (R) and a greater total
number of objects sorted in comparison to low ED subjects.
The current findings are also congruent in many
respects with results from studies comparing creative and
disordered thought. Dellas and Gaier (1970) found that
creative subjects deployed their attention more widely and
tended not to screen out the irrelevant, much as
schizophrenics have been postulated to utilize a wide
attentional strategy and to make wide associations (Dykes
& McGee, 1976; Lehmann, 1966; Wallach, 1970). The
overinclusion measures associated with thought disorder
and the ideational fluency measures associated with
creativity often involve similar tasks and scoring
methods, but deficits are attributed to high scores of
schizophrenics and creative features are attributed to
samples based on nonclinical criteria.
Given the findings that support the similarity of
creativity and psychosis (see also Andreason & Powers,
197 6), one can ask whether ICS as manifested in the
current study also exemplifies this similarity, or whether
225
there are any data that can suggest a distinction. Barron
(1957) found a strong correlation between a composite of
numerous divergent thinking tests measuring originality
and the W+ Rorschach response; he emphasized that the
response produces a unitary synthesizing image that
integrates diverse stimuli. Allison and Blatt (1963)
characterized W+ answers as expressions of abstractive and
integrative strivings. The findings of the current study
indicate that high ED subjects give a high number of W+
responses in comparison to low ED subjects. Barron (1957)j
i
also emphasized the presence in his high scoring subjects
of persistence, organization, and control over premature
closure, marked by the scanning of all possible stimuli
before a configuration is selected. Similarly, the
subjects in the present study gave significantly more M
responses, which are reflective of delaying operations and
deliberate processing, and higher Zd scores, a variable
that speaks to an unusual effort to integrate diverse
stimuli.
Considered together, none of the above variables are
associated with the Schizophrenic Index of the Rorschach
devised by Exner and his colleagues (see Exner, 1986).
The high scores on the hypothesized variables are more
associated with ego resources than with ego deficits--as
in the hypothesis of maximal utilization and synthesis of
226
information by emotionally mature, "integrated"
personalities (Seeman, 1959? 1989). The implication
derived from the preceding comparisons is that individuals
at the high end of ED reflect signs of effective
information processing and adaptation rather than signs of
disarray or diffusion, as might othrwise be inferred from
an isolated consideration of their GEFT scores.
Measurement of ICS
Rorschach variables. The ICS Index (including both
Rorschach and OST variables) can now be added to other
available measures that have the potential to discriminate
among ELs on the basis of key phenomenological
distinctions. A qualification is that two of the
Rorschach variables are completely nondiscriminating
between the two ED groups: W:D and F0o%. This is
considered a good finding in terms of its theoretical
value and practical utility for reducing the set of
significant ICS indicators. Thus, the presence of an
unusual number of whole responses in relation to the rest
of the record does not appear to matter, in contrast to
the importance of the quality of developmental
sophistication of any response. Both ED groups were lower
than average on the experimental index of perceptual
conventionality (F0o%), but they were not different from
each other. Although no theoretical explanations readily
227
obtain for this result, it nevertheless suggests that
cognitive style differences between ED groups do not
depend on distinctions related to reality testing.
Although two other variables (Zd and COM) were less
effective indicators of ICS than the remaining four (W+,
DQsyn, M, and Zf) , results indicate that they function as
important contributions to the overall discrimination of
ICS. Evidence for this conclusion includes the fact that
both the median and mode for Cl are 6 for the high ED
group, compared to 3 for the low ED group. As it stands
now, the Rorschach ICS Index could be fairly effective
when comprised of the six variables mentioned above and a
criterion that stipulated the identification of ICS when a
record was positive for at least five variables. Seventy
percent of the high ED subjects meet this criterion,
compared to 15% of the low ED subjects. The Rorschach ICS
Index should be used cautiously, especially at this stage
of its development. As an example, if five variables are
positive, the likelihood of ICS is considerable and that
of a false positive is quite low. When four variables are
positive, the presence of ICS is even more substantial,
but the greater possibility of a false positive cannot be
ignored. The possibility of using the six variables as
the nucleus of an eventually refined and expanded ICS
Index seems promising, considering that, of the high ED
228
subjects who had at least five positive variables, 15% had
7 and 15% had all 8 positive variables on their record,
compared to only one low ED subject who had 7 positive
variables.
OST scales. The adaptation of the OST for the
differentiation of various types of overinclusive object
sortings (Harrow, Himmelhoch, Tucker, Hersh, & Quinlan,
1972? Harrow, Tucker, & Adler, 1972? Himmelhoch, Harrow,
Tucker, & Hersh, 197 3) proved amenable to the
investigator's further modification of its Overinclusion
scale. This scale functioned as the most discriminating
variable among all 11 variables selected to duplicate the
ICS construct.
The present study's findings suggest that the general
concept of overinclusion effectively parallels the concept
of integrative processing. Scores 4 or 5 on the modified
Overinclusion scale are specific indices of ICS. A
possible cutoff for the Richness of associations scale is
not as definite. The score of 1 documents a minimal
presence of richness that could normally be expected.
However, a score of 2 or above can be conceptally defended
as indicative of presence of ICS. In fact, the findings
of the present study show that 85% of the low ED subjects
received a 0 or 1 on the Richness scale, compared to only
20% of the high ED group. This finding again suggests a
229
link between creativity and high ED (cf. the description
of allusive thinking on object sorting tasks in highly
adaptive students and their parents, McConaghy & Clancy,
1968).
Methodological comments. It is hoped that the
conceptual preparation and research design of this study
have minimized the possibility of an overcontribution of
methods variance to its findings. As an example of
potential complications, the SCT represents a measure of
resources and is not designed to test for the presence of
pathology— a source of error that one must hope is
randomly distributed. Also, a high score on the SCT
cannot guarantee an unvaried presentation on other
personality variables or behaviors.
The type of "projective" and individualized testing
utilized in this study carry certain potential sources of
external invalidity, especially in reference to reactive
effects of the testing procedures. These contextual
variables include attempts by the subjects either to do
"well" or to sabotage the data; sensitivities to potential
