-, .zAYfi-‘i.
4 V

ACCEPTANCE vERsus REJEcTioN . * '
0F OTHERS AND sELE iN '
PERSONALITY SELF- REPORTS

Dissertation for the Degree of Ph. D, G

MICHIGAN STATE UNIVERSITY
LARRY M. GERSTENHABER
1974

 

llilli Ii Mil

‘ 3 1293 0

M“ ”II V W
83 5612 LIBRARY

Michigansw
Um'vcnity

 
 
   

This is to certify that the

thesis entitled

Hccepfwce Versus ReJecf/on 0/ Offlers Md ;
self I." Persona/ify Se/f-IQQPOr-ts

‘_ presented by

Larry M. Gerstenhafier

has been accepted towards fulfillment
of the requirements for

__Ehl_degree in M1119 y

[/7
1’ .

Major professor

 

 

  

z,

 
  

‘ 5 sons

'_ BLUUKBIN MDERY INC.

~~ may ambzn
[MAI

w

ACCEF

The research 11

211:; night be adequ

 

izexsions of accept
t'ersts rejection of

fzzrlation, the pre
gazil measures of p

:Erszzality Inventorj

:‘Efrlist (ICE), Ret‘

Static dlfferentiai

:E‘Q’vn '
on 5 Personal (

n. int undergraduate

mm cluster analvs 1'

12.5 factor analys IS

37‘1"
\uigS communality a
E: 3" .T

Munt for the do

The findings gen

.1

“4‘?
LETS concerned wi‘

ABSTRACT

ACCEPTANCE VERSUS REJECTION OF OTHERS AND
SELF IN PERSONALITY SELF-REPORTS

By

Larry M. Gerstenhaber

The research literature which I reviewed suggested that person-
ality might be adequately described by two minimally interdependent
dimensions of acceptance versus rejection of self (SAR) and acceptance
versus rejection of others (ARO). Designed to investigate this
formulation, the present study was limited to self-report, paper-and-
pencil measures of personality. A battery consisting of Eysenck's
Personality Inventory (EPI), LaForge and Suczek's Interpersonal
Checklist (ICL), Rotter's Interpersonal Trust Scale, a series of
semantic differential scales, and selected scales from the MMPI and
Shostrom's Personal Orientation Inventory (POI) was administered to
182 MSU undergraduates enrolled in elementary psychology courses. The
BCTRY cluster analysis technique and a non-communality principal
axis factor analysis were applied to these data. The BCTRY method
employs communality as a measure of the sufficiency of dimensions
to account for the domain in question.

The findings generally sustained expectations of two large

clusters concerned with the acceptance versus rejection of others

' ' "'JI'VL'I'

1‘!

 

35.3) and self

:agsr clusters r
E35 an ARO ClUSi
aoi acceptance c
hostility, or it
described by Eve
"ezctional" mode
33's social mod
assertiveness, w
related to £8811:
:31: $53 clusters
zeestres located
:L:s:er,}1,u_’ ap;
'I'iS aCCOunted for
217-3 of SAR'S r8j

c."“‘"
"“1 SEnerally t

Larry M. Gerstenhaber
(ARO) and self (SAR), as well as their relative independence. Four
major clusters and two or three minor ones were identified. Salient
was an ARO cluster characterized by behaviors of warmth, gentleness,
and acceptance of others, and by adjectives related to the affection-
hostility, or love—hate, Interpersonal Checklist dimension. SAR was
described by two major clusters, seemingly representing "social" and
"emotional" modes of self-acceptance (SARsoc’ SARemot’ respectively).
SAR's social mode was defined by measures related to extraversion and
assertiveness, while the emotional mode was defined by measures
related to feelings about oneself. The relationship between these
two SAR clusters was depicted as a gradient with two general SAR
measures located midway between SARSOC and SARem . The fourth major

0t

cluster, MAL, apparently dealing with more severe maladjustment than
was accounted for by SAR and ARO, seemed at least partly representa-
tive of SAR's rejection pole. MMPI scales employed in the present
study generally tapped MAL, but an Introversion scale loaded most
highly on SARsoc’ and a Suspicion scale loaded inversely nearly as
highly on ARO as on MAL. I interpreted these results as suggesting
that popular measures of maladjustment are substantially loaded on
SAR. In addition to interpreting these clusters, I cautioned that
selection of the measures may have partially influenced the outcome
of the study.

Measures of extraversion-introversion and anxiety (Eysenck's
EPI), as well as a social desirability response set scale, tended

to link firmly with SAR and MAL, but minimally with ARO. This find-

ing suggested that popular theories claiming the sufficiency of

 

:Tizensions of 9"”
Personality have t
5;;ears that the ‘
ggasion. Males
ercept that 3 min‘
faales but vi th 1
that a set of 58!:
5A3 and ARO adeqt.
The main all
ccllege students
suggested that fu
:easures in stud:
:larify SAR and A
serious error for
E2131 concern for
regard, Rogers' ‘
a:CEPtance of sel

Present findings

"Grantee:

.Cfm ' ;
lc‘ I

U31 H‘ Wakely
y‘ o

{ME-El
"‘4 Es F .

Larry M. Gerstenhaber
dimensions of extraversion-introversion and anxiety to represent
personality have overlooked the importance of ARO. In addition, it
appears that the tendency to "look good" might be part of the SAR
dimension. Males and females were nearly identical on all dimensions,
except that a minor Transparency factor was linked with ARO among
females but with SAR among males. I also inferred from the results
that a set of semantic differential scales could be used to assess
SAR and ARO adequately.

The main allegations of the present study were supported for
college students on self—report, paper-and-pencil measures. I
suggested that future researchers expand the number and variety of
measures in studies like this one, in order to further validate and
clarify SAR and ARO. Additionally, I concluded that it may be a
serious error for therapists to focus on enhancing self—esteem without
equal concern for the client's orientation toward others. In this
regard, Rogers' (1953) belief in an intrinsic linkage between the
acceptance of self and others seems sharply inconsistent with the

present findings.

.\
." 1

Committee: 4~ 3,- '4 // /’

“if / if. 01’ CL A

John R. Hurley, chairperson
I

'\

I
0

John H. Wakely
Martha Karson

Charles F. Wrigley

 

u‘
.harA-r

in Pa

ACCEPTANCE VERSUS REJECTION OF OTHERS AND
SELF IN PERSONALITY SELF—REPORTS
By

a (

Larry M? Cerstenhaber

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Psychology

1974

i
'.
l
I
'd
V
|

 

 

‘
I

all?"

'9
C25‘

S

C

o
5

17'

Iwish t0 t
:1 teacher, SUPe
and support help
dissertation can
the years of my
35 the data unde
tsstion such in
:13 reliability.
Iver the past se

its inception ,

fitting of this

..
Ll‘:

ether, th

Q

I also wish
‘5sz at St .
‘ c

minue‘i Supper

ling the Past

ACKNOWLEDGEMENTS

I wish to thank Dr. John Hurley, who over the years has been
my teacher, supervisor, therapist and friend. His participation
and support helped make this project as enjoyable as writing a
dissertation can be. Also, Dr. Jack Wakeley, who followed me through
the years of my graduate training and who made statistical analysis
of the data understandable to the point that I finally began to
question such institutionalized psychological concepts as validity
and reliability. Dr. Martha Karson was a source of personal esteem
over the past several years, and fully supported this project from
its inception. Also important was the occasional aid, during the
writing of this dissertation, of my friend Dr. George "Bill"
Fairweather, who introduced me to BCTRY while I served as his research
assistant several years ago. A

I also wish to thank the outpatient and child and adoleSCent
staffs at St. Lawrence Community Mental Health Center for their
continued support. Finally, my heartfelt thanks to Deborah and
Lisa Norton, my housemates, who shared many agonized hours with me

during the past year.

ii

REVIEW

METHOD.

TABLE OF CONTENTS

OF THE LITERATURE . O O O O O O O O O O O C 0

Dimensions of Personality. . . . . . . . . .
Two Salient Dimensions. . . . . . . .

The Self-Other Dichotomy ~ Interpersonal

Acts Defined. . . . . . . . . . . .
Postulations from Theory. . . . . . .
Postulations from Factor Analysis .

Postulations from Encounter Group Research. . .

Evidence for a Circular Order of Personality
Relationship of the Dimensions . . . . . . .
Response Set . . . . . . . . . . . . . . . .
Cross-Cultural Stability . . . . . . . . . .
Summary - Statement of Partial Hypotheses.

Subjects . . . . . . . . . . . . . . . . . .
Procedure. . . . . . . . . . . . . . . . . .
Test Battery . . . . . . . . . . . . . . .
Self-Report Data vs. Behavior Ratings
Instruments . . . . . . . . . . . . .
Time of Battery. . . . . . . . . . . . . . .
Order of Battery . . . . . . . . . . . . . .
Statistical Method . . . . . . . . . . . . .
The BCTRY Cluster Ana ysis System . .
The Communality Problem . . . . . . .
Factor Analysis . . . . . . . . . . .
Statistical Treatment . . . . . . . .
Statement of Hypotheses . . . . . . .

RESIJIITS O O O O O O O O O O O O O O O O O O O O I 0

Pilot Analysis . . . . . . . . . . . . . . .
Empirical Cluster Analysis (Total Sample). .
Clusters. . . . . . . . . . . . . . .
Cluster 1 (ARO) . . . . . . . . . . .
Cluster 2 (MAL) . . . . . . . . . . .

Cluster 3 (SAR ). . . . . . . . . .
soc

iii

Variables

Page

22

22
22
23
24
24
35
36
36
36
41
42
43
44

45

45
46
46
47
53
54

CIUSter 4 (SARemOt) o o o o o o o

Cluster 5 . . . . . . . . . . . . .

Cluster 6 (Transparency). . . . . . .
Relationship Between Clusters. . . . . . . . .

Sufficiency of Clusters to Account for the Domain.

Principal Axis Analysis. . . . . . . . . . .
Preset Cluster Analysis - BCTRY. . . . . .
Introduction to Preset. . . . .
Preset Clusters . . . . . . . . . . .
Relationship Between Clusters - Preset .
Preset Summary Note . . . . . . . . . .
Non-Clustering Variables . . . . . .
Response Set Measures. . . . . . . . . .

Comparison Between Subgroups - Male vs. Female .

Comparison Between Subgroups - Non—
Specific Groups . . . . . . . . .

DISCUSSION. 0 '0 O O O O O O O O O O O O O O O O O O O

The Salient Dimensions . . . . . . . . . . . .

Cluster 1: Acceptance versus Rejection

of Others (ARO) . . . . . . . . . . .

Clusters Representing Acceptance versus

Rejection of Self (SAR) . . . . . . .

The Maladjustment Cluster (MAL) . . . .

A Minor Cluster: Transparency (TR) . .

Other Groupings . . . . . . . . . . . .

Relationship of the Dimensions . . . . . . . .

Validity of the Item Sample. . . . . . . . . .

Eysenck's "Super Factors". . . . . . . . . . .

Response Set Measures. . . . . . . . . . . . .
Subsample Comparisons. . . . . . . . .

Measurement of ARO and SAR (Clustering of Variables)

Implications for Research and Psychotherapy. .
Research. . . . . . . . . . . . . . . .
Psychotherapy . . . . . . . . . . .

Summary and Conclusions. . . . . . . . . . . .

APPmDICES O O O O O O O O O O O O O O O O O O O O O O

WERENCE S O O O O O O O O O O O O O O O O O O O O 0

iv

Page

57
59
60
61
63
66
67
67
69
79
80
81
81
83

87
93
93
93

95

97
101
101
102
103
105
106
107
108
114
114
118
119

122

148

Table

10

11

12

13

14

15

LIST OF TABLES
Page

Intercorrelations between cluster scores I, B, S,
T (Tryon and Bailey, 1970, p. 164) . . . . . . . . . . 10

Alpha reliabilities of full and shortened TSC
scales (Tryon, 1966) O O O O O I I O O O O O O O O O O 26

Description of POI categories. Categories on the
POI are presented as a series of opposites . . . . . . 30

Variables for cluster analysis . . . . . . . . . . . . 37

Clusters on empirical V-analysis - total sample
cluster structure of expanded clusters . . . . . . . . 49

Oblique factor coefficients - empirical V-analysis . . 51
Correlations between oblique cluster domains . . . . . 61

Number of residuals in .1000-.1999 range (empirical
anaIYS1S) O O O O O O C O O O O C O C O O O O O O O O O 65

Clusters on preset V-analysis - total sample cluster
structure of expanded clusters . . . . . . . . . . . . 70

Oblique factor coefficients - preset analysis. . . . . 72

Correlations between purported measures of response
set. 0 O O I O O O O O O O O O O O O O O O I O I O O O 82

COMP analysis for males and females. . . . . . . . . . 83

V¥analysis performed on preset clusters as part of
COMP analysis - males vs. females. . . . . . . . . . . 85

Empirical clustering of variables - male vs. female. . 88

COMP analysis for non-specific groups - cosines
among dimensions . . . . . . . . . . . . . . . . . . . 89

 

.a'ole

15 V-analysis 1
COPSD analys

17 Oblique fat
tionship of

18 Expected an

Table

16

17

18

Page
V-analysis performed on preset clusters as part of
COMP analysis - random samples, N-llO, N = 72. . . . 90
Oblique factor coefficients relevant to the rela-
tionship of SAR and MAL (empirical analysis) . 99

Expected and actual clustering of variables. . . . . . 110

vi

Figure

LIST OF FIGURES

Relationship of the four temperaments to Berne's
life positions . . . . . . . . . . . . . . . . .

Summary of dimensional analyses. . . . . . . . .
Lorr and Mc'Nair primary factors . . . .

Diagrammatic representation of the SAR gradient.

vii

Page
3

. 13
. 15
. 98

 

 

 

FYJ

(s

LIST OF SPAN DIAGRAMS
Diagram Page
1 Empirical analysis . . . . . . . . . . . . . . . . . . 52
2 Empirical analysis . . . . . . . . . . . . . . . . . . 55
3 Empirical analysis . . . . . . . . . . . . . . . . . . 64
4 Preset analysis. . . . . . . . . . . . . . . . . . . . 73
5 Preset analysis. . . . . . . . . . . . . . . . . . . . 74

6 COMP analysis for males and females (higher order
ClusterS) . O C O O C O C O . O O C O O O O O O O O O O 86

7 COMP analysis for non-specific groups (higher order
ClusterS) O O O O O O O O O O O O O I O O O O O O O O O 92

viii

Appendix

A

LIST OF APPENDICES

AN ARBITRARY CLASSIFICATION OF MMPI ITEMS
BY CONTWT O O O C O O O O C O O O O O O O

MMPI ITEMS . . . . . . . . . . . . . . . .
EPI ITEMS. . . . . . . . . . . . . . . . .
ICL ITEMS. . . . . . . . . . . . . . . . .
POI ITEMS. . . . . . . . . . . . . . . . .

ROTTER INTERPERSONAL TRUST SCALE . . . . .

TWO-DIMENSIONAL INTERPERSONAL INVENTORY. .

SEMANTIC DIFFERENTIAL SCALES . . . . . . .

RAW CORRELATION MATRIX OF ALL VARIABLES. .

CORRELATIONS BETWEEN DEFINING VARIABLES OF
IN THE EMPIRICAL ANALYSIS. . . . . . . . .

PRINC IPAL AXIS MATRIX O O O O O O O O O O O

VARIMAX ROTATION ANALYSIS - ROTATED FACTOR

CORRELATIONS BETWEEN DEFINING VARIABLES OF
IN THE PRESET ANALYSIS . . . . . . . . . .

ix

CLUSTERS

LOADINGS.

CLUSTERS

Page

122
123
128
130
132
136
138
139

140

142
143

145

147

 

q
uuvfi
_.
»

an! IE}

U l‘!
H a . c I
a. «.1. MI firiJ
u‘» .i. ~ .Q ‘ ‘

N‘

9‘

v;

 

REVIEW OF THE LITERATURE

Dimensions of Personality

 

For ages man has attempted to classify himself and his environ-
ment. The classical view that all nature is composed of four basic
elements, air, earth, fire and water, was expanded, first by
Hippocrates (c. 400 3.0.) and later by Galen (c. 150 A.D.), to
describe men in relation to others. Four temperaments were described,
roughly corresponding to the four elements. Thus the sanguine
temperament corresponded to air, the melancholic to earth, the
choleric to fire, and the phlegmatic to water. Allport (1937)
described feelings that correspond to the temperaments. The
sanguine person is hopeful, the melancholic sad, the choleric prone

to anger, and the phlegmatic apathetic and sluggish.