evaluative connotations of the tests; and susceptibilities
to the self-presentation of the examiner. Any future
research designed to include testing of this type can
benefit from a careful consideration of these potential
confounding variables.
230
Conclusions
1. High and low ED individuals are well
differentiated on variables that reflect integrative
cognitive operations— that is, variables distinguished by
features of synthesis, category breadth, and richness of
associations. The constructs of both ED and ICS, as well
as their associated measures, share in the mutual support
of their validity that the current study has established.
2. Two different kinds of cognitive styles can be
distinguished among individuals with low GEFT scores:
global or FD, and ICS. The global holistic attitude is
relatively undeliberate and immature and can be positively
represented by low scores on the ICS measures rather than
by default. Other possible indicators of the presence of
FD suggested by the findings of the current study include
a greater tendency toward an ambitent response style,
lower overall available resources, and a more marked
tendency toward underincorporative organizational
operations.
3. FD and ICS correspond to the lower and upper ends
of the ED continuum.
4. Features of ideational fluency and uniqueness,
and of an attentional strategy that samples an unusually
wide range of available environmental stimuli, similarly
characterize the information processing of individuals
1
231
whose theoretical sources of deviation may vary, including
those identified as creative, thought-disordered, or
developmentally advanced (in reference to EL).
5. Unusual thought must therefore be considered in
context. Strengths can also be detected with assessment
tools that have more frequently been associated with the
detection of pathology. Thus, creativity appears to be an
increasing aspect of higher level thought within the
context of the general advance of ED.
Recommendations for Future Research
Several of the findings of the current study are
worthy of future investigation. In particular, the
following conceptual relationships and methodological
strategies need to be established more clearly;
1. There is a need for a retrospective examination
of Rorschach records of high ED individuals, to identify
any variables throughout the entire range of variables
that can be linked conceptually or empirically with the
overall profile of high ED. Unlike the current study,
which posited a priori definitions of a variable's
presence based on its probable affinity to ICS and on
previous empirical normative data, the retrospective
method would focus on extracting the structure of a high
ED group of subjects from their overall Rorschach
performance. Any variable that was conceptually or
232
empirically linked to high ED, plus any others that
appeared in more than 50% of the records, could be
included in the computer search of positive criteria.
Eventually, the cluster of variables could be retested
prospectively as an experimental index for high ED in
general.
2. The ICS Index needs to be refined.
Nonsignificant and nondiscriminating variables can be
dropped, and Rorschach ICS cutting scores must be
reexamined in a thorough post hoc analysis of the data
from both high and low ED groups. False negatives and
false positives need to be analyzed in hopes of yielding
consistent patterns that may subsequently be incorporated
into the ICS Index criteria. The revised Experimental ICS
Index may then be tested on a new sample and its
generalizability assessed by means of cross-validation to
another sample.
3. If ICS becomes the focus of inquiry, independent
of its postulated relationship to high ED, other
validating criteria for ICS need to be found against which
the experimental Index can be tested.
4. Alternative methods for assessing an analytical
orientation at high ED need to be explored, in order to
clarify whether an analytic capacity has been either
undeveloped or lost at these ELs, rather than replaced by
233
an alternative stylistic preference. Perhaps analytical
instructions could be alternated with integrative
instructions or no instructions. Another possibility is
the administration of an untimed GEFT, which may allow the
latent presence of analytical processing to emerge
although it is not ordinarily deployed spontaneously (see
Kepner & Neimark, 1984).
5. To complement the approach used in the current
study of treating the ED variable as categorical for the
purpose of inferring significant differences between high
and low ELs, the vicissitudes of cognitive styles across
the developmental continuum can be addressed by treating
ED as a continuous variable. The nature of the relation
between ED and ICS or between ED and FI can be elucidated
by determining whether the regression of cognitive style
on ED is linear or curvilinear? and if curvilinear, by
determining its specific form. A larger sample than that
utilized for the current study, one that is heterogeneous
in respect to ELs, is needed to tap the postulated
curvilinear trends in analytic processing and monotonic
trends in integrative processing.
6. The psychometric relationship between the
dimensions of analytic (FI) and integrative cognitive
processing needs to be determined, in terms of possible
unidimensionality, bipolarity, or orthogonality (versus
234
positive or negative correlation). These relationships
are likely to vary at different ELs.
The final comments of this study are directed toward
those issues of general significance that underlie its
specific lines of inquiry— namely, the interplay between
differentiation and integration and the phenomenon of
deviation from the norm. Together, they represent a
natural human ambivalence towards difference versus
"belonging." Although these issues are of broad
intellectual interest, they also inform the curiosity and
interest of the clinician who wants to know the person who
has come seeking his or her help, and who is often
gratified by the detection of "false positives" and "false
negatives."
In the fields of mathematics and biology, the entity
that one differentiates is derived from a whole, and the
whole is restored in the process of integration. Although
it is certainly prudent to avoid overextended and
unexamined metaphors in scientific research, the
transferred meanings of these two processes through
cultural history can be of legitimate interest and
heuristic value. Thus integration implies the renewal and
i
restoration of an individuals integrity, in a manner that
is at once a healing and a fresh beginning. It thus
235
functions metaphorically as a central concern for the
discipline of counseling psychology.
In the clinical practice of counseling psychology,
development and its associated reciprocal operations of
differentiation and integration can be of great value in
understanding clients* current positions and how they move
to new developmental levels. Ivey (198 6) refers to such a
practice as developmental therapy; he sees development as
a holistic relational enterprise, and the participation of
therapy in that process as productive of creative
contradictions and new syntheses. His views are consonant
with Kegan's (1982), who suggests that therapists look to
instances of "natural therapy" as the intellectual
foundation for practice. Natural therapy, proposes Kegan,
consists of those relations and human contexts that