 

Two Salient Dimensions

This ancient system and its behavioral correlates become even
more interesting when viewed with the recently accumulating evidence
for two prepotent dimensions in personality theory. This evidence,
reviewed by Adams (1964), strongly suggests that these dimensions are
best defined by "the degree of acceptance or rejection of self" and

"the degree of acceptance or rejection of others." This chapter is

~e
1'}:

CCU;

 

» A
a;

2
concerned with reviewing evidence in the literature relevant to the

appearance of these dimensions.

The Self-Other Dichotomy - Interpersonal Acts Defined

 

If one employs Adams' (1964) definers of the predominant
dimensions of personality, it becomes apparent that any description
of personality, and therefore interpersonal behavior, must take into
account both Self-Acceptance-Rejection (SAR) and Other—Acceptance-
Rejection (ARO). Foa (1961) described an interpersonal act as

"an attempt to establish the emotional relationship of

the actor toward himself and toward the other, as well

as to establish the social relationship of the self

and the other with respect to a larger reference group."

The division in viewing behavior as meaningful to both self and other

is referred to here as the self~other dichotomy.

Postulations from Theory

 

The late Eric Berne developed a system, one of the main tenets
of which is based on the self—other dichotomy. Berne (1966) believed
that an individual's fundamental life position could be defined by
two polar dimensions: I'M 0K - I'M NOT OK, YOU'RE OK - YOU'RE NOT OK.
According to Berne, the individual who has made the life decision
I'M OK.- YOU'RE OK is the individual most likely to be described as
healthy. A further discussion of the evolvement of the four possible
life positions is presented in Thomas Harris', I’M’OK - YOU’RE OK:

A Guide to Transactional Analysis.

It is easy to see how Berne's dimensions correspond to the

earlier systems described above. A description of this relationship,

including current psychological terminology, is presented in Figure l.

.mdoaufimoa owHH m.ocuom on mucoamuomamu snow onu mo magmGOfiumHom .H ouswam

 

 

 

 

 

 

Mo uo: ou.=ow I Mo uoc a.H mwouwnom haumm< owumawoanm

Mo uo: ou.=ow I mo a.H vaoqmumm huwaanwummuH ownoaono

MO mu.now I Mo was E.H o>awmmuamn mmmacmm UHHonoamHmz
wdfinwamauum

Mo mu.:ow I Mo a.H Imamm .hnuammm mmmaaamomom onwawnmm

Amcuomv >woaoafiahmu oumaouuoo Aaoaou

coauwmoa omwg Hmofiwoaonohmm uamuuao uow>mnom .mmumuoommamv

 

 

 

unmauuomaua

 

 

 

Li 1"" Iran—r-

a-n Ya

.. ,
but .

I
.. v.1
V..S

.‘: r!
a. '<
I, . ,
.it a

r.

‘10
[J

4

The work of Freedman, Leary, Ossorio and Coffey (1951) also
generated a two-dimensional model. Seeking a comprehensive schema
for personality description, Freedman et al. developed a system of
variables based on four criteria: systematic interrelatedness of
variables, interpersonal reference, encompassing of normal as well
as abnormal functioning, and ability to transform into an operational
statement. They believed that personality exists at three levels -
public, conscious, and private (unconscious). The public level can
be defined as interpersonal mechanism and is best described by verbs.
The conscious and private levels represent interpersonal traits and
are described by adjectives. Freedman et a1. separated their list
of variables into self and other categories and considered them to
be arranged in a circular order of sixteen divisions. Seeking to
measure the conscious level, Suczek and LaForge (1955) developed the
Interpersonal Checklist (ICL). This measure is scored on the six-
teen possible modes of behavior. The checklist itself can be
reduced further to two major dimensions: Affection-Hostility and
Dominance-Submission. LaForge (1973) reported factor analytic results
in accordance with this reduction to two dimensions. A lengthy
summary of the earlier work is found in Leary (1957).

These two dimensions ally closely with Adams' (1964) suggestion
that Self-Acceptance-Rejection and Other-Acceptance—Rejection play
a major role. According to Adams,

"The dominance-submission axis defines the degree of

acceptance or rejection of self, while the affection-

hostility axis defines the degree of acceptance or
rejection of other."

 

....

1 l
.ub
.ua

-\
al.-

-.__

 

 

:1

n-‘

.,_

l,—

R!
r
.

’5

 

5
Hurley (in press, a), in agreement with this position, observed that
positive self—references on the ICL ("self—respecting", "self—
confident", "self-reliant", "assertive") align "exclusively" with
the Dominance pole, while negative self-referent terms ("always
ashamed of self", "self—punishing", "lacks self-confidence") align
with the Submissive pole.

Foa (1961) also has redefined Leary's dimensions to fit the
self-other dichotomy. Foa began by dividing the dimensions into
four discrete facets: Dominance, Submission, Love and Hostility.
Believing that actions are meaningful both toward other and self,
he defined eight profiles: hostility to self, submission to self,
dominance of other, submission to other, etc. He also included
another variable which he described as mode of action. This can be
social or emotional. For example, he suggested that Leary's
managerial—autocratic behavior is a blend of social and emotional
rejection of other and of social and emotional acceptance of self.
Restated in Berne's system, this is the I'M OK - YOU'RE NOT OK
position. This can be done for each of Leary's types. In addition,
Foa too has placed these interpersonal variables in a circular order.

The concept of circular order is discussed in a later section.

Postulations from Factor Analysis

 

Behavior Studies:
In addition to evidence compiled from a theoretical base, the
development of factor analytic methods has made the reduction of large

numbers of variables to a small number of dimensions a relatively

 

 

 

18510!

I. .
"yup!-

ILQALAA

carter

'3-3.’
.rtul

IS 20!

| 'h‘.

tu’h}.

’ic-b.“
'\
“~U“.

.r)
.;
H5

(I!

In

~|
I.

I t
an
r)

A":
'5‘,“

u

6
simple process. Foa (1961) reported on a variety of studies. Carter
[(1954) (Foa, 1961)] in a variety of situations emphasized three
factors: Individual prominence and achievement, Aiding attainment
by the group, and Sociability. Borgatta, Cottrell and Mann (1958)
reported that two of Carter's three factors obtained in a study in
which graduate students were rated by peers on variables based on
Carter's three factors. The investigators labelled the factors
Individual Assertiveness and Sociability. Individual assertiveness
is comparable to the SAR dimension, while Sociability is closer to
ARO. Schaefer (1959, 1961), in studies of maternal and child
behavior, found evidence for two factors which he labelled love vs.
hostility and autonomy vs. control. The closeness of these dimensions

to Leary's, and therefore to SAR and ARO, is apparent from the names.

Questionnaire Studies:

Perhaps the best known researchers in the area of personality
questionnaire research are H. J. Eysenck and Raymond Cattell.

Eysenck (1953), factoring a large number of traits measured by a
variety of instruments, found evidence for three main dimensions:
Neuroticism, Extraversion-Introversion, and Psychoticism. Neuroti-
cism and Extraversion became the basis for development of the
Maudsley and Eysenck personality inventories.

Cattell (1964 and earlier) described fifteen primary personality
factors which have been embodied in the l6-PF questionnaire. General
intelligence was added as the sixteenth factor. Cattell's own
factoring of these 16 first order factors suggested four higher order

factors: (1) Dynamic integration vs. anxiety, (2) Extraversion—

 

istrovers

Crbaroicen

analyses

only two

these Adj

.AI“"T

7
Introversion, (3) Cyclothyme vs. Schizothyme constitution, and (4)
Unbroken success vs. frustration, of which factors 2 and 4 are
highly significantly positively correlated. Peterson (1965) compared
analyses of Cattell data for adults and children and found that
only two higher order dimensions could be defended. He labelled
these Adjustment and Extraversion—Introversion. Wetzel (Peterson,
1965) administered the l6-PF Questionnaire to undergraduate psychology
students and obtained peer and parent ratings on traits purportedly
measured by the questionnaire. He found two second order factors,
which be labelled Dynamic integration (self-perception of adjustment)
and Introversion-Extraversion.

Although Cattell held that primary factors are most important,
Eysenck argued that it is the larger dimensions that are more meaning-
ful psychologically (Eysenck, 1972). Using Cattell's correlations
for the l6—PF instrument, Eysenck corrected for attenuation and
concluded:

"As far as we can see, practically all the information

contained unreliably in the primaries is contained

reliably in the second-order factors."

Wiggins (1968) struck a responsive chord to this argument. In a vast
review of the personality-structure research he reported:

"If consensus exists within the realm of temperament

structure, it does so with respect to the importance

of the large, ubiquitous, and almost unavoidable

dimensions of extroversion and anxiety (neuroticism)."

While it is apparent that Eysenck, Cattell and others reported
in this section are dealing with similar dimensions, and that these

dimensions are somewhat comparable to SAR and ARO, there may be some

controversy over which dimensions align with which others. Hurley

8
(personal communication) identified introversion with the submission
pole and extraversion with the dominance pole of Leary. In this
formulation extraversion~introversion relates closely to SAR.
Eysenck's account, however, in a description to prospective buyers
of the Eysenck Personality Inventory, is somewhat different (Eysenck
and Eysenck, 1968). Neuroticism here seemed more related to SAR,
and extraversion-introversion to ARO. Suggestions from Wiggins
(1968) and Kassebaum et a1. (1959) also appeared indirectly to tie
extraversion—introversion more closely to ARO. The question remains
an empirical one which can be viewed within the framework of a cluster

analysis of the data of this study.

MMPI Studies:

Factor analysis of another widely used measure, the MMPI,
shmilarly led to the interpretation of a small number of factors.
Welsh (1956) claimed that the MMPI, which purports to measure a large
variety of factors, can be adequately described in terms of two.
Welsh identifies these as A and R. A may align more closely with
SAR, while R may align more closely with ARO (Adams, 1964). An
alternative interpretation suggests some overlap, with high A scorers
falling below the mean on both dimensions (Adams, 1964). Some question
may also be raised about the validity of A and R given the lack of
computer sophistication at the time of Welsh's work (Charles Wrigley,
personal communication). Kulik and Kulik (1972), however, using a
revised cluster analysis technique (VCLUST), identified two major

clusters whose similarity to A and R was commented on.

9

Kassebaum, Couch and Slater (1959) factor analyzed 32 MMPI
scales and identified two major factors as ego-strength-ego-weakness
(I) and introversion—extraversion (II). Welsh's A loaded highly on
the first while R boasted the highest loading on the second. In
addition, "R" was virtually the only variable independent of "I".
The R factor has also been interpreted as a measure of "acquiescence"
(Abbott, Fry, Abbott, 1972). The effect of response sets will be
discussed in a later section. Both studies mentioned incidentally
used already constructed scales of items in their analyses.

A later, more comprehensive effort by Robert Tryon, Kenneth
Stein and Chen-Lin Chu involved a cluster analysis of the total 550
item MMPI on a sample of patients and normals (Tryon, 1966; Stein,
1968; Tryon and Bailey, 1970). The analysis employed the BIGNV
component of the BCTRY cluster analysis system, which eliminates the
small N limitation of other analyses. Initially the program identi-
fied seven item-clusters: Introversion (I), Body Symptoms (B),
Suspicion (8), Depression (D), Resentment (R), Autism (A), and
Tension (T). Four of the dimensions, Introversion, Body Symptoms,
Suspicion, and Tension, accounted for most of the total communality
of the item pool. The remaining three were considered dependent
clusters. Eliminating nearly two-thirds of the total number of items
based on low communalities, the clusters could then be shown to form
scales (the TSC Scales). The full-length scales and shortened
versions were reported by Tryon (1966). Intercorrelations between

cluster scores are reported in Table l.

 

10

TABLE 1. Intercorrelations between cluster scores I, B, S, T
(Tryon and Bailey, 1970, p. 164)

 

 

I B S T
Introversion .47 .27 .68
Body Symptoms .32 .75
Suspicion .48
Tension

 

 

 

It is important to note that Introversion and Suspicion shared the
lowest intercorrelation of the clusters. Their low dependence on
each other is suggested by an}:2 of only .07. In addition, these two
clusters appear to have some of the same defining characteristics as
SAR and ARO. Suspicion probably relates to ARO, while Introversion
relates to SAR. Further study by Tryon (Tryon and Bailey, p. 200),
consisting of a comparison of the patient and normal subgroups of his
sample, revealed that for the normal group only the dimensions of
Suspicion and Introversion held up. Thus, once again two major

dimensions, similar to SAR and ARO, adequately described a domain.

Postulations from Encounter Grouijesearch

 

Hurley and Force (1971), in an extensive study of pre-lab, lab,
and post-lab data for 50 participants in an eight-day laboratory
experience, identified two clusters of personality variables. These
clusters conformed to their expected dichotomy of self and other

orientation. What is noteworthy is that:

 

11

"Within the Self-Acceptance and Other-Acceptance clusters,

46 of the 48 (96%) correlations are positive and sta-

tistically significant. A sharply different pattern is

apparent within the Across—Cluster Rectangles where only

15 of the 42 (36%) correlations are significant and

positive."

In addition, the ICL variables, Love-Hate and Dominance-Submission,
correlated highly in their expected clusters.

Data on the effects of trainers on their T—groups have also been
discussed by Hurley (in press, a). In addition to his own findings
which further confirm the saliency of SAR and ARO, Hurley has kept
up a correspondence with Bolman (Hurley, 1973; Bolman, l971a,b;
Hurley, 1971; Bolman, 1973, in press) (in Hurley, in press, b). During
this time Bolman conducted and reported on two independent studies of
the effects of trainers on T—groups. Most recently, Hurley (in press,
b) found that only one cluster accounted for nearly all of the sta-
tistically significant correlations among the eleven most important
variables utilized by Bolman (1971, 1973). This cluster's positive
pole was marked by a trio of measures (Trainer's Congruence-Empathy,
Liking for trainer, and Trainer's affection toward the group), while
the negative pole was denoted by another trio (Tension of group,
Conditionality of trainer, and Discomfort with trainer). Hurley
regarded this cluster as measuring ARO. Interestingly, the most
peripheral member of this cluster was the variable Dominance of
trainer, which seemed to represent SAR. This latter variable did not
correlate significantly with any of the ten other variables in

Bolman's 1971 study, and correlated only marginally (-.28) with

Other's Learning in Bolman's 1973 study. The independence of

 

12
Dominance of trainer from the six polar markers in both studies sug-
gests that Dominance is unrelated to ARO.

Lieberman, Yalom and Miles (1973) also reported on the leaders of
encounter-type groups. From data based on ratings by group members
and outside observers, four rotated factors of leadership function
were isolated: (l) Emotional stimulation, (2) Caring, (3) Meaning-
attribution, and (4) Executive function. Only caring and meaning-
attribution correlated significantly with outcome. Caring was
described as incorporating protecting, loving, supporting behavior.

It was clearly a warm/cold, love-not-love dimension, or ARO. Meaning—
attribution by contrast is representative of SAR. Meaning-attribution
was that function of the leader concerned with the translation of
feelings and behaviors into ideas. In so much as the leader gives
meaning to the experiences of the participants, he may be seen as
dominant. In the schema of this report, this type of dominance

aligns with SAR.

Given these data, it is plausible to conclude that further
research on encounter groups will continue to reveal the prepotency
of SAR and ARO. It is also likely that continued factor analytic
studies will reveal these dimensions and provide stronger empirical
backing for theoretical formulations such as those of Berne and
Leary. This study, as its first goal, using Tryon's Cluster Analysis
techniques with selected self-report personality measures, was designed

to do this. A summary is contained in Figure 2.

13

Accepts others

 

 

You're OK
Love Caring
Melancholic Sanguine
Depressive Healthy
Submission Dominance
I'm nut OK I'm 0K
Rejects self Accepts self
Introversion (Tryon) Meaning attribution
Phlegmatic Choleric
Schizoid Paranoid
Hostility

You're not OK
Rejects others

Suspicion and mistrust (Tryon)

Eysenck Extraversion—Introversion}
Eysenck Neuroticism-Stability

Figure 2. Summary of dimensional analyses.

 

 

va1

Ci.

¢_L.
("
I];

I!)