spontaneously support people through the process of growth
and change. The natural facilitation of development
comprises a life history of the holding environment
described by Winnicott (1965) in reference to the infant.
Kegan proposes that people are held in qualitatively
different ways through an evolutionary state of
embeddedness throughout the life cycle. Psychological
support can be evaluated according to the model of the
intrauterine environment (Kegan, 1982):
. . . [it is the] provision of a medium in
which the growing organism can thrive . . .
the framework . . . for a kind of lifelong
'psychological amniocentesis,' by which
the quality of holding environments, their
capacity to nourish and keep buoyant the
life project of their 'evolutionary
guests,1 might be assessed, (p. 257)
The particular strength of an integrative cognitive
style is in the identification of far-ranging and perhaps
even unusual commonalities among items of interest. If
the ultimate items of interest are human individuals, then
the assets of an ICS approach to phenomena can also teach
theoreticians and research designers to be mindful of
their reasons for discriminating and separating among
groups. For instance, findings of gender and racial
differences may be distorted or exaggerated by their
unavoidable societal context, either during the research
or after findings have been disseminated.
Interestingly, the negative impact of such
consequences are located by Gilligan (1978) and other
feminist psychologists in the very language of growth and
development that has informed the concerns of this study.
The stereotypically male overemphasis on differentiation
is favored with attributions of continuing development,
whereas the stereotypical female overemphasis of
integration is depicted in terms of dependency and
immaturity (Kegan, 1982). In an integrative approach to
phenomena, the stance of isolation versus relation is not
237
pertinent. Psychological evolution is intrinsically about
both differentiation and integration, and it thus provides
a useful model to guide one's consideration of the goals
that are ultimately intended by one's research efforts.
238
REFERENCES
Achenbach, T. M. (1978). Research in developmental
psychology: concepts, strategies, methods. New York:
The Free Press.
Adcock, C., & Martin, W. (1971). Flexibility and
creativity. Journal of General Psychology. 85, 71-76.
Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., &
Sanford, R. N. (1950). The authoritarian
personality. New York: Harper & Row.
Al-issa, I. (1972). Stimulus generalization and
overinclusion in normal and schizophrenic subjects.
Journal of Clinical Psychology. 39. 182-186.
Alumni newsletter (1987). Asheville, NC: Rorschach
Workshops.
Alumni newsletter (1988). Asheville, NC: Rorschach
Workshops.
Alumni newsletter (1989). Asheville, NC: Rorschach
Workshops
Anderson, C. C., & Cropley, A. J. (1966). Some
correlates of originality. Australian Journal of
Psychology. 18. 218-229.
Anderson, R. C., & Ortony, A. (1975). On putting apples
into bottles— A problem of polysemy. Cognitive
psychology. 7., 167-180.
Andreasen, N. J. C., & Powers, P. S. (1975). Creativity
and psychosis. Archives of General Psychiatry. 32.
70-73.
Ansbacher, H. L., & Ansbacher, R. R. (Eds.). (1956).
The individual psychology of Alfred Adler. New York:
Basic Books.
Arieti, S. (1976). Creativity: The magic synthesis.
New York: Basic Books.
239
Asch, S. E., & Witkin, H. A. (1948). Studies in space
orientation. I. Perception of the upright with
displaced visual fields. Journal of Experimental
Psychology. 38. 325-337.
Avery, R. R., & Ryan, R. M. (1988). Object relations
and ego development: Comparison and correlates in
middle childhood. Journal of Personality. 56, 547-
567 .
Bakeman, R., & Gottman, J. M. (1986). Observing
interaction: An introduction to sequential analysis.
Cambridge: Cambridge University Press.
Baldwin, J. M. (1902). Social and ethical
interpretations in mental development. New York:
Macmillan. (Original work published 1897)
Barron, F. (1954). Complexity-simplicity as a
personality dimension. Journal of Abnormal and Social
Psychology. 48. 163-172.
Barron, F. (1957). Originality in relation to
personality and intellect. Journal of Personality.
25. 730-742.
Barron, F. (1963a). Creativity and psychological
health: Origins of personality and creative freedom.
Princeton, NJ: Van Nostrand.
Barron, F. (1963b). The needs for order and for
disorder as motivation in creative activity. In C . W .
Taylor & F. Barron (Eds.), Scientific creativity: Its
recognition and development (pp. 153-160). New York:
Wiley.
Barron, F. (1972, June). The creative personality: Akin
to madness. Psychology Today, pp. 2, 42.
Barron, F., & Harrington, D. M. (1981). Creativity,
intelligence, and personality. Annual Review of
Psychology. 32. 439-476.
Barron, F., & Welsh, G. S. (1952). Artistic perception
as a possible factor in personality style: Its
measurement by a Figure-Preference Test. Journal of
Psychology. 33. 199-203.
240
Bartko, J. J., & Carpenter, W. T. (1976). On the
methods and theory of reliability. Journal of Nervous
and Mental Disease. 163. 307-317.
Beck, S. J. (1960). The Rorschach experiment; Ventures
in Blind Diagnosis. New York: Grune & Stratton.
Behrends, R. S. (1986) . The integrated personality:
Maximal utilization of information. Journal of
Humanistic Psychology. 26. 27-59.
Berkowitz, L. (1957). Leveling tendencies and the
complexity-simplicity dimension. Journal of
Personal itv. 25., 743-751.
Bertalanffy, L. von. (1961). Modern theories of
development. New York: Harper Torchbook. (Original
work published 1928)
Bieri, J., Bradburn, W. M., & Galinsky, M. D. (1958).
Sex differences in perceptual behavior. Journal of
Personality. 26., 1-12.
Blasi, A. (1972). A developmental approach to
responsibility training. Unpublished doctoral
dissertation, Washington University, St. Louis.
Blasi, A. (1976). Concept of development in personality
theory. In J. Loevinger (with the assistance of
Augusto Blasi), Ego development (pp. 29-53). San
Francisco: Jossey-Bass.
Blasi, A. (1983). The self and cognition: The roles of
the self in the acquisition of knowledge, and the role
of cognition in the development of the self. In B.
Lee & G. C. Noam (Eds.), Developmental approaches to
the self (pp. 189-214). New York: Plenum Press.
Blatt, S. (1986). Where have we been and where are we
going? Reflections on 50 years of personality
assessment. Journal of Personality Assessment. 50.
343-346.
Blatt, S. J., & Allison, J. (1963). Methodological
considerations in Rorschach research: The W response
as an expression of abstractive and integrative
strivings. Journal of Projective Techniques. 27. 269-
278.
241
Bloom, B. S. (1963). Report on creativity research by
the examiner's office of the University of Chicago.
In C. W. Taylor & F. Barron (Eds.), Scientific
creativity; Its recognition and development (pp. 2 51-
264). New York: Wiley.
Bloomberg, M. (1967). Inquiry into the relationship
between field independence-dependence and creativity.
The Journal of Psychology. 67, 127-140.
Bloomberg, M, (1971). Creativity as related to field-
independence and mobility. The Journal of Genetic
Psychology. 118. 3-12.
Bogen, J., & Bogen, G. (19 69). The other side of the
brain: III. The corpus callosum and creativity.
Bulletin of the Los Angeles Neurological Society. 34,
191-220.
Borders, L. D. (1989). Developmental cognitions of
first practicum supervisees. Journal of Counseling
Psychology. 36. 163-169
Broughton, J. (1989). Ego and ideology: A critical
review of Loevinger's theory. In C. Power & D.
Lapsley (Eds.), Self, ego and identity (pp. 179-208).
New York: Springer.
Brown, R. (1976). Reference: In memorial tribute to
Eric Lenneberg. Cognition. 4., 125-153.
Bruner, J. (1985). Actual minds, possible worlds. New
York: Cambridge University Press.
Busse, T. (1968). Establishment of the flexible
thinking factor in fifth-grade boys. Journal of
Psychology. 69. 93-100.
Cameron, N. (1939). Schizophrenic thinking in a
problem-solving situation. Journal of Mental Science.
85. 1012-1035.
Candee, D. (1974). Ego developmental aspects of new
left ideology. Journal of Personality and Social
Psychology. 30. 620-630.
Cattell, R., & Warburton, F. (1967). Objective
personality and motivation tests. Chicago:
University of Illinois.
242
Chase, W. G., & Simon, H. A. (1973) . Perception and
chess. Cognitive psychology. 4, 55-81.
Claridge, G. S. (1972). The schizophrenias as nervous
types. British Journal of Psychiatry. 121. 1-17.
Cooley, C. H. (1964). Human nature and the social
order. New York: Schocken. (Original work published
1902)
Cooper, L. A., & Podgorny, P. (1976). Mental
transformations and visual comparison processes:
Effects of complexity and similarity. Journal of
Experimental Psychology— Human Perception and
Performance. 2, 503-514.
Cox, N. (1974). Prior help, ego development, and
helping behavior. Child Development. 45. 594-603.
Cummings, A., & Murray, H. G. (1989). Ego development
and its relation to teacher education. Teaching &
Teacher Education. 5, 21-32.
Dana, R. H. (1968). Six constructs to define Rorschach
M. Journal of Projective Technigues and Personality
Assessment. 32. 138-145.
Davidson, M. (1983). Uncommon sense. Los Angeles: J.
P. Tarcher.
Deitch, H. L., & Jones, J. A. (1983). The relationship
between stages of ego development and personal
constructs. Journal of Clinical Psychology. 39., 2 3 5-
239.
Dellas, M., & Gaier, E. L. (1970). Identification of
creativity: The individual. Psychological Bulletin.
73, 55-73.
Di Nuovo, S. (1984). Administration times for Witkin's
Group Embedded Figures Test. Perceptual and Motor
Skills. 58, 134.
Dobzhansky, T. (1964). Mankind evolving. New Haven:
Yale University Press.
Dykes, M., & McGhie, A. (1976). A comparative study of
attentional strategies of schizophrenic and highly
creative normal subjects. British Journal of
Psychiatry. 128. 50-56.
243
Epstein, S. (1985). The implication of cognitive-
experiential self-theory for research in social
psychology and personality. Journal for the Theory of
Human Social Behavior. 15. 283-310.
Erickson, V. L. (1977). Beyond Cinderella: Ego maturity
and attitudes toward the rights and roles of women.
Counseling Psychologist. 7, 83-88.
Exner, J. E. (1985). Rorschach workbook for the
comprehensive system (2nd ed.). New York: Wiley.
Exner, J. E. (1986). The Rorschach: A comprehensive
system (Vol. 1, 2nd ed.). New York: Wiley |
i
Eysenck, H. J. (1967). The biological basis of
behaviour. Springfield, England: C. C. Thomas.
Festinger, L. (1957). A theory of cognitive dissonance, j
Stanford, CA: Stanford University Press. j
Fingarette, H. (1963). The self in transformation. New
York: Basic Books.
I
Fuller, A. R. (1983). Synthesizing the everyday world.
The Journal of Mind and Behavior. 4, 369-388.
Gabel, S. (1988). A review of research supporting a ;
dualistic view of mental operations. Bulletin of the
Menninaer Clinic. 52. 410-422.
Gardner, H. (1985) . The mind's new science: A history
of the cognitive revolution. New York: Basic Books.
Garner, W. (1979) . Letter discrimination and
identification. In A. Pick (Ed.), Perception and its
development: A tribute to Eleanor Gibson (pp. 111-
144). New York: Erlbaum.
Garwood, D. S. (1964). Personality factors related to
creativity in young scientists. Journal of Abnormal
and Social Psychology. 68. 413-419.
Getzels, J. W., & Jackson, P. W. (1962). Creativity and
intel1igence. New York: Wiley.
Goldstein, K., & Scheerer, M. (1941). Abstract and
concrete behavior: An experimental study with special
tests. Psychological Monographs. 53.(2) .
244
Goodenough, D., & Karp, S. (1961). Field dependence and
intellectual functioning. Journal of Abnormal and
Social Psychology. 63. 243-246.
Haan, N. (1977). Cooing and defending. New York:
Academic Press.
Hadamard, J. (1945). The psychology of invention in the
mathematical field. Princeton, NJ: Princeton
University Press.
Hall, A. E. (1978). Review of GEFT. In Buros, O. K.
(Ed.), The eighth mental measurements yearbook (Vol.
1, pp. 572-573). Highland Park, NJ: Gryphon Press.
Harrow, M., Harkavy, k., Bromet, E., & Tucker, G. J.
(1973). A longitudinal study of schizophrenic
thinking. Archives of General Psychiatry. 28. 179-
182 .
Harrow, M., Himmelhoch, J., Tucker, G., Hersh, J., &
Quinlan, D. (1972). Overinclusive thinking in acute
schizophrenic patients. Journal of Abnormal
Psychology. 79, 161-168.
Harrow, M., Rattenbury, F., Marengo, J., & King, G.
(1990). A manual to assess positive thought disorder,
using the object sorting test. Unpublished
manuscript, Michael Reese Medical Center, University
of Chicago, and Northwestern University, Chicago.
Harrow, M., Tucker, G., & Adler, D. (1972). Concrete
and idiosyncratic thinking in acute schizophrenic
patients. Archives of General Psychiatry. 26. 433-
439.
Hartmann, H. (1958). Ego psychology and the problem of
adaptation. New York: International Universities
Press. (Original work published 1939)
Harvey, O. J., Hunt, D. E., & Schroder, H. M. (1961).
Conceptual systems and personality organization. New
York: Wiley.
Hasenfus, N., & Magaro, P. (1976). Creativity and
schizophrenia: An equality of empirical constructs.
British Journal of Psychiatry. 129. 346-349.
245
Hauser, S. T. (1976). Loevinger1s model and measure of
ego development: A critical review. Psychological
Bulletin. 83. 928-925.
Hawtrey, R. S. W. (1981). Commentary on Plato's
"Euthydemus.1 1 Philadelphia, PA: American
Philosophical Association.
Heidbreder, E. (1933). Seven psychologies. Englewood
Cliffs, NJ: Prentice-Hall.
Hertz, M. (1986). Rorschachbound: A 50-year memoir.
Journal of Personality Assessment. 50. 396-416.
Himmelhoch, J., Harrow, M., Tucker, G. J., & Hersh, J.
(1973). Manual for assessment of selected aspects of
thinking: Object Sorting Test (No. 02206). New York:
American Society for Information Sciences/National
Auxiliary Publications Service, Microfiche
Publications.
Hock, H. S. (1973). The effects of stimulus structure
and familiarity on same-different comparison.
Perception & Psychophysics. 14, 413-420.
Hofstadter, D. R. (1981). Prelude . . . ant fugue. In
D. R. Hofstadter & D. C. Dennett (Eds.), The mindfs
eve (pp. 149-2 01). Toronto: Bantam Books.
Holt, R. R. (1980). Loevinger1s measure of ego
development: Reliability and national norms for male
and female short forms. Journal of Personality and
Social Psychology. 39., 909-920.
Holzman, P. S. (1954). The relation of assimilation
tendencies in visual, auditory, and kinesthetic time-
error to cognitive attitudes of leveling and
sharpening. Journal of Personality. 22., 375-394.
Hoppe, C. F., & Loevinger, J. (1977). Ego development
and conformity: A construct validity study of the
Washington University Sentence Completion Test.
Journal of Personality Assessment. 41, 497-504.
Hoppe, K. D. (1986). Dialogue in the future. In L.