‘v
.“

u.‘ ‘

‘ h

‘:
0

n
on

14

Evidence for a Circular Order of Personality Variables

 

Up to this section the author has been concerned with selecting
evidence for a two-dimensional theory of personality. Further examina-
tion of Leary's system, however, suggested the possibility of
describing interpersonal behavior in a meaningful circular order
(statistically known as circumplex).

Foa (1961), in redefining Leary's types, discussed ordering them
in a circumplex. Statistically such an order may be said to be
present when

"the correlations between any two variables Xi and are

a function of their sequential separation. The correla-

tions of variable Xi should decrease monotonically and

then increase monotonically as we move around a circular

order." (Lorr and Mc'Nair, 1963)

Lorr and Mc’Nair (1963) began their study of interpersonal behavior
with 13 hypothesized factors of interpersonal behavior. They ended
up with nine primary factors which could be arranged in a circumplex
order. These are shown in Figure 3. From these primary factors they
obtained three higher order factors described as control, dependence,
and affiliation vs. detachment. Expanding the circle (Lorr and
Mc'Nair, 1965), they confirmed 15 interpersonal behavior categories

in a circumplex order. It may be noted from Figure 2 that the axes

of the Lorr—Mc'Nair circle are comparable to Leary's dimensions.

Relationship of the Dimensions
Research in the areas of self-acceptance and other-acceptance in
time late 1940's and early 1950's often suggested a strong positive ggx

‘relationship between self and other acceptance (Sheerer, 1949; Stock,

 

15

Dominant-
Controlling

    
    
  

Rebr lious

Susp cious Affi iative—
Trusting

Inhibit-d- Nur urant—

Supportive

Dependent

Figure 3. Lorr and Mc'Nair primary factors.

16
1949; Omwake, 1954). Frequently using five-point scale type measures,
experimenters reported correlations between self-acceptance and
other-acceptance ranging from .37 to .74. All correlations reported
in these studies were significant.

In contrast, factor analytic studies like those reported earlier
suggest the saliency of orthogonal factors. Remember that by defini-
tion orthogonal factors have a correlation of .00 with each other.
They are drawn by successive factoring from the intercorrelations and
residuals of the intercorrelations of the variables involved. Thus,
the appearance of dimensions interpretable as self and other
acceptance will, by necessity, be independent of each other. In
this case it is impossible to study those variables which may offer
some clues to the possibly complex relationship between the two. By
contrast, employing oblique clusters allows this. Oblique clusters
are assumed to be correlated until they demonstrate otherwise.
According to Tryon and Bailey (1970), oblique clusters provide a more
natural description of the domain.

The question of the independence of dimensions of self and other
acceptance-rejection, however, is influenced by the distinction
between statistical vs. theoretical definitions of independence.
Statistically, independent dimensions must be uncorrelated (g - .00).
Theoretically, however, it is unlikely that wholly independent
constructs exist in the personality questionnaire domain. By hypothe-
sizing independence for statistical purposes, the author is merely

suggesting a minimal interdependence between SAR and ARO.

 

Cu

 

.. Plyvfl‘

 

17

Clinically, Carl Rogers (1951, p. 520) suggested that the raising
of self-acceptance will automatically result in the raising of other-
acceptance. This is a currently held belief among many clinicians
and has led to techniques of psychotherapy which are essentially
client-centered and relate little to the client's relationship with
the therapist and others. Hurley and Force (1971), however, in their
study of these dimensions in the T-group, reported that

"only limited overall gains (in interpersonal competence)

are possible when increments in other-acceptance are

small, no matter how great the gains in self-acceptance

are."
This, too, is the conclusion reached clinically by this author and
several colleagues (personal communication).

The controversy over the relationship between SAR and ARC is
addressed in this paper. The BCTRY system presents a method for
describing the relationship of clusters. The second goal, therefore,

of this study is to support the contention that SAR and ARO are

functionally independent dimensions.

Response Set

 

In any study employing self-report inventories, it is necessary
to confront the possibility that one's chosen measures are actually
measuring a "response set" of the subject. The two most often dis—
cussed response sets are social desirability and acquiescence.
Wiggins (1968) attributed the variance involved in personality
inventories to four sources: (1) bybproducts of the strategy under

which the scale was constructed, (2) item characteristics which

A

l8
produce method variance, (3) organized response styles which exist
in_§s, and (4) manifestations of the particular content domain.

Edwards et a2. (1962), in an analysis of 58 MMPI scales and three
other personality scales, examined three factors. All these factors,
according to Edwards, are based on the response style of the subject.
Edwards stated: "We believe that the first factor can be best
described as a social desirability factor and the second factor as
an acquiescence factor." The third factor suggested the tendency of
_§s to lie. Social desirability (SD) is defined here as the tendency
to give a

"'True' response to an item with a socially desirable

scale value (SDSV) or a 'False' response to an item with

a socially undesirable scale value."

Acquiescence is simply the tendency to respond "True" to personality
items. Thus Edwards (1964) concluded:

"It may be argued that individuals who obtain high scores

on the trait scale are responding to the items in terms

of the trait which the scale was designed to measure,

but it is equally plausible that they are responding

to the items in terms of a trait which the scale was

not designed to measure, namely, the tendency to give

socially desirable responses."

To measure the SD factor, Edwards developed two social desira-
bility scales drawn from MMPI items, one of 79 and one of 39 items.
Substantial correlations were reported between the SD scale and
other MMPI scales (Edwards, 1957). Edwards (1964) also reported that
the correlations between 43 MMPI scales and the SD scale were
directly related to the proportion of items in the MMPI scales keyed

for socially desirable responses. Fox (1966) and Dicken and Wiggins

(1964) were unable to substantiate Edwards' claims.

'5

19

The relationship of SD and acquiescence to other interpretations
has been noted. Adams (1964) interpreted SD as comparable to the
Affection—Hostility axis of Leary (and therefore ARO). Wiggins
(1968) and Crowne et al. (Hurley, in press, a) reported a higher
correlation of SD with SAR. A comparison of Edwards et al. (1962)
and Kassebaum et a1. (1959) is also revealing. Welsh's A and R
appear to differentiate between factors I and II no matter what they
are called. In the Edwards et a1. rotated factor data Welsh's A
loaded -.93 on I (SD) and R loaded —.90 on II (Acq). In Kassebaum
et al. A loaded +.88 on I (ego strength), while R loaded +.69 on II
(extrav-introv). In fact, R "boasts the highest loading on II and
is virtually the only variable independent of I" (Kassebaum et al.,
1959). That R is measuring an acquiescence factor has also been
suggested by Abbott et a1. (1972, 1973).

It appears likely, then, that response sets not only account
for some of the variance in factor studies, but that they are probably
related to important personality clusters themselves. Whole books,
notably by Edwards (1970), have been written attempting to clarify
the role of response sets, with no satisfactory conclusions presented.
The present study does not attempt to deal with this subject compre-
hensively, or perhaps even adequately. A hypothesis relevant to the

response set question is, however, suggested.

Cross-Cultural Stability

 

Foa (1961) described interpersonal actions in terms of a series
of components (see earlier review under "Interpersonal Theory"). The

components are (1) content: rejection or acceptance, (2) intensity

7’:

 

r!-

 

AP
‘bb

g:

’5‘
s»

.IJ

Q; ,

. \t.

 

20

of action, (3) object: self or other, and (4) mode: social or
emotional. Foa suggested that "the direction of the third and fourth
principal components is culturally determined." Apparently, intensity
also plays a major role as Foa defined four cultural types. These are:

(l) The emotional mode is more intense than the social mode and
the individual comes before the collectivity.

(2) The emotional mode is also more intense but the collectivity
comes before the individual.

(3) The social mode is more intense and the individual comes

 

before the collectivity.

(4) The social mode is also more intense but the collectivity
comes before the individual.

The comparison of cross—cultural profile patterns is clearly

beyond the scope of this paper.

Summary_- Statement of Partial Hypotheses

 

Evidence has been reviewed relevant to the appearance of two
prepotent dimensions of personality which may best be described as
Self-Acceptance-Rejection and Other-Acceptance-Rejection. Further,
it appears likely that these dimensions are minimally interdependent.
The major goal of the present study is to assess this evidence.

Thus, two partial hypotheses can be stated.

H1: In the administration and subsequent analyses of a series

of personality inventories, two salient reliable dimensions will

occur. These dimensions can be described in terms of Self-

Acceptance—Rejection, and Other-Acceptance-Rejection.

21

H2:

These dimensions will be independent of each other.

Several sub-hypotheses may be stated in partial form. These
will become more defined in later sections of this proposal.

H3: A small group of items or scales can be drawn from the
large pool to adequately measure the dimensions.

H4: Eysenck's extraversionrintroversion factor will cluster
closely with SAR, while Neuroticism will cluster with ARO.

H5: Response set variables will not lessen the meaningfulness

of the dimensions.

In:

 

METHOD

Subjects

Subjects for this study were 182 undergraduate psychology
students at Michigan State University. A sample size of at least
160 was desired on the basis of an arbitrary acceptable standard
error of .08 in a commonly accepted formula for the standard error
of measurement. Students were offered extra credit in their classes
for participating in research programs. As additional incentive,
the students were offered feedback on their performance. A numbered

identification system preserved the students' anonymity.

Procedure
Subjects completed a battery of self-report personality measures

lasting approximately one and one-half hours. The shortness of the
battery presented a limitation to adequately sampling the domain of
personality measures of this type. The length of one and one-half
hours was justified by the experimenter on three grounds: (1) past
experience; acceptability to college students because of past
experience with tests of this length, (2) fatigue; one would expect
some fatigue to accumulate and negatively affect a much longer
battery, and (3) motivation; one would expect motivation to answer

thoughtfully and truthfully to decrease with the battery's length.

22

23

Test Battery

 

Choice of Instruments:

The goal of this study was stated as an assessment of evidence
for a two-dimensional interpersonal theory of personality based on
the dimensions Self-Acceptance-Rejection (SAR) and Other-Acceptance-
Rejection (ARO). Faced with a vast domain of available instruments,
the experimenter was forced to choose among them. The initial choice
in this case was to limit the study to self-report measures. While
this confined the study to the conscious level (Freedman et al.,
1951) of interpersonal traits, it also simplified and hastened data
collection and allowed for a larger sample than otherwise might have
been possible. The problems encountered by omitting data from the
"public" and "unconscious" sectors in favor of the "conscious" are
discussed below.

While the experimenter's degree of comfort and familiarity with
a given instrument undoubtedly influenced the domain from which he

chose, some broad criteria were operative: (1) simplicity-instru-

ments were chosen that were made up of yes-no, checklist, and semantic

differential type items; (2) clustering-—instruments were chosen from
which meaningful clusters were expected to emerge; and (3) self and

other dimensions-instruments were chosen which appeared related to

SAR and ARO [to better assess prior evidence, the experimenter was
obliged, for at least part of the battery, to use instruments and
scales employed in earlier studies; this criterion highly influenced
the choice of scales to be used when it was necessary to choose only

certain scales from a larger instrument], (4) shorter measure-

 

’3.

24
instruments were chosen which might reveal SAR and ARO in a shorter
instrument [this criterion was of import in choosing some shorter,
semantic scales of an experimental nature]; and (5) response set-
instruments were chosen which would incorporate ways of testing the
effects of response sets. The reader can see how these broad

criteria operationalized in the choice of each instrument below.

Self-Report Data vs. Behavior Ratings

 

Before going on to the instruments themselves, however, perhaps
some note should be made of the split between self—reports and
reports by others. The present study was limited to self-report
data. Unfortunately, the correspondence between this type of
questionnaire data and behavior ratings has been uniformly low.
While Cattell has argued that there is a "secure linkage" between
behavior rating and questionnaire factors, Becker (1960) disagreed.
Becker reviewed Cattell's data and arguments, along with his own,
and drew opposite conclusions. Peterson (1965) wondered whether
higher order dimensions might do better. In reporting a study by
Wetzel, Peterson concluded that for second order factors there is
some minimally significant correspondence between the methods. This
evidence is plainly an insufficient base for concluding that self-
report and report by others will yield identical results. The

present study, then, is restricted to self-reports of personality.

Instruments

 

(1) Minnesota Multiphasic Personality Inventory (MMPI):
The MMPI is so widely used that little description is necessary.

According to Dahlstrom and Welsh (1960), the MMPI "was designed to

 

 

 

 

'q
.4

a
b~¢

:g,a

fl)

25
provide an objective assessment of some of the major personality
characteristics that affect personal and social adjustment." The
instrument is made up of 566 True—False items. Initially these were
divided up into scales based on psychopathology. The major scales
are Hypochondriasis (Hs), Depression (D), Hysteria (Hy), Psychopathic
deviate (Pd), Masculinity-Femininity (Mf), Paranoia (Pa), Psychastenia
(Pt), Schizophrenia (Sc), and Hypomania (Ma). Another scale, Social
introversion (Si), was later added and routinely included in the
profile. Three validating scales, L (lie), F (validity), and K
(correction), were also included. It is common knowledge, however,
that more than 200 MMPI scales now exist. Many of these describe
"normal" rather than pathological behaviors. According to Dahlstrom
and Welsh (1960), the item pool may be arbitrarily classified into
a broad spectrum of content areas as shown in Appendix A, page 122
Factor analytic studies of the MMPI were presented earlier (Postula-

tions from Factor Analysis).

 

Because of the excessive length of the MMPI, three abbreviated
versions were considered: Tryon's four major clusters (Tryon and
Bailey, 1970), scales A and R (Welsh, 1956), and the Mini-mult
(Kincannon, 1968). Tryon's clusters (I, introversion; B, body
symptoms; S, suspicion and mistrust; T, tension and worry) plus a
shortened version of Edwards' (1970) Social Desirability scale form
the portion of the MMPI used in this study. Tryon's technique of key
cluster analysis was used to identify his four clusters (I, B, S, and
T). The technique, to be discussed later, eliminates and retains

items based on their communality. Thus, Tryon's clusters may be

 

26
presented as scales which purport to represent the full MMPI. Three
additional clusters found by Tryon were discarded because they were
extremely dependent on the other four. The instrument used in this
study, with items keyed for the appropriate scale, is included as
Appendix B, page 123. Internal consistency reliabilities for the
full scales (to be used in this study) and their shortened versions

are reproduced below (Table 2).

TABLE 2. Alpha reliabilities of full and shortened TSC scales
(Tryon, 1966)

 

 

Scale n Alpha (rtt') n Alpha (rtt')
I (Introversion) 26 .925 17 .911
B (Body Symptoms) 33 .919 17 .886
S (Suspicion) 25 .854 17 .830
T (Tension) 36 .923 17 .882

 

 

 

 

 

 

 

A discussion of Edwards' 18 item SD scale is included below (Response
Set).

Dimensions A and R (represented by corresponding scales) were also
considered. It was concluded that because of the lack of sophisti-
cation of computer technique at the time of that study.(l956), A and
R might be less valid a short form than others.

Although worthy of note, the Mini—Mult was also rejected for
the present study. It consists of 71 items drawn from the MMPI by

a process of clustering and choosing items within clusters that were

27
scored on the largest number of clinical and validity scales (Kincannon,
1968). Research using the Mini-Mult (Kincannon, 1968; Armentrout and
Rouser, 1970; Harford et al., in press) does not encourage its use
in this study. Armentrout (1970) obtained significant correlations
between the raw scores on the MMPI and the Mini-Mult (except males on
scale F). The Mini-Mult profiles did not, however, predict full
MMPI profiles, and the author discourages use of the shorter instru-
ment with a college population. In addition, the Mini-Mult seems

to predict the full measure more closely when extracted from it,

 

rather than administered separately (Streiner et aZ., 1973).

(2) Eysenck Personality Inventory (EPI):

The EPI measures two "superfactors" which, according to Eysenck
(1953), account for most of the variance in the personality domain.
These are extraversion-introversion and neuroticismestability. In
a review of the test, Lingoes (Buros, 1970) accepts the orthogonality
of the factors for non—psychotic samples. The instrument consists
of 57 "Yes-No" items and is available in parallel forms (A 8 B). The
availability of American college norms is useful. The test also
contains a "lie" scale. Split-half reliabilities are reported
between .74 and .91 (Eysenck and Eysenck, 1968). In addition to
measuring a small number of dimensions based on a factor analytic
model, the EPI provides an opportunity to examine the "fit" of
Eysenck's factors with SAR and ARO. A reproduction of the EPI is

included as Appendix C, page.128.

28
(3) Interpersonal Checklist (ICL):

The ICL is a

"l34-item list of words or phrases which may be used to

obtain self-descriptions or description by others with

respect to an interpersonal domain of personality."

(LaForge, 1973)
The instrument, an outgrowth of the work of Leary, was not copy-
righted by the authors (Suczek and LaForge, 1955). The experimenter
is encouraged to use his own judgment and creativity. The ICL
represents something of a marker measure in the present study. The
summary factors LOVE-HATE (Lov) and DOMINANCE-SUBMISSION (Dom),
to which the instrument may be reduced, appear to be suitable definers
of the hypothesized SAR and ARO dimensions.

In addition to the Lov and Dom summary scores, the ICL provides
two additional factors. Ain (Average Intensity) is based upon the
assignment of an intensity level to each item in the list. There
are four levels and each of the 16 subscore scales contains an equal
number of items at each level. Average intensity (Ain) can be used
as a measure of social desirability (LaForge, 1973). Number of Items
Checked (Nic) is self—defining and may be viewed as a measure of
acquiescence (LaForge, 1973). Internal consistency reliability

estimates (based on communalities) for the summary scores are all

above .90. The ICL is reproduced as Appendix D, page 130.

(4) Personal Orientation Inventory (POI):
The POI (Shostrom, 1964) purports to provide a comprehensive
measurement of values and behavior seen to be of importance in the

development of self-actualization. The inventory consists of 150

 

29
two-choice value and behavior items and is scored for the following:
Time Competence (Tc), Other/Inner Orientation (I), Self Actualizing
Value (SAV), Existentiality (Ex), Feeling Reactivity (Fr), Spontaneity
(S), Self Regard (Sr), Self Acceptance (Sa), Nature of Man (Nc),
Synergy (Sy), Acceptance of Aggression (A), Capacity for Intimate
Contact (C). A description of these categories is presented in
Table 3.

In a factor analytic study of the P01, Tosi and Hoffman (1972),
using an oblique rotation, identified three factors: (I) Extraversion,
(II) Openemindedness, and (III) Existential Non-Conformity. Highest
loadings on factor I were obtained by Acceptance of Aggression,
Spontaneity and Feeling Reactivity. Highest loadings on factor II
were obtained by Nature of Man, Synergy, and Time Competence. Highest
loadings on factor III were obtained by Existentiality, Self Accep-
tance, and Capacity for Intimate Contact. Silverstein and Fisher
(1972) reported somewhat comparable factors. The latter authors,
however, set out to measure what they believed to be a "built-in"
factor structure in the P01 based on item overlap from scale to scale.
The authors concluded that

"there is some suggestion that item overlap may be responsible

for at least the first two factors in the empirical data,

although the possibility remains that the item overlap

itself is due to relationships among the latent variables

that the POI was designed to measure."

The POI has also been compared with the EPI and the MMPI. Knapp
(1965) reported high negative correlations between EPI Neuroticism and

P01 measures of time competence (Tc) and self regard (Sr), while high

positive correlations obtained between EPI Extraversion and POI

30

 

TABLE 3. Description of POI categories. Categories on the POI are
presented as a series of opposites
Category

 

Tflme competent (Tc)

Lives in the present

(Time incompetent)
Lives in past or
future

 

Inner-Directed (I)

Independent, self-
supportive

(Other-directed)
Dependent, seeks sup-
port of others' views

 

Self-Actualizing
Value (SAV)

Holds values of self-
actualizing people

Rejects values of
self-actualizing
people

 

Existentiality (Ex)

Flexible in applica-
tion of values

Rigid in application
of values

 

Feeling Reactivity
(Fr)

Sensitive to own
needs and feelings

Insensitive to own
needs and feelings

 

Spontaneity (S)

Freely expresses
feelings behaviorally

Fearful of expressing
feelings behaviorally

 

Self-regard (Sr)

Has high self-worth

Has low self-worth

 

Self-acceptance (Sa)

Accepting of self in
spite of weaknesses

Unable to accept self
with weaknesses

 

Nature of man,
Constructive (Nc)

Sees man as essen-
tially good

Sees man as essen-
tially evil

 

Synergy (Sy)

Sees opposites of
life as meaningfully
related

Sees opposites of
life as antagonistic

 

Acceptance of
aggression (A)

Accepts feelings of
anger or aggression

Denies feelings of
anger or aggression

 

Capacity for Intimate
Contact (C)

 

 

Has warm interpersonal
relationships

 

Has difficulty with
warm interpersonal
relationships

 

 

31
variables spontaneity (S), self regard (Sr) and acceptance of
aggression (A). Shostrom and Knapp (1966) offered evidence suggest-
ing that the POI predicted therapeutic growth more completely than
the MMPI. In addition, POI Time competence was shown substantially
correlated (r = -.65) with the MMPI Psychastenia scale, the highest
intercorrelation between the scales of the two instruments.

Three POI scales were chosen to represent the instrument in this
study. Nature of Man (Nc) was chosen as a purported measure of ARO.
SAR was represented by the Self—Regard (Sr) scale. An additional
seven item scale was chosen (Idef) (Shostrom, 1964) which purported
to be a shortened measure of the Inner vs. Other Directedness scale.

The scales are presented in Appendix E, page

(5) Interpersonal Trust Scale:

Rotter (1967) defined interpersonal trust as

"an expectancy held by an individual or a group that the

word, promise, verbal or written statement of another

individual or group can be relied upon."
A 40-item Likert-type scale was developed on the basis of this defini—
nition. Inclusion of an item on the scale was based on internal
consistency with other trust items, relative independence of social
desirability (measured by the Marlowe-Crowne SD scale), and spread of
scores on the 5-point scale. Fifteen filler items were included to
partially disguise the purpose of the scale. These filler items were
eliminated in the present battery. The scale is "additive", meaning

that a high score "shows trust for a variety of social objects"

(Rotter, 1967).

32

Although the scale offered a number of items important to this
study, seemingly in relation to the dimension of Other-Acceptance-
Rejection, the present author would like to offer a cautionary note.
Viewing a high score on the interpersonal trust measure as necessarily
healthy and contributing to interpersonal competence may be naive.
Current events suggest that trust in social institutions (government,
police, etc.) is often misplaced. A study accentuating this point
was reported by Hochreich and Rotter (1970): "Have College Students
Become Less Trusting?" Data obtained on the Interpersonal Trust
Scale from 1964 to 1969 show a drop in mean scores (male and female
combined) from K = 72.41 to'K = 66.64. An examination of individual
items, however, clearly demonstrated that the drop in trust was less
likely in items where the students had personal contact with the social
agent (e.g., government). The items in this scale appear most
closely related to other-acceptance variables. The Rotter Inter-

personal Trust Inventory is presented as Appendix F, page 136.

(6) Semantic Differential Inventory (SEM):

This section represents a series of seven-point semantic dif-
ferential scales drawn from three sources.

(a) Eight scales were drawn from the Two Dimensional Inter-
personal Inventory, Experimental Version (Hurley, in press, c)
(Appendix G, page 138). The eight were chosen on the basis of inter-
correlations which demonstrated their ability to cluster into dimensions
of self-acceptance-rejection and other-acceptance-rejection. SAR
was comprised of shows-hides feelings, active-passive, strong-weak,

likes-dislikes self, and open-guarded. ARO was comprised of

33
gentle-harsh, accepting-rejecting, and warm—cold. Because of the
computer answer sheet, the scales were shortened from ten to seven
points.
(b) OK-ness ratings (Hurley-Force, 1971): Two semantic differen-
tial scales were chosen based on how OK the individual views both
himself (OKs) and others (OKo). These items were based on the theory

of Berne (1966) (described in Postulations from Theory). In the

 

Hurley-Force paper (1971), OKs correlated significantly and positively
with LaForge's Dom factor. OKo correlated significantly and positively
with LaForge's Lov factor. The two (OKs and OKo) also correlated
positively and significantly with each other, leaving some question

as to the ability of this simple measure to differentiate between

SAR and ARO. Unfortunately, by error, OKo was left off the instrument
in this study.

(c) A series of five semantic scales was included which appeared
to ask more directly about the hypothesized dimensions. These were
extraverted-introverted, dominant-submissive, stable-unstable, accepts
self-rejects self, accepts others-rejects others.

In addition to individual scale scores, two summary scores were
used accounting for the two hypothesized dimensions (SAR, ARO). SEM
1 (for SAR) included the SAR scales, from (a) above, I'M OK—I'M NOT
OK, Extraverted-Introverted, Dominant—Submissive, and Accepts-Rejects
Self, a total of nine scales. SEM 2 (for ARO) included the ARC
scales from (a) above, along with Stable-Unstable and Accepts-

Rejects Others, a total of five scales. The attempt to gain informa-

tion by using both overall scale scores and individual item scores

." .‘Mi""

 

34
presented a problem. Because of this mixture of scale and item
scores one might have expected the items to correlate highly with
their respective summary scores. Clusters built around these
groupings may have been somewhat artifactual.
The semantic differential inventory is reproduced in Appendix

H, page 139.

(7) Response Set Measures:

"set" to

For purposes of this study, measures of a tendency or
respond in a manner unrelated to the content of the items were drawn
from the larger instruments.

Social desirability, defined as the tendency to give a "True"
response to an item with a socially desirable scale value (Edwards
et aZ., 1962), was measured by Edwards' revised (1970) 18 item scale
derived from the MMPI item pool, and separately by the ICL Ain score.
Edwards' revised scale represented an attempt to control for
acquiescent responding by dropping all but nine of the False-keyed
items from the 39 item SD scale. Of the remaining 18 items, half
were keyed True and half False. Edwards reported the same mean level
of responding on both scales.

The Ain score is a measure of the average intensity of respond-
ing on the ICL. Items on the ICL were divided into four intensity
levels. Intensity and social desirability were assumed to be related
(Kogan, in LaForge, 1973). For a set of 126 items, Kogan found a
correlation of intensity with judged social desirability of -.74 in

a sample of 46 neuropsychiatric patient raters and -.73 in a sample

of 94 university student raters. LaForge suggested that this

35
correlation resulted from a tendency to judge extremely intense
interpersonal behaviors as inapparopriate in most situations, and
hence as socially undesirable. For this reason Ain was employed in
this study as an estimate of social desirability.

In addition to social desirability, acquiescence (tendency to
give a "true" response) and lying (tendency to fake) are response sets
worthy of note. The present study could only include one measure of
each of these. Acquiescence was measured by the ICL Nic score.

This score, number of items checked, in its extremes, was a measure

of a tendency to answer true (or in this case yes) to personality
items. The EPI Lie scale was used as a measure of the tendency to
lie. This tendency was regarded as different from social desirability
(Edwards et al., 1962).

It was not intended that these scales would comprehensively deal
with the question of response sets. That challenge was beyond the
scope of this study. Their purpose was to provide some measure of the
level of responding due to response set, and an idea of where these

response sets cluster with major personality dimensions.

Time of Battery

 

Times suggested for the measures were (in minutes):

(1) iMMPI scales 25 (5) Interpersonal Trust 10

(2) EPI 15 (6) Semantic differential 10

(3) ICL 15

(4) POI scales 10 Total 85 mins

The time above did not include time to read and understand instructions.

bl

36

Order of Battery

The tests were administered in the following randomly selected
order:

(1) Interpersonal Checklist (ICL)
(2) Personal Orientation Inventory Scales (P01)
(3) Interpersonal Trust Scale
(4) Semantic Differential Inventory
(5) Eysenck Personality Inventory (EPI)

(6) MMPI Scales (MMPI)

Statistical Method

 

Variables for Cluster Analysis:

An examination of the test battery described above resulted in
a tally of 33 possible variables for cluster analysis. Variables are
presented in Table 4. Next to each measure is its code name for the
study. These appellations were used interchangeably with full names
throughout this report. A table identifying the predicted clustering
of each variable, and actual clustering in the study, is included as

Table 18 (page 110).
Statistical Technique:

The BCTRY Cluster Analysis System

In order to determine a statistical technique appropriate to
this study, the author reviewed several techniques of factor and
cluster analysis. The system of techniques seemingly most suited

to both study and author is the BCTRY, developed by the late Robert

 

 

..fl...

 

ntu misfih uwnuns

\uauv Uni-u ~ .1-

i flu H

Q§EMU~ F-.N‘H‘1>

av

HQ .~ ¢\ (.3

 

37

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mm A<Hoa
H xmm xum
uzomam xmmzwmaouum
m nmzz suaafinmuamon Hmaoom ammuo< was
mzmazz AHV coname qumanIwaHuamoo<
u>Hau< m>auu<au>ammmm
mmomzz Ame aofiuaamsm maszmu amuse-caucus
scones Ame maouaasm scam mmmomm swan
Hszz aHv :oamum>ouuaH Hmzz nammm msonmnmoeam AZMmV swab
N ZMm ouoom mumaaom Qm< lawanMHn
H Sam muoom humssbm m<m ofiucmamm
m mHAHmm «Ha
smszm amfiUHuouamz
meHmm sowmum>muuxm Hmm H HmDMH madam umauH umaua Houuom
0H myooo< mumeuo muoufiumnmumuuu<
ammoo< mamm muumfimmumuamuo< m mmHom wummmm-mamm
mam<am magnumumanmumaa ozHom Auzv am: no musumz
szoo m>ammaanamuuamafiaoa zHHom emuumuauuumacH Hem
><MHNM .>ouuaHIvmuum>muuxm voxomno
eozH eo no: a.Hueo a.H Ae.uaouV e UHz unamuauumaasz
AmmeHa mamm mmxaamfinummxfia Hmauam zH< suamaouaH mwmum><
zm<3 sumznwaoo upmwmfia >ogaoH mum: .m> o>oa
zmmo smegmaouamao caucmamm zGQuUH aofimmaa
upam .m> mucmqaaon 40H
0.3323 2800 Cam: . mamum 9:5me ouamwmai «woo ABM”; . mason mgmfli
moanmfium> .muommmav maanum> moanwaum> .oHSwmuav manmaum>
no .02 mo .oz

 

 

 

mammamcm umumsau now moanmaum>

.e mgm<a

38
Tryon and associates and reported in Cluster Analysis (Tryon and
Bailey, 1970).

The system emphasizes the key—cluster analysis method of Tryon,
although more traditional factor analytic techniques are available
in the BCTRY computer network. The complete BCTRY system of programs
(except for BIGNV) is available at Michigan State University on the
CDC 6500 computer.

In Tryon's system, clusters of variables are formed (composites)
that are similar. This similarity is determined by observing the
objects in their relations to a common body of other referent enti-
ties. "To the degree to which the patterns of their relations to
these referents are the same, we judge them to be similar." This
leads to an index labelled the Index of’Proportionality, which is a
function of the degree to which the columns of correlations of the
two variables follow the same pattern. Variables having the same
profile of correlations are said to be collinear. An index of
collinearity, 2?, measures the degree to which the correlations of
two variables are consistently proportional across all the other
variables of the study.

The early steps in cluster analysis are based on these concepts.
The system can solve for clusters in two ways:

(1) V-analysis: reduces many specific attributes to a smaller
number of general attributes (dimensions).

(2) O-analysis: reduces many individuals to a small number of
general types.

‘The present study employed the V-analysis technique.

 

 

39

Two major steps make up the usual V-analysis and lead to the
definition of oblique rather than orthogonal clusters. Step 1,
Key-cluster analysis, meets three objectives: (1) selects from the
variables of the study the mutually collinear variables defining
each of the clusters; (2) discovers the minimal or salient number
(k) of clusters sufficient to reproduce all intercorrelations among
variables and the communalities of the variables; and (3) when there
are more clusters of variables than the minimally sufficient number,
provides information with which to select the most nearly independent
clusters on which to score individuals (Tryon and Bailey, p. 71).
These objectives are accomplished by factoring the correlation matrix
of variables to yield major dimensions. The program then chooses
as definers of each dimension the most collinear subset of variables
that is also most nearly independent of the definers of other dimen-
sions. Other variables can be added to these "pivot" variables based
on their}:2 value.

The second step, Cluster Structure Analysis, has at its objec-
tives: (1) to provide information permitting improvement in the
sets of definers of each cluster that have been selected by the
factoring procedures; (2) to discover how important each cluster or
factor is and how the different clusters or factors are related to
one another and to the variables of the study (p. 85). Cluster
Structure Analysis is performed both statistically and geometrically.
The cluster-defined dimensions are considered oblique in that they
vunild normally be expected to yield non-zero correlations with other

clusters. Tryon considers oblique representation superior to

40

orthogonal representation in that it reveals more "natural" clusters.
In contrast, the orthogonal dimensions derived from factoring
necessarily correlate .00 with each other. The difference is readily
apparent on the geometric diagram (SPAN) provided by the program.
Here, if the clusters were independent they would appear grouped
around the ends of their respective factor dimensions. Alpha relia-
bilities of the clusters are routinely computed.