Robinson (Ed.), Psychiatry and religion: Overlapping
concerns (pp. 119-132). Washington: American
Psychiatric Press.
246
Hoppe, K. D. (1989). Psychoanalysis, hemispheric
specialization, and creativity. Journal of the
American Academy of Psychoanalysis. 17, 253-269.
Husserl, E. (1965). Phenomenology and the crisis of
philosophy (Q. Lauer, Trans.). New York: Harper &
Row Torchbooks.
Hy, L. X., & Bobbitt, K. H. (1990, March). The proper
use of the Washington University Sentence Completion
Test. Paper presented at the meeting of the Society
for Personality Assessment, San Diego, CA.
Inhelder, B. & Piaget, J. (1964). The early growth of
logic in the child. New York: Harper & Row.
Jahoda, M. (1958). Current concepts of positive mental
health. New York: Basic Books.
James, W. (1890). The principles of psychology (Vol.l).
New York: Dover.
Kagan, J., Moss, H. A., & Sigel, I. E. (1960).
Conceptual style and the use of affect labels.
Merrill-Palmer Quarterly. 6, 261-278.
Kagan, J., Moss, H. A., & Sigel, I. E. (1963).
Psychological significance of styles of
conceptualization. In J. C. Wright & J. Kagan (Eds.),
Basic cognitive processes in children (pp. 7 3-112).
Monographs of the Society for Research in Child
Development. 28. (Whole No.2, Serial No. 86).
Kant, I. (1961). Critique of pure reason. Selections in
Walter Kaufmann (Ed.), Philosophic classics: Bacon to
Kant. Englewood Cliffs, NJ: Prentice-Hall. (Original
work published 1781).
Katz, J. (1972). Semantic theory. New York: Harper
and Row.
Kegan, R. (1983). A neo-Piagetian approach to object
relations. In B. Lee & G. C. Noam (Eds.),
Developmental approaches to the self (pp. 267-308).
New York: Plenum Press.
Kepner, M. D., & Neimark, E. D. (1984). Test-retest
reliability and differential patterns of score change
on the Group Embedded Figures Test. Journal of
Personality and Social Psychology. 46, 1405-1413.
247
Kirkcaldy, B. (1987). Personality correlates of
Benton*s Visual Retention Test. Personality and
Individual Differences. 8., 141-143.
Koestler, A. (1964). The act of creation. New York:
Macmillan.
Koffka, K. (1935). Principles of gestalt psychology.
New York: Harcourt, Brace & Co.
Kogan, N. (1971). Educational implications of cognitive
styles. In G. S. Lesser (Ed.), Psychology and
educational practice (p. 244). Glenview, IL: Scott &
Foresman.
Kogan, N. (1973). Creativity and cognitive style: A
life-span perspective. In P. B. Baltes & K. W.Schaie
(Eds.), Life-span developmental psychology (pp. 145-
178). New York: Academic Press.
Kohlberg, L. (1964). Development of moral character and
moral ideology. In M. L. Hoffman and L. W. Hoffman
(Eds.), Review of child development research (Vol. 1,
pp. 3 8 3-431). New York: Russell Sage Foundation.
KOhler, W. (1969). The task of Gestalt psychology.
Princeton, NJ: Princeton University Press.
Kris, E. (1952). Psychoanalytic explorations in art.
New York: International University Press.
Kuhn, T. S. (1970). The structure of scientific
revolutions (2nd ed.). Chicago: University of
Chicago Press.
Kurtz, R. (1969). A conceptual investigation of
Witkin's notion of perceptual style. Mind, 78. (312).
Lakoff, G. (1977). On linguistic gestalts. Chicago
Linguistic Society Papers. 13. 236-287.
Lambert, H. (1972). Comparison of cognitive
developmental theories of ego and moral development
[Summary]. Proceedings of the 80th Annual Convention
of the American Psychological Association. 7, 115-116.
Landau, E., & Maoz, B. (1978). Creativity and self-
actualization in the aging personality. American
Journal of Psychotherapy. 32, 117-129.
248
Langer, S. (1942). Philosophy in a new kev. Cambridge,
MA: Harvard University Press.
Langer, S. (1953). Feeling and form. New York:
Scribner1s.
Langer, S. (1967). Mind: An essay on human feeling
(Vol. 1). Baltimore, MD: The Johns Hopkins Press.
Lapsley, D. K., Jackson, S., & Rice, K. (1988). Self
monitoring and the "new look" at the imaginary
audience and personal fable: An ego-developmental
analysis. Journal of Adolescent Research. 3_, 17-31.
Lashley, K. S. (1951). The problem of serial order in
behavior. In L. A. Jeffress (Ed.), Cerebral
mechanisms in behavior (pp. 112-136 ). New York:
Wiley.
LaVoie, A. L. (1984). Embedded figures test. In D. J.
Keyser & R. C. Sweetland (Eds.), Test critigues (pp.
259-265). Kansas City, MO: Test Corporation of
America.
Lawrence, D. M., & MacDonald, K. L. (1984). Ego
development and mature adult career change.
Unpublished manuscript, California State University,
Los Angeles, Department of Psychology.
Lehmann, H. (1966). Pharmacotherapy of schizophrenia.
In P. Hoch & J. Zubin (Eds.), Psychopathology of
schizophrenia pp. 388-411). London: Grune &
Stratton.
Lerner, J. W. (1976). Children with learning
disabilities (2nd ed.). Boston: Houghton Mifflin.
Loevinger, J. (1962). Measuring personality patterns of
women. Genetic Psychology Monographs. 65, 53-136.
Loevinger, J. (1966). The meaning and measurement of
ego development. American Psychologist. 2JL, 195-2 06.
Loevinger, J. (1969). Theories of ego development. In
L. Breger (Ed.), Clinical-cognitive psychology: Models
and integrations (pp. 83-135). Englewood Cliffs, NJ:
Prentice-Hall.
249
Loevinger, J. (1973). Ego development: Syllabus for a
course. In B. Rubenstein (Ed.), Psychoanalysis and
contemporary science (Vol. 2, pp. 77-98). New York:
Macmillan.
Loevinger, J. (with the assistance of Augusto Blasi).
(1976). Ego development. San Francisco: Jossey-Bass.
Loevinger, J. (1979a). Construct validity of the
Sentence Completion Test of ego development. Applied
Psychological Measurement. 3., 281-311.
Loevinger, J. (1979b). The idea of the ego. The
Counseling Psychologist. 8, 3-5, 39-40.
Loevinger, J. (1979c). Scientific wavs in the study of
ego development. Worcester: Clark University Press.
Loevinger, J. (1980). Some thoughts on ego development
and counseling. Personnel and Guidance Journal. 58,
389-390.
Loevinger, J. (1984). On the self and predicting
behavior. In R. A. Zucker, J. Aronoff, & A. I. Rabin
(Eds.), Personality and the prediction of behavior
(pp. 43-68). New York: Academic Press.
Loevinger, J. (1985). Revision of the Sentence
Completion Test for ego development. Journal of
Personality and Social Psychology. 48., 420-427.
Loevinger, J., & Knoll, E. (1983). Personality: Stages,
traits, and the self. Annual Review of Psychology.
34, 195-222.
Loevinger, J., & Hy, L. X. (1989). Measuring ego
development: Supplementary manual and exercises for
Form 81 of the Washington University Sentence
Completion Test (Vols. 1-2). (Pre-publication version
available from department of Psychology, Washington
University, St. Louis, MO 63130)
Loevinger, J., & Sweet, B. (1961). Construction of a
test of mothers* attitudes. In J. C. Glidewell (Ed.),
Parental attitudes and child behavior (pp. 110-123).
Springfield, IL: Thomas.
Loevinger, J., & Wessler, R. (1970). Measuring ego
development I: Construction and use of a sentence
completion test. San Francisco: Jossey- Bass.
250
Loevinger, J., Wessler, R., & Redmore, C. (197 0).
Measuring ego development: II: Scoring manual for women
and girls. San Francisco: Jossey-Bass.
Looney, J. (1988). Ego development and Black identity.
The Journal of Black Psychology. 15, 41-56.
Lucas, R. (1971). Validation of a test of ego
development bv means of a standardized interview.
Unpublished doctoral dissertation, Washington
University, St. Louis.
MacDonald, K. (1986). Individual differences in field