These initial clustering procedures may be followed by a decision

to preset new definers and redo the analysis. This decision, like

_._‘-

the decision to keep the final clusters, should be based on five
criteria: (1) degree of collinearity of the definers of each
dimension, (2) degree of independence of the oblique cluster that
defines each dimension, (3) meaning of each oblique factor as a
construct, (4) contribution of the definers to the reliability of a
cluster score on the oblique cluster, and (5) generality of each
oblique cluster and of the variables that define it. For more
detailed discussion, the reader is referred to Tryon and Bailey,
Cluster Analysis.

In addition to computing basic dimensions, the BCTRY includes a
number of options employed in this study to deal with additional
questions. One such question reflected the concern that any given
battery was not representative of the total domain of personality
items related to a given cluster. The Cluster Structure Analysis
(CSA) component routinely computes a domain validity coefficient.
This coefficient is quantified by comparing the value of the correla-

tion coefficient of observed scores with the domain scores that would

41
be earned by the SS on an indefinitely large battery. The latter
is an estimate drawn from the inter-correlations between the definers
of a cluster based on their communality. It followed, then, that
an additional testable hypothesis was: "The item sample in this
study had a high validity coefficient in the personality questionnaire
domain."

A second question was the reproducibility of the factors over
several samples. The COMP component of BCTRY, based on preset
definers from the inclusive group, figures dimensions for each of
two subsamples and compares them. They are compared using both a
metric expression (the index of similarity) and their geometric
proximity on the SPAN component. This component can provide a
measure of cross validation within a large sample.

Finally, the choice of items to make up a smaller, more valid
measure was also a goal of the study. A testable hypothesis reads:
"A small group of items or scales can be drawn from the large pool

' Scales may be

to adequately measure the revealed dimensions.‘
selected as measurements of a given dimension by being best able to
both represent that dimension and distinguish it from others. The
scale's ability to do this may be determined by examining its factor
coefficients on the oblique clusters and/or its partial communalities.

Testing of any proposed new instrument was beyond the scope of this

study.

The Communality Problem

The use of communality estimates in factor analysis is the sub-

ject of longstanding controversy. Communality estimates are employed

42
to attempt to sort out common and unique variance. A researcher
interested only in the common factors which produce the intercorre-
lations between variables may employ communalities (Thurstone,
1961). This method seemingly omits the possibility of discovering
factors representing unique variance in the matrix, and might lead
to significant omission in the results. Nunnally (1967), however,

reports:

'm

"There have been numerous demonstrations that the

results obtained with unities in the diagonal spaces

tend to closely match the results obtained with

different communality estimates" (p. 355).

In order to note possible sources of unique variance, since

BCTRY employs communalities, a non-communality principal axis analysis

was included in this study.

Factor Analysis

The principal axis method selects factors to maximize the amount
of variance on the first factor, and so on. The resulting factors
were rotated to the varimax criterion. Varimax maximizes the sum of
variances of squared loadings in a column of the factor matrix. In
this way each factor is left with some high loadings and some loadings
near zero, thus making them more interpretable. The analysis was
done with unities in the diagonals of the correlation matrix and with
the program instructed to use up the total variance of the matrix
in selecting factors. This adjunct analysis, then, was not dependent

on communality.

43
Statistical Treatment

The present study employed the BCTRY system of cluster analysis
described above. Initially, in what amounted to a de facto pilot
analysis, a key-cluster analysis and cluster structure analysis were
carried out on the first 50 subjects. This allowed the meeting of
three objectives: (1) ironing out "bugs" in the author's use of
the BCTRY system; (2) providing the author with an initial "sense"
of if and how the chosen variables (scales) cluster; and (3) providing
data for the elimination of unreliable and unique variables (scales)
from future analysis.

The next step involved submitting data from the inclusive
sample of 182 subjects to the major components of cluster analysis:
key-cluster analysis and cluster structure analysis. At the same
time the total set of data was submitted to a non-communality
principal axis factor analysis with varimax rotation. Following
this, the author made decisions based on the empirical analyses,
while.at the same time running another analysis, presetting definers
for the clusters. This permitted addition or subtraction of definers,
or even whole clusters, based broadly on the criteria described on
page 40.

In addition, two comparative analyses (COMP) were run on sub-
sets of the data. The technique permitted the test of a new
hypothesis: "Subsamples of the larger inclusive sample reveal
identical clusters in the domain." The first COMP analysis compared
Inales and females within the sample. The second provided a cross

*walidation.within the sample. The total sample was randomly split

44
to form subgroups 0f.E = 72, and N = 110. The numbers conformed
roughly to a rule of thumb for this type of analysis where a sample

is divided into A = (2/3) N and B = (l/3)_N.

Statement of Hypotheses

The statistical treatment dealt with seven hypotheses. Restated

and renumbered, the hypotheses were:

H1: In the administration and subsequent analyses of a series

of self-report personality inventories to college students, two

 

salient and reliable dimensions occur. These dimensions can be

described in terms of Self-Acceptance (SAR) and Other-Acceptance-

Rejection (ARO).

H2: These dimensions are independent of each other.
H3: The item sample in this study has a high validity coef-
ficient with respect to the total personality questionnaire domain.

H Eysenck's introversion-extraversion factor will cluster

4:
closely with SAR,°while the Neuroticism-Stability factor will cluster

closely with ARO.

H5: Measures of response set variables will not occupy "pivotal"

positions in the major clusters, nor will their variance be explained

by the major clusters.

H Subsamples of a larger inclusive sample will reveal identical

6:
clusters in the personality questionnaire domain.

H7: A small group of items or scales can be drawn from the

larger pool to adequately measure the revealed dimensions.

RESULTS

Pilot Analysis

 

In order to satisfy the three objectives of a pilot analysis
(page 43), an empirical V-cluster analysis was obtained on a sub-
sample of the first 50 subjects. The results are reviewed briefly
before presenting the main findings.

The first objective of the pilot analysis was to iron out "bugs"
in the author's use of the BCTRY system. This was partially
accomplished. The second objective related to whether meaningful
clustering would occur. Clustering did occur, generally in the
hypothesized directions. Only four clusters of sufficient strength
to report were selected. Cluster 1 was defined by three semantic
differential scales (like self vs. dislike self, accept self vs.
reject self, and I'm OK vs. I'm not OK) and by the POI self-regard
(Sr) scale (POISR). Cluster 2 was defined by SEM 2 (a semantic
differential summary score related to ARO), the ICL Lov factor
(ICLLOV) and two semantic differential scales (accepting vs.
rejecting, and gentle vs. harsh). Cluster 3 was defined by MMPI
Tension and Body Symptoms, EPI Neuroticism, and sex. Finally, the

cluster 4 definers were SEM l (the SAR counterpart of SEM 2), MMPI

Introversion (reflected), and two semantic differential scales

45

46
(extraverted vs. introverted and dominant vs. submissive). These
four dimensions exhausted 90% of the initial communality.

A third objective, the elimination of non—clustering variables,
proved irrelevant. A sample of 50.§s on 33 variables failed to
amply demonstrate either uniqueness or unreliability. Furthermore,
the inclusion of all variables in the study was viewed as a positive
step. With some flexibility it was decided to accept the BCTRY

definition of not clustering; that is, less than .20 of the communality

accounted for by the domain.

Finally, another decision was based on the pilot findings. It
was noted above that the four retained clusters accounted for .90 of
the communality. A review of the residual-correlation matrix follow-
ing the removal of the common variance revealed a sizable number of
residuals in the .1000-.2999 range. This suggested that perhaps the
BCTRY version of clustering was not accounting for as large a number
of clusters or clusters comprehensive enough to best represent the
domain. It was decided to subject the total data to a non-

communality principal axis factor analysis concurrent with BCTRY.

Empirical Cluster Analysis (Total Sample)

Clusters

The empirical V-analysis identified six key cluster dimensions.
These dimensions accounted for .97 of the initial communality. This
section deals individually with the clusters selected in the analysis.
Raw correlations between variables of the study, used sparingly in

this analysis, are presented for the readers' interest in Appendix I,

47
page 140. Table 18 (page 110) provides a review of clustering by

variables.

Cluster 1 (ARC)

 

The first cluster contained SEM 2, the variable which contri-
buted the largest portion of the general variance, and was defined
by variables related to the acceptance and rejection of others (ARO).
The defining variables were SEM 2, the summary score on the semantic
differential purportedly measuring ARO, and three individual semantic
differential scales: Accepting vs. Rejecting (ACCEPT), Accepts
Others vs. Rejects Others (ACCOTH), and Warm vs. Cold (WARM). SEM
2 was additively made up of five scales, including the three just
cited. Of the other two components making up SEM 2, Gentle vs.

Harsh (GENTLE) was highly related to ARO, while Stable vs. Unstable
(STABLE) was relatively unrelated to any variables of the study. The
results of the initial clustering suggested that the ARO cluster may
have been somewhat artifactual due to the additive nature of SEM 2.
The subsequent addition of ICLLOV strengthened the interpretation

of the cluster.

The defining variables had a mean intercorrelation of .52, and
are presented in Appendix J, page 142. The diagonal values represent
estimated communalities of the variables which were included in
computing the mean. The correlations are generally above .5, although
tw° are “tab” 1”“ (lg-ACCEPT vs. ACCOTH ' '36 and EACCEPT vs. WARM ’
.26).

The first cluster accounted for .26 of the initial communality

estimate of the domain. The cluster had a domain validity coefficient

 

48
of .91 and an internal consistency reliability (for definers) of .82.
Validities and reliabilities of clusters on the empirical analysis are
found in Table 5, column 4. Column 2 of this table lists the total
communality (hz) of each variable in the domain. The communality of
SEM 2 was 1.00, while the communalities of the other three definers
were low, all between .33 and .36. Column 1 defines the proportion
of the totally2 that was accounted for by each cluster. For the
definers of ARO all or nearly all of their common variance within
the domain of this study was accounted for by that cluster. This can

be viewed by reading across from 01— in column 1 (e.g., for WARM,

C6
C1 accounts for .35, CA for .01, C6 for .01, with a totalh2 of .36).
Notably there are other variables that have a sizable proportion of
their communality on this cluster. MMSUSP shares .17 of an_h_2 of
.44 on this cluster. EPIEXT shares .15 of an};2 of .48. The figures
in column 1 were rounded to the nearest hundredth accounting for
minor disparities between partial and total_h2.

The oblique factor coefficients for the variables on each
cluster are presented in Table 6. The oblique factor coefficients
for the definers of ARO range from .56 to 1.00. Other variables with
notable factor coefficients on this cluster were ICLLOV (.62),
GENTLE (.49), and MMSUSP (-.42). The cluster may be viewed geo—
metrically in SPAN diagram 1. It is pictured at the lower left of
the diagram, clustered around the dimension marker "Z". SPAN diagrams
are plotted as three-dimensional spaces accounting for three

dimensions. Read the diagram so that the letters X, Y, 2, which

represent factored dimensions, are points on a sphere 90° apart.

 

49

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

em . bed @331
am. mquHHm> season
mm. mm. ow. oo. oo. oo. «H. no. Ho. m>Heo<
he. oo. No. oo. on. oo. ma. 9 ememm
Hm. oo. oo. Ho. we. Ho. oo. o zuzon
«s. oo. oo. oo. em. mm. HH. 9 Hszz
co. co. Ho. oo. mm. Ho. co. m ><maxm
Ibomll. m
m<m " o
a. 53333.5
3. 533m» 338
cm. om. «q. oo. He. no. no. NH. NH. mmamzz
no. oo. oo. oo. no. em. mo. a ems:
me. oo. oo. oo. Ho. so. so. a unomzz
on. oo. Ho. oo. oo. we. no. a szHmm
an. oo. oo. oo. so. «u. so. a mzmmmm N
qez “ u
we. musaanwaaum
Hm. hquHHm> damaog
cw. ow. mm. Nam
mm. mm. em. so. Ho. No. oo. no. «a. mqezmo
em. oo. as. oo. co. co. mm. a ammoo<
em. oo. oo. oo. co. co. cm. a msoou<
on. as. oo. Ho. oo. oo. mm. a 2M<2
cm. em. as. Ho. co. «on No. No. mm. >oaqu
co.H No. oo. oo. Ho. oo. HN.H n N mmm H
om< ” u
m . a no .35 aawfim so no so no No Ho 2: “333
museum Houmaau .Amfiamu vmvnnmwm N: Hoops (Nufiamaaaaoo Hmfiuumm mandamus
muoumsao wovomoxm mo ousuoouum noumsao oaoamm HouOu I mfiwhamaml> Hmoaufimsw do muoumoao .n mam¢a

50

 

 

 

 

 

 

 

 

 

 

 

so. suaaanmfiaum
ow. sufiufiamp nausea
om. so. am. ma. so. oo. co. so. as. zHHom
es. «m. oo. oo. no. no. No. a museum
Hm. mm. oo. Ho. mo. Ho. ea. c zmmo
em. em. oo.H 5H. co. am. an. «a. «a. H 2mm
UGO
memmmmmw "so
am. suaaanmaamm
mm. suaufiam> nausea
Na. co. es. oo. Ho. aH. no. a on
es. oo. mm. oo. oo. Hm. so. a 2H4 m
82.22 " 0
mm. suaafinmaaom
as. suauaaa> season
mm. ow. mN. oo. oo. mo. AH. no. No. ozomam
mm. am. am. Ho. oo. «a. co. mo. mo. mqm<sm
as. as. Ho. as. oo. Ha. om. ma. no. zonaoH
om. No. oo. HH. mo. om. mo. 9 mmHom
so. as. oo. ow. so. as. ma. a ammoo<
«a. co. so. on. No. as. so. a gmmMHq
«a. so. oo. on. No. Ha. as. a MQzH
luoawlll. v
m<m " o
m.n no .asu mawcfim so no so no No Ho any “unmade
museum umumoau .nmfiamu vwvcmmwm : Hmuoa muuaawon

 

 

N

 

huwauaoaaoo Hosanna

 

 

 

as.uaooV n magma

 

51

TABLE 6. Oblique factor coefficients — empirical V-analysis

 

 

 

Clusters
Variables l-ARO 2-MAL 3—SAR 4-SAR 5-AIN 6—Tran8w
soc emot
parency

POISR .30 .57 .49 .68 .28 .49
POINC .31 -.11 .13 .02 .06 .18
POIIN .08 -.25 .ll .19 -.35 .41
TRUST .21 -.18 .17 .13 -.02 .19
EPILIE -.04 -.13 -.20 —.05 -.03 -.10
EPIEXT .38 -.16 .64 .27 .ll .32
EPINEU -.16 .86 -.30 -.52 .51 -.26
MMTENS -.21 .89 -.20 -.48 .40 -.25

USP -.42 .50 -.06 -.19 .29 -.24

ODY -.19 .79 -.18 -.39 .40 -.23
MMINT -.33 .59 -.79 -.56 .25 -.49
MMSD .29 -.78 .49 .57 -.36 .31
NIC .22 .37 .05 -.09 .83 .00
AIN -.ll .48 —.23 -.34 .85 -.27
ICLDOM .19 -.46 .64 .66 -.14 .44
ICLLOV .62 -.02 .04 -.02 .14 .21
SHOFEE .15 -.22 .31 .23 -.14 .65
GENTLE .49 .04 .09 -.00 .21 —.03
ACTIVE .11 -.25 .44 .25 -.ll .26
ACCEPT .57 -.14 .21 .24 .12 .27
STRONG .13 -.24 .40 .44 -.11 .19
OPEN .39 -.21 .36 .34 -.08 .70
WARM .59 -.13 .22 .17 .00 .30
LIKSEL .24 -.38 .30 .84 -.16 .30
IMOK .33 -.40 .32 .84 -.14 .25
EXTRAV .24 -.14 .79 .28 -.07 .38
DOMIN .02 -.12 .66 .30 -.08 .24
STABLE .29 -.34 .17 .52 -.09 .15
ACCSEL .38 -.52 .43 .79 -.24 .30
ACCOTH .58 -.19 .23 .29 -.00 .27
SEM 1 .37 -.45 .79 .82 -.20 .82
SEM 2 1.00 -.27 .31 .42 .07 .31
SEX .29 .17 .10 .ll -.02 .04

 

 

 

 

 

 

 

 

 

52

LIKSEL ///// \\\\\
IMOK
STABLE SARemot // \\\\\

 

 

 

 

 

ACCSEL /
POISR // \\\\
I M§§USP
ACCOTH 1. / \
AC EPT a/’ \‘\
/" ‘\\ MMBODY
Ilflfll ‘\\\ \\\ MMTENS
WARM ‘ le MMSD
TRUST \_ / EPINEU
GENTLE "““‘ ‘*’““” lull.
ICLLOV

SPAN diagram 1 - Empirical analysis

53

In order to be plotted, a variable must have .80 of its communality
explained in the subspace represented by the diagram.

The Cluster Structure Analysis component added three variables
to the ARO cluster, each with its highest factor coefficient (£9) on
the cluster. ICLLOV was highly and predictably related with an_§g
of .62. Gentle vs. Harsh (GENTLE) was also predictably related, being
one of the semantic differential definers of SEM 2. SEX was also added
but was very minimally related (f3 = .29). ICLLOV had a large propor-
tion (.38 of .47) of its communality accounted for by the ARC
dimension. GENTLE was slightly more diffuse but still had .24 of

a total‘h2 of .34 accounted for by the cluster.

Cluster 2 (MAL)

 

Cluster 2 was a dimension not originally predicted that appeared
made up of variables related to pathology or maladjustment (MAL).
The cluster was defined by the Eysenck Neuroticism-Stability scale
(EPINEU) and three MMPI scales: Tension (MMTENS), Body Symptoms
(MMBODY), and Social Desirability (MMSD). MMSD was reflected in this
cluster, meaning that it was a reversal of the SD score that was
positively related to the others. The definers had a mean correlation
among them of .65 (Appendix J, page 142).

MAL accounted for .28 of the initial communality estimate. Its
domain validity coefficient was .95. Its internal reliability was
.91 (Table 5). The reliability and validity coefficients of this
cluster were the highest for the clusters selected. The communalities
of the defining variables on MAL were also high (Table 5, page 49),

ranging from .65 to .81. Again, as with ARO, the defining variables

 

54
had nearly all their communality accounted for by this dimension
(Table 5, column 1).

The oblique factor coefficients on MAL were high, ranging from
-.78 for MMSD to .89 for MMTENS (Table 6, page 51). Several other
variables were strongly related to this cluster. Two of these were
the other MMPI scales employed in the study. MMSUSP had a factor
coefficient of .50 on the cluster. The MMPI Introversion scale
(MMINT) had an £2 of .59 on the cluster. MMINT appeared later on as
a definer of one of the SAR clusters.

The second cluster may also be viewed in its relationship to
the other important clusters in SPAN diagrams 1 (page 52) and 2
(page 55). It seemed very tightly knit. MMSD's reflected status
is denoted by a minus sign on the diagram. On both diagrams the clus-
ter occupies a space right near the dimension definer. In diagram
1, MMSUSP, a related measure, appears about one-third of the way up
the sphere from the other tightly clustered variables.

Cluster Structure Analysis added MMSUSP to the maladjustment
cluster. MMSUSP was related to both MAL and ARO. It shared about
equal parts of its communality (.17 of .44 on each) with these two
clusters (Table 5, page 49) and had its highest oblique_fg's on the

two clusters.

Cluster 3 (SAR )
‘soc~
Clusters 3 and 4 were both related to the hypothesized self-
acceptance-rejection (SAR) dimension. Each consisted of a less

tightly grouped bunch of variables than ARO and MAL. The spread of

 

MAL

55

 

 

 

 

 

 

 

 

SAR
emot
LIKSEL
IMOK
/’//’ |'\\
/ \ STRONG
I SEM 1
/ \ ICLDOM
/ ' MMINT
/ \ \ "
/ O
/ EPILIE ACTIVE
MMSD ./’ \\\ I SARsoc
Z ’"MBODY \ DOMIN
MENS / EXTRAV
EPINEU \ /

SPAN diagram 2 - Empirical analysis

56
these variables can be viewed in SPAN diagram 2. ICLDOM and SEM 1,
both key to the concept of self-acceptance-rejection, did not occur
as definers of either dimension but were about equally dependent
on each.

Cluster 3 appeared related to a social description of SAR and
was labelled SARsoc' The cluster was defined by two semantic dif-
ferential scales, Dominant vs. Submissive (DOMIN) and Extraverted
vs. Introverted (EXTRAV), the reflected MMPI Introversion scale
(MMINT), and the Eysenck Extraversion—Introversion scale (EPIEXT).
These variables were related to the originally hypothesized construct
of acceptance and rejection of self (SAR). The defining variables
had a mean intercorrelation of .41 (Appendix J, page 142).

The cluster accounted for .18 of the initial communality estimate.
The cluster reliability coefficient was .84. Its domain validity was
.91. The total communalities were high for EXTRAV (.66) and MMINT
(.73) and moderate for DOMIN (.51) and EPIEXT (.47) (Table 5, page
49). EXTRAV and DOMIN, the semantic differential scales, had most of
their_h_2 accounted for by this cluster. MMINT shared a large amount
(.28) with MAL and a smaller amount (.11) with ARO of its total h? =
.73. EPIEXT shared .15 of its total h? of .47 with ARO. The factor
coefficients of these variables on ARO were -.33 and .38, respectively.
Factor coefficients of definers on this cluster ranged from .64 to
.79 (Table 6, page 51). In terms of these coefficients, MMINT was
the least specific of the definers. Its f3 on SARsoc was -.79, on

cluster 4 -.56, and on MAL .59. Other notable_£g's on this dimension

Twere.ICLDOM (.64) and SEM 1 (.79). Interestingly, for SEM 1 this was

 

57
only its third highest coefficient, being also highly related to
clusters 4 and 6.

The diffuse character of the SAR clusters seems clear. SARSOC
appears in the lower right of SPAN diagram 2 (page 55). DOMIN and
EXTRAV are closer to the dimension definer "X". The reflected
variable MMINT lies farther away from the definer. Its position is
somewhat indicative of its positive relationship with MAL.

The cluster structure analysis component added one variable to
the cluster: Active vs. Passive, a semantic differential scale. The
variable's totalh2 (Table 5) was so low (.20) that the variable is

peripheral to this cluster. Both SEM l and ICLDOM could have been

added but had slightly higher oblique factor coefficients elsewhere.

Cluster 4 (SARemot)

 

Cluster 4 appeared to represent the emotional aspects of SAR and
was labelled SARemot' The cluster was defined by three semantic dif-
ferential scales and by the self-regard scale of the Personal Orien-
tation Inventory (POISR). The semantic scales are Like self vs.
Dislike self (LIKSEL), I'm OK vs. I'm not OK (IMOK), and accepts
self vs. rejects self (ACCSEL). The mean correlation between the
items of this cluster was only .33 (Appendix J, page 142). The lower
correlations were those of POISR with the other definers.

This cluster, selected fourth, accounted for .12 of the total
initial communality. The cluster's reliability was reported as .88.
Its domain validity coefficient equalled .94. The defining variables

of this cluster had total communality values ranging from .56 for

POISR to .74 for LIKSEL (Table 5, page 49). The definers had a

 

58
large portion of their variance accounted for by the domain.
The partial communalities in column 1 of Table 5 further reflected
the diffuse structure of the cluster definers. IMOK, with a total_l_1_2

of .74, had .50 of this accounted for by SARem but also had .11

ot’
each accounted for by ARC and MAL. ACCSEL shared a much larger por-
tion of its communality with ARO and MAL (.15 and .19, respectively),
leaving only .26 to be accounted for on SARemot' POISR had twice as

much of its communality (.26) accounted for by MAL as by SAR (.11),

emot
the cluster it helped define. The oblique factor coefficients were
also illustrative (Table 6, page 51). The defining variables of the
cluster had fg's ranging from .68 to .84 on the cluster. POISR and
ACCSEL also had fg's over -.50 on MAL. Looking from left to right on
the definers of this cluster revealed that the definers had some
fairly sizable factor coefficients with all the clusters. Other
notable factor coefficients on this cluster were EPINEU (-.52),

MMINT (-.56), MMSD (.57), ICLDOM (.66), STABLE (.52) and SEM l (.82).
For SEM 1, this was only its second highest coefficient.

SPAN diagram 1 (page 52) offers a view of SARemot' The cluster
is spread out and lies slightly off the sphere and away from the
dimension definer Y. It is also leaning toward ARO, reflecting
their relationship. SPAN diagram 2 (page 55) also pictures the
cluster although it does not include all variables. The diagram
reveals the relationship between this cluster and SARsoc’ pictured
at the lower right.

Cluster Structure Analysis added three variables to this cluster,

based on having their highest oblique factor coefficients with it.

59
The variables were ICLDOM, Stable vs. Unstable (STABLE), and Strong
vs. Weak (STRONG), the last two semantic differential scales. STABLE
and STRONG both have low total-h2 in the domain (.31, .25, respec-
tively). The addition of ICLDOM, according to the reliability
estimates (Table 5, column 3, page 49) would raise the reliability
of the cluster.

It should perhaps be noted that the mathematical relationships
between definers of this cluster might seem insufficient to the
reader. The author will provide an interpretation which will, hope-
fully, add meaning to these relationships beyond the pure mathematical

intercorrelations.

Cluster 5

Cluster 5 was a doublet, defined by the NIC (number of items
checked) and the AIN (average intensity of items checked) measures
of the ICL. AIN purported to be a measure of the social desirability
response set and NIC purported to measure acquiescence. The raw
correlation between the two variables was .67. The cluster had a
reliability coefficient of .84 and a domain validity of .92. The
cluster accounted for about .06 of the total initial communality.
The partial communalities of cluster 5 (Table 5, page 49) revealed
that the variables had much of their communality accounted for by
cluster 5, but also shared a sizable amount with MAL. For AIN, with
a total h? of .76, .52 was on cluster 5 and .21 on MAL. For NIC,
'with a total};2 of .72, .46 was on cluster 5 and .19 on cluster 2.
The oblique factor coefficients (Table 6, page 51) were .82 and .85,

respectively. AIN and NIC were also related to MAL (fg's = .36 and

60

.47, respectively). One other notable relationship was that of EPI
neuroticism to this cluster (f9 = .51). The other two purported
response set variables correlated -.35 (MMSD) and -.03 (EPILIE)
with the cluster.

The CSA analysis added no variables to cluster 5; it remained a
doublet. It also appeared to the author that the AIN-NIC combination
may have been more an artifact of the ICL structure than psycho-

logically meaningful.

Cluster 6 (Transparency)

 

Cluster 6 (Transparency) was defined by the semantic differen-
tial scales Open vs. Guarded (OPEN) and Shows feelings vs. Hides
feelings (SHOFEE). The two were originally included as SAR scales.
The raw correlation between the two variables was .44. This cluster
was less reliable than the others (3 = .64). Its validity coefficient
in the domain was .80. The cluster accounted for about .06 of the
initial communality.

OPEN and SHOFEE had total hz's of .51 and .44, respectively.

.A survey of their partial communalities (Table 5, page 49) showed
that SHOFEE had most (.34) of its communality accounted for by the
Transparency cluster. OPEN had .28 accounted for by this cluster and
.16 accounted for by ARO. Oblique 23's of OPEN and SHOFEE were .70
and .65, respectively.

Cluster structure analysis added two variables to the Transparency
cluster: SEM l (oblique £3 = .82), and POIIN (oblique £3 = .41).
POIIN had a communality of only .27, which it shared among clusters

6, 5 and 2. The variable seems useless to the cluster at this point.

 

61
SEM 1, it has already been noted, was nearly equally related to this
cluster and to the SAR clusters (oblique_§g's - .82, .81). This
purely mathematical placement of the variable in this cluster was

not necessarily its best placement theoretically.

Relationship Between Clusters

 

The correlations between oblique cluster domains are presented

in Table 7. This is the simplest measure of the relationship between

TABLE 7. Correlations between oblique cluster domains

 

Clusters ARO MAL SAR SAR TR NIC-AIN
soc emot

 

A. Empirical Analysis

 

ARO -.25 .34 .39 .40 .06

MAL -.35 -.59 -.32 .50

SAR .49 .49 -.10
soc

SAR .42 -.26
emot

TRANSPARENCY -.16

NIC-AIN

B. Preset Analysis*

 

ARo - .21 .29* . 30* . 39

MAL - .35 - .59 - . 31

SAR .63* .49
80C

SAR .55*
emot

TRANSPARENCY

 

 

 

*
Asterisk indicates difference of at least .05 from empirical
analysis.

 

62
clusters. It is also useful to look at the relationships of indi-
vidual variables to the clusters to examine areas of overlap. This
is done by referring again to partial communalities and oblique
factor coefficients.

Cluster 1, ARO, related minimally to the other clusters. ARO
correlated -.25 with MAL. MMSUSP was highly related to both ARO and
MAL (oblique_fg's = -.42, .50, respectively). The correlation between
ARO and SARsoc was .34. EPIEXT had half as much of its communality

accounted for by ARC as by SARSOC ARO was slightly more related to

SARemot (r = .39). MMINT had factor coefficients with ARO and
SARemot of -.33 and -.56, respectively. ACCSEL showed some overlap
based on its partial communalities (.15, .19, .26 on C1’ 02 and C4,

respectively). SEM l was also a general variable in this sense.
Its partial communalities were .14, .14, .39, .21, .17 on C1, C2, C3,
C4, C6’ respectively. ARO's relationship of .40 with cluster 6
seemed partly accounted for by the overlap of the variable OPEN
{partial};2 = .16, —.28, respectively).

MAL was highly negatively related to SARem -.59, and

In
H

ot’

less related to SARSOC and Transparency, £_= -.35, r = -.32. It

was highly related to cluster 5,_£ = .50. MMINT was clearly a factor
in the relationship between MAL and the SAR clusters. Its oblique
factor coefficients were .59, -.79, -.56, respectively. In general,
variables with high positive loadings on SARemot had relatively high
negative loadings on MAL, and vice versa. ICLDOM and SEM 1 had high
positive loadings on both SAR clusters, and relatively high negative

loadings (-.46, -.45) on MAL.

 

63

The relationships of the SAR clusters with MAL are notable.
SARemot was more highly related to MAL (g = -.59) than to SARSOC
(g = .49). The SAR clusters were only modestly linked, however, to
ARO (r = .34, .39, respectively). These are key correlations as they
most directly represent the relationship of ARO and SAR.

The interdomain correlations, partial 2?: and oblique_fg are
not the only measures of the relationships between dimensions. An
examination of the SPAN diagrams 1 (page 52), 2 (page 55) and 3
(page 64) is useful. SPAN diagram 1 presents the subspace containing
ARO, MAL, and SARemot' Immediately seen is that ARO and MAL are
virtually 90° apart and cluster close to their orthogonal defining
coordinates. SARemot tends to cluster loosely and away from ARO
but is not 90° away. Diagram 2 is the space defined by MAL, SARSOC
and SARemot' The SAR clusters are both more than 90° away from MAL.
The SAR clusters appear related by the spread of dependent variables
between them. If the variables of these clusters were centered
closer to ICLDOM, we could probably interpret them as a polar oppo—
site of MAL. Diagram 3 is less useful. It does, however, demonstrate

a distance of less than 90° between MAL and cluster 5, and a distance

of more than 90° between MAL and Transparency.

Sufficiency of Clusters to Account for the Domain

 

The residual correlation matrix following the clustering pro-
cedure provided an initial estimate of the sufficiency of the clusters.
The number of residual correlations in the .1000-.l999 range was

included in Table 8. While several variables had five or six values

SPAN diagram 3 - Empirical analysis

65

.xmm auHa om. Hmstmmu maHa

“sum auHs mm. no Hmsvamu msH

m

ume nuHB ON. was UHz nuHB me.