dependence-independence as an indicator of ego
developmental stages. Unpublished master's thesis,
California State University, Los Angeles.
Mahler, M., Pine, F., & Bergman, A. (1975). The
psychological birth of the human infant. New York:
Basic Books.
Markus, H., Smith, J., & Moreland, R. L. (1985). Role of
the self-concept in the perception of others. Journal
of Personality and Social Psychology. 49, 1494-1512.
Maslow, A. (1962). Toward a psychology of being. New
York: Van Nostrand.
McCloskey, M. E., & Glucksberg, S. (1978). Natural
categories: Well-defined or fuzzy sets? Memory and
Cognition. 6, 462-472.
McConaghy, N., & Clancy, M. (1968). Familial
relationships of allusive thinking in university
students and their parents. British Journal of
Educational Psychiatry. 114. 1079-1087.
McKenna, F. S. (1984). Measures of field dependence:
Cognitive style or cognitive ability? Journal of
Personality and Social Psychology. 47., 593-603.
Mead, G. H. (1934). Mind, self, and society. Chicago:
University of Chicago Press.
Mednick, S. (1962). The associative basis of the
creative process. Psychological Review. 69. 220-232.
Mednick, S. A., & Mednick, M. A. (1967). Examiner's
manual: Remote Associates Test. Boston: Houghton
Mifflin.
251
Mercer, C. D. (1979). Children and adolescents with
learning disabilities. Columbus, OH: Merrill.
Merleau-Ponty, M. (1963). The structure of behavior.
Boston: Beacon Press. (Original work published 1942)
Messick, S., & French, J. (1975). Dimensions of
cognitive closure. Multivariate Behavioral Research.
10. 3-16.
Mervis, C. B. (1980). Category structure and the
development of categorization. In R Spiro, B. C.
Bruce, & W. F. Brewer (Eds.), Theoretical issues in
reading comprehension. Hillsdale, NJ: Erlbaum.
Mervis, C. B., & Rosch, E. (1981). Categorization of
natural objects. Annual Review of Psychology. 32. 89-
115.
Messick, S. (1980). H. A. Witkin (obituary). American
Psychologist. 35. 99-100.
Miller, L. (1988). Ego autonomy, creativity, and
cognitive style: A neuropsychodynamic approach.
Psychiatric Clinics of North America. 11. 383-397.
Missler, R. A. (1986). Analytic and synthetic cognitive
functioning: A critical review of evidence bearing on
field dependence. Journal of Research in personality.
20. 1-33.
Moran, A. P. (1985). Unresolved issues in research on
field dependence-independence. Social Behavior &
Personality. 13. 119-125.
Noppe, L. D., & Gallagher, J. M. (1977). A cognitive
style approach to creative thought. Journal of
Personality Assessment. 41. 85-90.
Nunberg, H. (1948). The synthetic function of the ego.
In H. Nunberg (Ed.), Practice and theory of
psychoanalysis (pp. 120-136). New York: International
Universities Press. (Original work published 19 31)
Ornstein, R. E. (1977). The psychology of consciousness
(2nd ed.). New York: Harcourt Brace Jovanovich.
Payne, R. W., & Friedlander, D. (19 62). A short battery
of simple tests for measuring overinclusive thinking.
Journal of Mental Science. 108. 362-367.
252
Payne, R. W. , & Hewlett, J. H. G. (1960). Thought
disorder in psychotic patients. In H. J. Eysenck
(Ed.), Experiments in Personality (Vol. 2, pp. 3-104).
London: Routledge & Kegan Paul.
Pettigrew, T. F. (1958). The measurement and correlates
of category width as a cognitive variable. Journal of
Personality. 26. 532-544.
Piaget, J. (1952). The origins of intelligence in
children. New York: International Universities Press.
(Original work published 193 6).
Piaget, J. (1970). Structuralism. New York: Basic
Books.
Plato. (1965). Euthvdemus (R. K. Sprague, Trans, and
Intro.). Indianapolis, IN: Bobbs-Merrill. (Original
work produced ca. 3 67BC)
Quinlan, D. M., Schultz, D., Davies, R., & Harrow, M.
(1978). Overinclusion and transactional thinking on
the Object Sorting Test of schizophrenic and
nonschizophrenic patients. Journal of Personality
Assessment. 42., 401-407.
Rabinow, P., & Sullivan, W. M. (Eds.). Interpretive
social science: A reader. Berkeley: University of
California Press.
Rank, O. (1932). Art and artist (C. F. Atkinson,
Trans.). New York: Knopf.
Rapaport, D., Gill, G. M., & Schafer, R. (1968).
Diagnostic psychological testing. New York:
International Universities Press.
Raychaudhuri, M. (1966). Perceptual preference pattern
and creativity. Indian Journal of Creativity. 3., 67-
70.
Redmore, C., & Waldman, K. (1975). Reliability of a
sentence completion measure of ego development.
Journal of Personality Assessment. 39. 236-243.
Renna, M., & Zennhausern, R. (1976). The Group Embedded
Figures Test: Normative data. Perceptual and Motor
Skills. 43, 1176-1178.
253
Richardson, J. T. E. (1985). Integration versus
decomposition in the retention of complex ideas.
Memory & Cognition. 13., 112-127.
Rips, J. (1975). Inductive judgments about natural
categories. Journal of Verbal Learning and Verbal
Behavior. 14, 665-681.
Rogers, C. (1959). Toward a theory of creativity. In
H. H. Anderson (Ed.), Creativity and its cultivation.
New York: Harper.
Rorschach, H. (1942). Psychodiagnostics. Bern: Hans
Huber Verlag. (Original work published 1921).
Rosch, E., & Mervis, C. B. (1975). Family resemblances:
Studies in the internal structure of categories.
Cognitive psychology. ]_, 573-605.
Rothenberg, A. (1986). Artistic creation as stimulated
by superimposed versus combined-composite visual
images. Journal of Personality and Social Psychology.
50. 370-381.
Rotter, J. B. (1954). Social learning and clinical
psychology. Englewood Cliffs, NJ: Prentice-Hall.
Rychlak, J. F. (1981). Introduction to personality and
psychotherapy (2nd ed.). Boston: Houghton Mifflin.
Schachtel, E. G. (1959). Metamorphosis. New York:
Basic Books.
Scheerer, M. (1954). Cognitive theory. In G. Lindsey
(Ed.), Handbook of Social Psychology (Vol. 1, pp. 91-
142). Cambridge, MA: Addison-Wesley.
Scruggs, T. E., & Mastropieri, M. A. (1984).
Reliability of performance on a whole-part perception
task. Perceptual and Motor Skills. 58, 416-418.
Seeman, J. (1959). Toward a concept of personality
integration. American Psychologist. 14. 633-637.
Seeman, J. (1983). Personality integration: Studies and
reflections. New York: Human Sciences Press.
Seeman, J. (1989). Toward a model of positive mental
health. American Psychologist. 44. 1099-1109.
254
Shade, B. J. (1984). Field-dependency: Cognitive style
or perceptual skill? Perceptual & Motor Skills. 58.,
991-995.
Strickland. B. (1989). Internal-external control
expectancies: From contingency to creativity.
American Psychologist. 44. 1-12.
Sullivan, C., Grant, M. Q., & Grant, J. D. (1957). The
development of interpersonal maturity: Applications
to delinquency. Psychiatry. 20. 373-385.
Sullivan, E. V., McCullough, G., & Stager, M. (1970). A
developmental study of the relationship between
conceptual, ego, and moral development. Child
Development. 41. 399-411.
Sullivan, H. S. (1953). The interpersonal theory of
psychiatry. New York: Norton.
Teusch, R. (1988). Level of ego development and
bulimics* conceptualizations of their disorder.
International Journal of Eatincr Disorders. 7, 607-615.
Thomas, M., & Seeman, J. (1972). Personality
integration and cognitive processes. Journal of
Personality and Social Psychology. 24., 152-161.
Thompson, D. (1942). On growth and form (2nd ed.).
Cambridge: University Press. (Original work
published 1917)
Thurstone, L. (1944). A factorial study of perception.
Chicago: University of Chicago.
Torrance, E. P. (1962). Guidincr creative talent.