 

 

«aaax «tan
«0 MHmovamu msHeaak .on :uHB me. no Hmavamu mzHaxa “MN.|oN. mo mHmanmou wousu maHo«
com
N m m<m HNMHmM
oom
N H m<m szon
com
N H m<m ><NHNM
com
m o M<m Hszz
m o MHm<Hm
e m Hmnma e m uzomum
H o ozHom m e zonHoH
uoam
seesaw m mHHHmm H H m<m MmHom
««««m e zHHom o o uoamm<m Humou<
H H MB mmmomm o o uoamm<m HumMHH
uoaw
H H my zmmo o o m<m MQZH
«Ram o 0H2 N N H mum
««N o zH< «e o Nmm
N H mmamzz N m mghzmu
o o H<z nmzz m N om< mHooo<
o 0 H4: Naomzz N H om< ammoo<
o 0 He: szHmm N H om< zm<3
o o H<z mzmazz N m >OHHUH
n n M>HHU< H N om< N sum
mHmovamu .oz MHmsvamu .oz .uHeam no moHanum> MHosvaou .oz mHmstmmu .oz .uHmao so mmHannm>
"ummmum ”HmoHuHesfl mumonmn "unmoum “HmoHuHQam muoaHmmn

 

 

 

 

 

 

 

 

 

AmHmmHmam HmoHuHmaov mwdmu mmoH.|oo0H. :H mHmsvaou mo umnazz .w mHm<H

 

66
in this range, the overall table suggested very little unresolved
variance. The proportion of communality accounted for by the
clusters also suggested little chance of any more meaningful

groupings in the data. The clusters were concluded to be sufficient.

Principal Axis Analysis

A principal axis factor analysis, with unities in the diagonals
of the correlation matrix, was performed on the total sample. The
analysis was run to conclusion, accounting for the total variance
of the correlation matrix. An examination of the factors before
rotation suggests similar groupings to those obtained by BCTRY
(Appendix K, page 143).

MAL variables EPI Neuroticism, MMPI Tension, and MMPI Body
Symptoms, had high negative loadings on factor 1. MMPI social
desirability, and SAR variables POISR, ICLDOM, LIKSEL, IMOK, ACCSEL
and SEM 1 had high positive loadings on this factor. In addition
'MMPI Introversion and SEM 2 were also highly related. ARO variables
ICLLOV, GENTLE, ACCEPT, WARM and SEM 2 all loaded highly on factor
2. Factor 3 was a maladjustment factor, loaded by EPINEU, MMTENS,
bflHSUSP and MMBODY, and interestingly by the semantic differential
scale Dominant vs. Submissive (DOMIN). Factor 4 was loaded by
EXIRAV, DOMIN, and STABLE.

Examination of the rotated factors (eigenvalue threshold set
.at 1n00) was confirming (Appendix L, page 145). Factor 1 (loaded
lxy EXPIEXT, MMINT, ICLDOM, EXTRAV, DOMIN, SEMl) was virtually identical
tr) SARsoc' Factor 2 was clearly ARO. Factor 3 was clearly MAL.

lflactor 4 was nearly identical to SARemot' Factor 5 was related to

 

67

EPILIE, which fell outside the communality of the cluster domain.
Factor 6 was the Transparency dimension. Factor 7, interestingly,
linked ACTIVE, WARM, and STRONG with loadings of .78, .44 and .46,
respectively. Factor 8 was the AIN—NIC dimension. Factor 9 was
important in that it linked POINC and TRUST along with MMSUSP and
ACCOTH. The factor loadings were POINC, .40, TRUST, .82, MMSUSP,
-.32, and ACCOTH, .37.

The grouping of POINC and TRUST was also noted in the unrotated
analysis, factor 10. POINC loaded .50, TRUST loaded .32, EPILIE
loaded -.30, and ACCOTH loaded -.37. The results suggested a
grouping represented by POINC and TRUST which lies somewhat outside

the study domain but is somehow related to ARO.

Preset Cluster Analysis — BCTRY

 

Introduction to Preset

 

The author believes that the clusters provided by the empirical
analysis are adequate for interpretation. Two major questions,
however, resulted in a preset analysis: (1) could the solution be
bettered by adding definers to some clusters and eliminating other
clusters altogether, and (2) specifically related to SAR clusters,
what happens if we add the important, but dependent, variables ICLDOM
and SEM 1 to one of the clusters? Financial and time constraints
limited the analysis to one preset. Since the aim of a preset of
definers and clusters is to aid in later interpretation, this was

deemed adequate.

68
The author performed several manipulations to prepare for the
preset analysis. First, new definers were added to two dimensions.
ICLLOV was added as a definer of ARO. ICLDOM and SEM l were added

to SARem because of a slightly stronger mathematical relationship

ot
with it than with SARsoc' It was decided, because of their relevance
to each other, to associate SEM l with ICLDOM rather than leaving it
to rescue the less reliable Transparency cluster.

The second manipulation involved elimination of cluster 5, made
up of variables NIC and AIN. The experimenter believed that this
cluster was useless to the current research, an artifact of the ICL
instrument. Further, Tryon and Bailey (1970, p. 102) recommended
eliminating clusters made up of doublets from the analysis. NIC and
AIN, because of their lack of relationship to other variables or
clusters of the study, were most clearly a doublet. Transparency,
however, also a doublet, was maintained because it did appear to
relate somewhat to other variables and clusters of the study.

Finally, the order of selecting the clusters was changed. This
manipulation changed the amount of variance drawn by each factor
during the initial factoring procedure. Theoretically the clusters
related to ARO and SAR are most important. Therefore those clusters
‘were factored out first. The reader should know the order used to
preset the clusters. Cluster 1, ARO, remained first. SARemot
(cluster 4), with added definers ICLDOM and SEM 1, was selected

second. SARSoc (cluster 3) remained third. MAL (cluster 2) was

selected fourth. Transparency (cluster 6) was selected fifth.

69

Preset Clusters

 

‘ARQ. ARO was defined by SEM 2, ICLLOV, Warm vs. Cold (WARM),
Accepts Others vs. Rejects Others (ACCOTH) and Accepting vs.

Rejecting (ACCEPT). The addition of ICLLOV strengthened the cluster
theoretically in that it then included two separate instruments both
related to ARO.

The mean correlation between the definers was very slightly
lower than it was on the empirical analysis (r, empirical = .51,

.5, preset = .49). This was accounted for by the slightly larger
distance of ICLLOV from SEM l as compared with the distance of the
semantic differential scales from SEM l. The proportion of communality
was also down very slightly from the empirical analysis. The prest
cluster accounted for .24 of the total common variance.

The reliability and validity of the cluster were very slightly
raised by the addition of ICLLOV (reliability = .84, validity = .92)
(Table 9, column 4). The table also contains the partial communali-
ties which are virtually identical with those on the empirical
analysis. Perhaps the most notable difference was seen by comparing
the oblique factor coefficients (£9) for ARO on the empirical analysis
(Table 6, page 51) and the preset analysis (Table 10). In general,
the oblique_£g's of the defining variables of the dimension, plus
that of GENTLE, went up. The_fg's of the other variables went down
on.this cluster. This represented an enhancement of the independence
of the cluster. The cluster may also be viewed in SPAN diagrams 4 and
5. The variables have remained around the dimension definer Z at the

lower left of the diagrams.

 

70

 

 

 

 

 

 

 

 

so. NuHHHanHom
No. hquHHm> oHoaog
No. No. HN. oo. NH. so. so. oN. N>Hao<
Ns. oo. no. NN. co. sH. a HHNHNN
mm. oo. NN. NN. No. oo. o szoa
NN. oo. mm. mN. No. No. a Hszz
so. oo. NH. ss. Ho. no. a ><NHNN
IUOQIII m
msm A o
Ha. NuHHHsmHHmN
ms. NuHsHHm> aHmaon
mm. mm. Hm. so. No. Ho. NH. no. on
so. om. os. NH. NH. NH. Ho. sz
om. oo. Ns. Ho. co. Ho. sN. NH. Nmomzz
ms. oo. on. No. NN. so. 9 nos:
so. oo. NH. Ho. as. no. a Naomzz
HN. oo. sN. oo. ss. No. a omzHNN
sN. oo. oH. Ho. Nm. no. a mzmmmm. N
H42 " o
sm. muHHHanHmm
No. NUHoHHm> aHmaon
no. mm. on. so. so. oo. oo. NN. NHazoo
sN. oo. Ho. co. co. Nm. a meooos
mm. oo. co. co. co. no. a amooos
Hs. oo. oo. co. so. as. a ass:
as. Ho. so. oo. oo. Ns. o >0HHoH
8H 8. Ho. 8. 8. 8H m N E H
ems " o
m.n so .azo oHoaHm no so No No Ho Hoe HmumaHo
monouo umumsHo .HMHHUH vmvamdwm N: HmHOH N: HMHuumm mumnHmoa

 

 

 

 

 

 

mumumoHo venomous mo musuoouum HoumoHo oHQBMm

Houou I mHthmaml> Hammad no muoumoHo

.m MHN<H

 

71

 

 

 

 

 

 

 

 

so. NuHHHowHHmN
oo. NuHoHHm> aHmaoo
so. so. mN. oH. so. oo. mo. oo. zHHoN
ss. No. No. No. oo. No. o monomm
oN. oH. Ho. oo. mH. o zmmo
a
.mmmmmmmw "no
mo. NuHHHomHHmm
so. 533.3 338
HA. No. Hm. No. NH. so. oo. oo. NHmsam
No. No. on. Ho. NN. Ho. Ho. Ho. ozomam
oo. oo. Hm. No. oo. Ho. o zooHoH
on. Ho. as. oo. no. oo. o mmHoN
so. no. on. Ho. Ho. HH. o Homoos
NN. Ho. oo. oo. Ho. oo. o onNHH
NN. so. oo. oo. Ho. No. o MozH
oo.H NH. sN. oo. Ho. oo. o H zoo
local s
Nam ” o.
m.o so .aoo «Hmch mo so mo No Ho Hoe youmsHo
mmuoom umumaHo .anHmu vwncmmxm N: Hmuoa N: HmHuumm mumaHmma

 

 

 

 

 

 

 

 

Ho.uaooo o moose

72

 

 

 

 

 

 

 

 

 

 

TABLE 10. Oblique factor coefficients - preset analysis
Clusters
Variables 1-ARO 2-MAL 3-SAR 4-SAR S-Transparency
soc emot
POISR .24 -.57 .49 .74 .48
POINC .30 -.12 .13 .04 .18
POIIN .01 -.25 .11 .20 .41
TRUST .22 -.18 .17 .16 .19
EPILIE .02 -.13 -.05 -.07 -.10
EPIEXT .37 -.16 .64 .32 .32
EPINEU -.14 .86 -.30 -.53 -.26
TENS -.17 .89 -.20 -.46 -.25
USP -.41 .50 -.06 —.16 -.24
ODY -.16 .80 -.18 -.39 -.23
INT -.28 .59 -.79 -.65 -.48
D .24 -.78 .49 .59 .31
NIC .23 .37 .05 —.06 .00
AIN -.07 .48 -.23 -.35 -.27
ICLDOM .ll -.46 .64 .71 .44
ICLLOV .65 -.02 .04 —.04 .21
SHOFEE .14 -.22 .31 .33 .65
WGENTLE .54 .04 .09 -.04 -.03
ACTIVE .06 -.25 .44 .36 .26
.ACCEPT .59 -.14 .21 .23 .26
STRONG .08 -.24 .40 .52 .19
OPEN .39 -.21 .36 .42 .70
WARM .63 -.13 .21 .18 .30
LIKSEL .18 -.38 .30 .79 .29
IMOK .26 -.40 .32 .79 .25
EXTRAV .22 -.14 .79 .40 .38
DOMIN -.04 -.12 .67 .44 .24
STABLE .25 -.34 .17 .49 .15
ACCSEL .32 -.52 .43 .77 .30
SEM 1 .31 —.45 .79 .91 .82
SEM 2 1.00 -.27 .31 .40 .31
SEX .28 .17 .10 .08 .04

 

 

73

wow

MMTENS EP INEU

NIC I \

/ l \

 

 

 

 

SEX
/ / \ \
/ \ LIKSEL
GENTLE / SEM
*ARM \ I OK STRONG
ICLLOV ACCEPT \ X SARemot
s a 2
AGGOTR\ flLmn
C x /
POIN
TRUST STABLE POISR
MMSD

SPAN diagram 4 - Preset analysis

74

/l\
’/” 1LEXTRAV DQQiE.

 

 

 

 

 

 

 

 

 

/// I SARsoc ACTIVE
POINC "MMINT \
GENTLE l
I LLO l
C V w RM / \ I DOM
/ TRUST \ SEM
[2:1ng \ STIYNG
‘\\
CCOTH
sm\ \
POISR
\ / SAR
I"”’*CCSEL emot
IMOK
LIKSEL
STABLE

SPAN diagram 5 - Preset analysis

75
The Cluster Structure Analysis again added GENTLE to the cluster.
Table 9, column 3 (page 70) demonstrated that the addition of GENTLE
to the ARO cluster would raise the reliability of the cluster. The
author therefore considers GENTLE to be a member of the cluster in

further discussion.

-§é§emot* SARemot represented the greatest change from empirical
to preset cluster analysis. The cluster was defined by IMOK, LIKSEL,
ACCSEL, POISR, SEM l and ICLDOM. The addition of the two key
dependent variables, ICLDOM and SEM l, greatly strengthened this
cluster. The mean intercorrelation between definers rose more than
.20 from the empirical analysis, to become .56. Partly because of
these additions and partly because of being preset second, the cluster
accounted for .35 of the initial communality.

The domain validity of the cluster was .96 (rather than .94) and
the reliability was .93, up from .88. The total communalities of the
definers remained high. The greatest changes were observed in the
partial communalities (Table 9, page 70). POISR, which had so much
of its variance accounted for by MAL, now had nearly all of it (.49
of a totalh2 of .58) on SARemot' ICLDOM and SEM 1, both of which

had been extremely diffuse in the empirical analysis, have most of

their communality accounted for by SARem on the preset. Their

0t
oblique factor coefficients (Table 10, page 72) were also higher on
the preset. POISR had a higher coefficient (.74 from .68), while
the other definers had 23's that were slightly lower than on the

empirical analysis. The reason for this can be seen by comparing

SPAN diagrams l, 2 and 5 (pages 52, 55, 74). NOte in diagram 1 that

76

POISR is the definer of SARemot farthest from the dimension definer
Y. Then in diagram 2 observe that ICLDOM and POISR are midway between
the clusters. The result of adding ICLDOM and SEM l to the cluster
can be viewed in diagram 5. POISR is now closer to being the center
of the cluster, while the semantic differential items are slightly
farther off.

The cluster structure analysis component again added STRONG and
STABLE to the cluster. Table 9 (page 70) demonstrated that neither
variable raises the reliability of the cluster. In addition, they

had low (.30) communalities. The author does not consider them part

of the cluster.

SAR was preset third, after SARem , because of the

SAR .
-—-soc- soc

addition of ICLDOM and SEM l to the latter. The cluster was defined

0t

by EXTRAV, DOMIN, MMINT (reflected) and EPIEXT, as before. The

decision to add ICLDOM and SEM l to SARem , and then to preset

ot
SARSOC third, in some ways mathematically weakened this cluster.
The mean correlation between the defining variables lessened from
.41 to .30, reflecting a spreading out of the cluster. Again, this
is pulled down by low, inaccurate communality estimates. The
cluster, in this position, accounted for .10 of the communality.

In other ways there was little change in the cluster. The
reliability remained virtually the same at .84, and its domain
validity coefficient also remained virtually equal (.92) to its

‘value on the empirical analysis. Interestingly, the oblique factor

(:oefficients for the defining variables of the cluster also remained

77

the same, reflecting their stability in relation to the other clusters.

A somewhat different note was sounded in viewing the communali-
ties, both total and partial (Tables 5 and 9, pages 49, 70). While
the total h? remained the same, partial hz's on SARSOC were greatly
lessened for MMINT, EXTRAV and DOMIN. For MMINT there was a definite
shift from sharing .28 and .11 of its variance on the empirical
analysis with MAL and ARO, respectively, to sharing .23 of h? with
SARemot’ and virtually none with ARO and MAL. The diffuse nature of
this cluster is also described in SPAN diagram 5 (page 74). In the
diagram it is difficult to suggest a center for the cluster. It can

be seen that MMINT, like ICLDOM and SEM l, is more closely related

to SAR than on the empirical analysis.
emot

.55E- The maladjustment cluster was selected fourth in the preset
analysis in order to allow the ARO and SAR clusters to pick up a
larger part of the correlation matrix. This, of course, led to a
lessening of the total communality accounted for by the cluster. Its
proportion of the communality was .15, as opposed to .28 in the
empirical analysis. The cluster definers remained the same; MMSUSP
was not added as a definer of MAL. Its relatively lowh2 (compared
with the cluster definers), relatively low oblique_£g, and position
on the SPAN diagram did not suggest any gain from adding it as a
definer. In addition, adding it would not have increased the cluster
reliability coefficient (Table 5, page 49).