Englewood Cliffs, NJ: Prentice-Hall.
Urbaitis, M. (1988). The relationship of mentoring to
stages of interpersonal maturity. Unpublished
doctoral dissertation, The Fielding Institute, Santa
Barbara, CA.
Vernon, P. (1972). The distinctiveness of field
independence. Journal of Personality. 9, 3 67-3 91.
Viglione, D. J., & McMillan, H. L. (1990, March).
Research criteria for Level I and Level II Special
Scores. Paper presented at the meeting of the Society
for Personality Assessment, San Diego, CA.
255
Vygotsky, L. (1962). Thought and language. Cambridge,
MA: MIT Press.
Wachtel, P. L. (1968). Style and capacity in analytic
functioning. Journal of Personality. 36. 202-212.
Wallach, M. A. (1970). Creativity. In P. H. Mussen
(Ed.), Carmichael1s manual of child psychology (pp.
1211-1272). New York: Wiley.
Wallach, M. A., & Kogan, N. (1965). Modes of thinking
in voung children. New York: Holt, Rinehart &
Winston.
Waugh, M. H. (1981). Reliability of the Sentence
Completion Test of ego development in a clinical
population. Journal of Personality Assessment. 45.
485-487.
Werner, H. (1957). The concept of development. In D.
Harris (Ed.), The concept of development.
Minneapolis: University of Minnesota.
Werner, H. (1964). Comparative psychology of mental
development. New York: International Universities
Press. (Original work published 194 0)
Werner, H., & Kaplan, B. (1956). The developmental
approach to cognition: Its relevance to the
psychological interpretation of anthropological and
ethnolinguistic data. American Anthropology. 58., 866-
880.
Wertheimer, M. (1945). Productive thinking. New York:
Harper & Bros.
Wexler, D. A. (1974). Self-actualization and cognitive
processes. Journal of Consulting and Clinical
Psychology. 42. 47-53.
Wilson, R. (1954). A factor analytic study of creative
thinking abilities. Psvchometrica. 19, 297-311.
Witkin, H. A. (1950) . Individual differences in ease of
perception of embedded figures. Journal of
Personality. 19, 1-15.
Witkin, H. A. (1965). Psychological differentiation and
forms of pathology. Journal of Abnormal Psychology,
70. 317-336.
256
Witkin, H. A., Dyk, R. B., Faterson, H. F., Goodenough,
D. R., & Karp, S. A. (1974). Psychological
differentiation. Potomac, MD: Erlbaum. (Original
work published 1962)
Witkin, H. A., & Goodenough, D. R. (1981). Cognitive
styles: Essence and origins. New York: International
Universities Press.
Witkin, H. A., Goodenough, D. R., & Oltman, P. K.
(1979). Psychological differentiation: Current '
status. Journal of Personality and Social Psychology. ,
37, 1127-1145. I
Witkin, H. A., Lewis, H. B., Hertzman, M., Machover, K.,
Meissner, P.B., & Wapner, S. (1972). Personality
through perception. Westport, CT: Greenwood Press.
(Original work published 1954)
Witkin, H. A., Moore, C. A., Goodenough, D. R., & Cox, P.
W. (1977). Field-dependent and field-independent
cognitive styles and their educational implications. j
Review of Educational Research. 47. 1-64. j
Witkin, H. A., Moore, C. A., Oltman, P. K., Goodenough, j
D. R., Friedman, F., Owen, D., & Raskin, E. (1977).
Role of the field-dependent and field-independent j
cognitive styles in academic evolution: A longitudinal j
study. Journal of Educational Psychology. 69., 197- i
211.
Witkin, H. A., Oltman, P. K., Raskin, E., & Karp, S. A.
(1971). A manual for the embedded figures tests.
Palo Alto, CA: Consulting Psychologists Press.
Woody, E., & Claridge, G. (1977). Psychoticism and ,
thinking. British Journal of Social and Clinical I
Psychology. 16. 241-248. I
257
APPENDIX A
Object Sorting Test Items
Item Material Predominant
Color
1.
Fork,large Metal Silver
2 . Fork, salad Metal Silver
3. Knife Metal Silver
4 . Spoon, demitasse Metal Silver
5. Toy hammer Plastic Red, black
6. Wood block with
nail half-way in
Wood/metal Brown/silver
7. Nail Metal Silver
8 . Nail Metal Silver
9. Pliers Metal Silver
10. Screwdriver Wood/metal Brown/silver
11. Toy musical instru
ment, tone block
Wood Brown
12 . Mallet, tone block Wood Brown
13 . Dog figurine, small Porcelain Brown/white
14 . Ball Rubber Red
15. Bicycle bell Metal Silver/red
16. Board game chip Plastic Red
17. Board game chip Plastic Yellow
18. Padlock with key Metal Silver
19. Candle, short taper Wax Red
20. Candle, short wide Wax White
21. Candle, apple-shaped Wax Red
22 . Sugar cube Food White
23 . Sugar cube Food White
24. Cracker Food Beige
25. Cracker Food Beige
26. Candy dish (or
ash tray)
Glass Red
27. Chocolate cigar Food Brown
28. Cigar in cellophane
with pull strip
Tobacco Brown/ red
29 . Pipe Corn cob Beige/black
30. Matchbox, filled,
partly open
Cardboard/
wood
White/red
31. Match, kitchen Wood Beige/red
32 . Match, kitchen Wood Beige/red
APPENDIX B
Ogive Scoring Rules
for the 18-Item Sentence Completion Test
Protocol Is: If There Are:
Integrated At least 1 Integrated and
1 Autonomous rating
Autonomous At least 2 Autonomous ratings
(or higher)
Individualistic At least 3 Individualistic ratings
(or higher)
Conscientious At least 6 Conscientious ratings
(or higher)
Self-aware At least 9 Self-aware ratings
(or higher)
Impulsive At least 3 Impulsive ratings
(or lower)
Self-protective At least 3 Self-protective ratings
(or lower)
Otherwise, protocol
Conformist
is
(Rules are applied in order given, from the top down.)
Note, Adapted from "The proper use of the Washington
University Sentence Completion Test" by L. X. Hy and K.
H. Bobbitt, 1990, Paper presented at the meeting of the
Society for Personality Assessment, San Diego, CA.
259
APPENDIX C
Summary of Key Abbreviations and Codes
Abbreviation Brief Introduction, elab-
or Code meaning oration of meaning
Ego Development
ED
EL
SCT
Ego Development
Ego Level
Sentence Completion Test
Chapter 1
Chapter 1
Chapter 1
—
Cognitive Style
FDI
FD
FI
GEFT
Field Dependence-Independence
Field Dependence
Field Independence
Group Embedded Figures Test
Chapter 1
Chapter 1
Chapter 1
Chapter 1
ICS Integrative Cognitive Style Chapter 2
Object Sorting Test
OST
Beh
O-in
Rich
Object Sorting Test
Behavioral Overinclusion Scale
Conceptual Overinclusion Scale
Richness of Associations Scale
Chapters 2-3
Chapters 2-3
Chapters 2-3
Chapters 2-3
(table continues)
260 j
Abbreviation Brief
or Code meaning
Rorschach Test
Preliminary
calculations
R Total protocol responses Table 5
W Whole response Chapter 2
D Detail response Chapter 2
D0+ Synthesized response with Chapter 2
form demand
DQV/+ Synthesized response with Chapter 2
no form demand
FQo Ordinary use of form features Chapter 2
ZSum Sum of Z scores (Organizational Chapter 2
activity)
Zest Best weighted prediction of ZSum Chapter 2
INCOM Single merged percept Chapter 2
FABCOM Loosely associated percepts Chapter 2
Research
Hypotheses
Cl Composite Index: all variables Chapter 4
scored positive for ICS
W: D Economy Index Chapter 2
W+ Integrated wholistic percepts Chapter 2
DOsyn Meaningful synthesis Chapter 2
M Human movement Chapter 2
FOo% Use of commonplace features Chapter 2
Zf Organizational activity Chapter 2
Zd Overincorporation Chapter 2
COM Sum of INCOM and FABCOM responses Chapter 2
Post Hoc
Analyses
EB Introversive or extratensive Chapter 4
response style
EA Extent of available resources Chapter 4
SumC Sum of chromatic color responses Chapter 4
X+% % of conventional form responses Chapter 4
X-% % of distorted form responses Chapter 4
FQu Unusual use of form Chapter 2
FQ+ Elaborated articulation of form Chapter 2
Introduction, elab
oration of meaning

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The relationship of cognitive, affective, demographic, and school-related variables to membership of community college students in four ethnic groups and to their academic achievement

PDF

The effects of a manual-guided cognitive intervention program upon substance abusers

PDF

Physical measures of female speech and voice variables as indices of retarded clinical depression

PDF

Adult attachment styles and relationship satisfaction

PDF

Family-of-origin intervention as a therapeutic resource in enhancing marital adjustment: An intergenerational approach

PDF

Physiological, emotional, and self-reported indices of stress response for women in dual-career marriages

PDF

Counseling and counselor training of the American rabbinate

PDF

The good child as the unidentified patient in the family system: A study of childhood antecedents of adult alienation from self

PDF

Intravenous drug use among women and men addicts and psychosocial and health belief correlates of AIDS risk reduction

PDF

Self-efficacy, job/home attitudes and career concept relationships in pairs of mothers and undergraduate daughters

PDF

Attitudes toward achieving peace : a measure of "peace through strength" and "peace through cooperation" and the relationship of these attitudes with affective, coping, personality, and gender co...

PDF

Christian counselors: A descriptive study of their counseling practices

PDF

Developmental study of the effects of media on cognitive style

PDF

Locus of control and motivation for counseling among cancer patients

PDF

A proposed synthesis for Festinger's theory of cognitive dissonance with Kohlberg's model of moral development: An exploratory study

PDF

Differences in life satisfaction, future orientation and locus of control between educationally active and non-active older adults

PDF

Religiosity, social support, meaning of life, and mental health in Indonesian Christians

PDF

A comparison of the validities of an item-sampling procedure and a traditional item-allocation method in the construction of achievement examinations in intermediate algebra for community college...

PDF

By the self defined : creating a lesbian identity

PDF

Successful and unsuccessful vocational rehabilitation of the legally blind: A multi-statistical approach

Asset Metadata

Creator MacDonald, Karen Lee (author)

Core Title The construct of integrative cognitive style and its measure as an indicator of advanced ego development

Contributor Digitized by ProQuest (provenance)

Degree Doctor of Philosophy

Degree Program Education-Counseling Psychology

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

Tag OAI-PMH Harvest,psychology, cognitive

Language English

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c26-538767

Unique identifier UC11245180

Identifier usctheses-c26-538767 (legacy record id)

Legacy Identifier DP25349.pdf

Dmrecord 538767

Document Type Dissertation

Rights MacDonald, Karen Lee

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