Clearly, then, MAL with its original definers remained a salient
cluster. Its reliability and domain validity remained about equal

to their values on the empirical analysis (.91, .95, respectively,

78
Table 9, page 70). Interestingly, the interrelatedness of this
cluster also suffered in the preset. The mean correlation among
the definers equalled .45, a reduction of about .20.

This is reflected in SPAN diagram 4 (page 73). The cluster
appears at the top of the diagram, grouped loosely around the
dimension definer Y. The defining variables MMBODY and MMTENS
clustered to the left and closer to MMSUSP than previously. EPINEU
spread the cluster to the right. MMSD appears in the lower right
of the diagram, reflecting its polarity with the other MMPI scales.
Had it been reflected and appeared at the top of the diagram, it
would have been off the page.

Total communalities of the definers, while remaining high,
were generally lower than on the empirical analysis (Table 9, page
70). The range of_112 on this cluster was from .65 to .74. Note
also that the partial communalities of the definers were more diffuse
(Tables 5 and 9, pages 49 and 70) than on the empirical analysis.
EPINEU, for example, had .30 of its total_l'_1_2 of .71 accounted for by
SAR . MMSD, as a more surprising example, had approximately equal

emot

portions of variance on MAL and SARemot° This was apparently related

to the addition of ICLDOM and SEM 1 to SARemot and was a result of
the sensitive relationship between the MAL and SAR clusters. The
oblique factor coefficients for all variables on MAL remained about
the same as they were on the empirical analysis (Tables 6 and 10,
pages 51 and 72).

Three variables were added in Cluster Structure Analysis: MMSUSP,

AIN, and NIC. MMSUSP was once again most highly related to MAL

79
(f9 = .50) but maintained its relationship with ARO (£3 = -.41)
Once again its addition would not have raised the reliability of
the cluster (Table 9, page 70). It is, therefore, not considered
a true definer of the cluster. AIN and NIC, having been eliminated
as a cluster pair, correlated most highly with MAL. Their oblique
_§g's were .48 and .37, respectively. Without their own cluster
they have very low total_h2's in the domain (.40, .31, respectively).
They were not added as cluster definers then, for purposes of this

analysis.

Transparency (TR). Because of its possible uniqueness as a

 

cluster, Transparency, while only defined by two variables, was
retained for the preset analysis. With SEM 1 assigned to SARemot’
there was no chance of raising the reliability of the cluster.

In fact, all the mathematical relationships of this pair of

definers remained virtually identical to the results of the empirical
analysis and therefore will not be repeated in this section. Cluster
structure analysis once again added POIIN to this cluster. The
reader may compare Table 9, columns 3 and 4 (page 70) to see that

its addition as a definer of this cluster would not have raised the

reliability of the cluster. This cluster then remained defined by

a doublet.

Relationship Between Clusters - Preset

 

The oblique factor correlations between clusters for the preset
analysis are presented immediately below those for the empirical

analysis for easy comparison by the reader (Table 7, page 61).

80

Mbst notable amongst these intercorrelations was the reduction in
the relationship between SAR and ARO. In the empirical analysis,

EARO vs. SAR = .34 and EARO vs. SAR = .39, while the preset
soc emot

reduced this to_r__ARO vs. SAR = .29 and-EARO vs. SAR = .30,
soc emot
more closely approaching independence. The relationship of ARO
with MAL was also lessened (£_= -.25 to.£ = -.21), but MAL's rela-
tionship with both SAR clusters remained essentially the same. The
relationship between the two SAR clusters was also higher on the
preset (E = .49 to_£ = .63). Finally, the relationship between
Transparency and SARemot changed from .42 to .55.

One may also observe the changes of the SAR clusters by comparing
their partial communalities (Tables 5 and 9, pages 49 and 70). The
definers of SARSOC empirically had most of their communality accounted
for on that dimension. On the preset they shared more of their
communality with SARemot' The definers of SARemot which empirically

shared a sizable portion of their communality with MAL, on the preset

had most of it accounted for on SAR .
emot

Preset Summary NOte

 

The preset key cluster analysis accounted for .92 of the total
communality of the domain. The decrease was primarily due to the
elimination of the NIC-AIN cluster. Total communalities of individual
variables within the domain were generally bettered, with the notable
exception of NIC and AIN (see Table 9). Again with these exceptions,
and some larger residuals for SEX, the number of residuals in the

.1000—.l999 range was not greatly increased. Tryon refers to this as

81

a criterion for accepting the statistical soundness of the preset

analysis.

Non-Clustering_Variables

The BCTRY system eliminates variables as non-clustering if their
common variance with the domain is less than .20. In the empirical
analysis this category included three variables: the Nature of Man
(POINC), Rotter's Trust Scale (TRUST), and the EPI Lie scale. SEX
was added by the preset. Four other variables had communalities of
below .30 on the preset: Inner-directedness (POIIN), ACTIVE, AIN, and
NIC. Note once again that POINC and TRUST, which lie outside this
domain, appear to share some unique factor as suggested by the
principal axis analysis.

Variables that do not cluster may be either unreliable, or a
source of unique variance. Rotter (1967) reported a split-half
reliability of .76 for a sample of 547 students, suggesting that the
TRUST variable may be unique. No appropriate measure of internal

consistency was reported for POINC.

Response Set Measures

 

Four supposed measures of response set were included in the
variables of the study: MMSD, EPILIE, AIN, NIC. AIN on the ICL
and the SD scale of the MMPI purported to measure social desirability.
The EPI lie scale was purported to measure a tendency to lie. NIC
on the ICL was supposed to be a measure (albeit a poor one) of

acquiescence.

82

EPILIE had a communality below .20 in the domain and therefore
did not cluster. A review of its raw correlations (Appendix I,
page 140) showed it to be virtually unrelated to other variables of
the study. It does appear in the principal axis analysis as a
definer of factor 5 (rotated). The MMPI Social Desirability scale
(reflected) was a definer of the MAL cluster. MMSD was also highly
related to the two SAR clusters (fg's = .49, .59, respectively).
AIN and NIC clustered together on the empirical analysis. When this
cluster was eliminated, AIN and NIC clustered loosely with MAL
(fg's = .48, .37, respectively).

The raw correlations between the purported measures of response

set are presented below in Table 11. AIN correlated .67 with NIC but

TABLE 11. Correlations between purported measures of response set

 

 

Variable AIN NIC MMSD EPILIE
AIN .67 -.37 .06
NIC -.23 -.12
MMSD .01

 

 

 

insignificantly with the others. This is puzzling in that AIN and
MMSD were both supposed to measure social desirability. NIC also
correlated very low with MMSD and EPILIE. MMSD and EPILIE did not

correlate at all.

 

83
Comparison Between Subgroups - Male vs. Female

In order to perform the analysis, the data were divided into
males and females. There were 97 females, 76 males, and 11 §s who
did not identify sex. The COMP analysis program of BCTRY demands
that clusters for each group be preset on the same definers. The
definers from the preset analysis were used. The program computes
an index of collinearity between the clusters for each group which
is geometrically equal to the cosine of the angle between them
(cosine 6). The cosine 6 values may be read like correlation coef-
ficients. The program also performs an empirical V-analysis using
the original clusters as variables.

The table of cosines between dimensions for males and females
is presented below (Table 12). The numbers in the diagonal repre-

sent the similarity between groups for each individual cluster.

TABLE 12. COMP analysis for males and females

 

Cosines among dimensions

 

 

Females
ARO SAR SAR MAL TR
emot soc

Males ARO .67 .30 .26 -.28 .37
.38 .82 .62 -.57 .53

emot
SAR .37 .60 .81 -.44 .54

soc
MAL -.38 -.61 -.45 .79 -.44
TR .36 .58 .62 -.41 .62

 

 

 

 

 

 

 

 

 

84

Thus, the relationship between ARO (male) and ARO (female) was repre-
sented by a cosine of .67, a moderately high value. The cosine
between SARemot (male) and SARemot (female) was .82. Between SARSOC
(male) and SARSOC (female) the cosine was .81. Between MAL (male)
and MAL (female) the cosine was .79. The relationship of Trans-
parency for males and females is .62.

The higher order V-analysis generated three clusters. Their
definers, oblique factor coefficients, reliabilities and validities
are reported in Table 13. In this case, without referring to the
specific statistics, note that the male and female clusters for each
dimension, ARO, SAR (both clusters), MAL, and Transparency, did
cluster with their counterparts, with one exception. In cluster 1,
the two SAR dimensions cluster together for both males and females.
Cluster 2 represents ARO, again for both males and females, and also
reflects the relationship of the Transparency cluster for females
to ARO. Cluster 3 represented the relationship of at least part of
the SAR dimension to MAL.

The reader is also referred to SPAN diagram 6, which is some-
what more difficult to read than the previous diagrams. Note that
the author has drawn in double arrows between clusters that represent
the same dimension for different groups. For example, ARO is repre-
sented in the lower right corner of the triangle. Its configuration
between males and females appears extremely close and is pictured
by the small double arrow. What is most important in comparing the
hypothesized dimensions over the two groups is that the reader note
the length of these arrows. Transparency is abbreviated as (TR) in

the SPAN diagrams.

85

 

 

TABLE 13. V-analysis performed on preset clusters as part of COMP
analysis - males vs. females
Oblique is on Total Cluster
Clusters Definers cluster communality scores
C1
SAR - MALE D .83 .75
emot
SAR - MALE D .82 .71
soc
SAR - FEMALE D .82 .74
emot
SAR - FEMALE D .81 .69
soc
TR.- MALE D .77 .62
TR - FEMALE D .70 .54
Domain validity .92
Cluster reliability .96
C2
ARO - FEMALE .92 .85
ARO - MALE .69 .50
TR - FEMALE .64 .54
Domain validity .91
Cluster reliability .82
C3
SAR - FEMALE D -.84 .74
emot
SAR - MALE D -.84 .75
emot
MAL - FEMALE D .80 .68
Domain validity .95
Cluster reliability .91

 

 

 

 

 

 

 

86

 

 

 

 

 

 

 

 

 

 

 

 

/ TR \
l ’ \
Cl / C2 ARO \’
\
SAR
soc
SAR

soc

C3
-MAL

SPAN diagram 6 - COMP analysis for males and females
(higher order clusters)

87
One more result to aid in identifying differences between
groups is presented. The author performed empirical V-analyses on
each group to see if any differences in clustering occur. The
variables in each cluster are presented side by side in Table 14.
The positions of individual variables will be discussed in the

discussion section.

Comparison Between Subgropps - Non-Specific Groups

 

In order to perform the analysis, the data deck was shuffled
and divided into subgroups of_N = 110 subjects and N = 72 subjects.
The definers of the preset cluster analysis were used to prepare
the groups for comparison. Cosines between clusters for the two
groups are presented in Table 15. Once again, it is the numbers in
the diagonal that are of greatest interest. The cosine value between
random groups for ARO was .68, again moderately high. For SARemot
the cosine was .80. For SARSOC it was .77. The highest cosine
between random groups is attained by MAL (cosine = .83). Trans—
parency attained a cosine of .62 between groups.

The V-analysis on the clusters of both groups resulted in three
clusters (Table 16). Cluster 1 appeared to include both SAR.dimen-
sions and MAL, which was highly negatively loaded on the cluster.

The reliability of the cluster is .92 and its domain validity is .96.
Transparency appeared to span clusters 1 and 2, representing perhaps
a bridging of these major dimensions. Cluster 2 was the ARC dimension
with ARO for both groups loading highly on the cluster. The cluster's

reliability was .82 and its domain validity .91. Cluster 3 repre-

sented a relationship between MAL for both groups and SARSOC for both

88

 

 

 

 

 

 

 

 

 

 

TABLE 14. Empirical clustering of variables - male vs. female
Variables
Cluster Male Female
SEM 2 (D) SEM 2 (D)
ICLLOV (D) ACCOTH (D)
WARM (D) GENTLE
ARO ACCEPT (D) ACCEPT (D)
ACCOTH ICLLOV
GENTLE WARM (D)
OPEN
SEM 1 (D) LIKSEL (D)
ICLDOM (D) ACCSEL (D)
POISR (D) IMOK (D)
IMOK (D) POISR
MMINT (D) STABLE (D)
SARemot ACCSEL (D)
MMSD (D)
LIKSEL (D)
STRONG
STABLE
ACTIVE
EXTRAV (D) SEM l
EPIEXT (D) MMINT (D)
DOMIN (D) ICLDOM (D)
SAR LIKSEL (D) EXTRAV (D)
8°C DOMIN (D)
ACTIVE
EPIEXT
STRONG
AIN MMTENS (D)
EPINEU (D) EPINEU (D)
MMTENS (D) MMBODY (D)
MAL MMBODY (D) MMSD (D)
MMSD (D) MMSUSP
NIC
MMSUSP
Other OPEN (D) AIN (D)
Clusters SHOFEE (D) NIC (D)
POIIN (D) POIIN (D)

 

 

89

 

 

 

 

 

 

 

 

TABLE 15. COMP analysis for non-specific groups - cosines among
dimensions
Group 1, N = 110
ARO SAR SAR MAL TR
emot soc
ARO .68 .34 .36 -.31 .37
Group 2
SAR .34 .80 .60 -.55 .57
N = 72 emot
SAR .29 .61 .77 -.43 .58
soc
MAL .27 -.57 -.42 .83 -.44
TR .45 .50 .55 -.32 .62

 

 

90

 

 

 

 

 

 

 

TABLE 16. V-analysis performed on preset clusters as part of COMP
analysis — random samples, N-llO, N = 72
Cluster
Oblique £2 Total scores
'Clusters Definers on cluster commun. (h ) on D's
C1
SAR - Group 1 D .84 .72
emot
SAR - Group 2 D .81 .66
emot
MAL - Group 1 D -.80 .81
MAL - Group 2 D -.77 .75
SAR - Group 1 .76 .80
soc
SAR - Group 2 .71 .67
soc
TR - Group 1 .70 .57
Cluster reliability .92
Domain validity .96
C2
ARO - Group 2 .88 .79
ARO - Group 1 .74 .57
TR - Group 2 .66 .53
Cluster reliability .82
Domain validity .91
C3
MAL - Group 1 D .46 .81
MAL — Group 2 D .44 .75
SAR - Group 1 D .43 .80
soc
SAR - Group 2 D .35 .67
soc
Cluster reliability .74
Domain validity .86

 

 

91
groups. The loadings on this cluster were comparatively low, as
was its reliability of .74. Its domain validity was .86. SPAN
diagram 7 presents these relationships geometrically. This diagram
is confusing because of the loose relationships in cluster 3. Once
again, the reader should note the close relationships for all but
the Transparency cluster between comparable dimensions on the two

groups (notated by ++).

92

 

 

 

 

 

 

 

 

 

 

 

 

\
\
/ / TR1 TR2 \
\
/ \
\
SAR 1 I ,SAR 2 C2 \ x
emOt emot

\ ARC 1/

C \

1 \ / ARO 2
-MAL 1

-MAL 2

SPAN diagram 7 - COMP analysis for nonrspecific groups
(higher order clusters)

DISCUSSION

The Salient Dimensions

 

The data generally supported the contention for two salient
and reliable dimensions representing self—acceptance-rejecrion
(SAR) and other-accepLance-rejection (ARO). More specifically,
data analyses revealed four major clusterings of variables, repre-
senting ARO, maladjustment (MAL), and two SAR clusters, representing
its social and emotional modes. MAL, while highly linked to both
SAR clusters, also appeared to represent a unique variant related
to more severe maladjustment. The findings suggested that SAR and
ARO were nearly sufficient to account for the personality question—
naire domain as represented in this investigation. The clustering

of individual variables is described in Table 18 (page 110).

'Clgsis£_l: _A:ceptange versus Rejection of Others (ARO) ‘“-HAI,

- ._ «.a_.

 

A reliable and valid cluster apparently representing the ARO
dimenSJOn was salient. The dimension contained constructs like
warmth, gentleness, being accepting, and accepting others. It was
also described by Interpersonal Checklist (ICL) statements like:
"can be frank and honest", "helpful", "considerate", etc. Further—
more, it was moderately negatively linked with MMPI items (MMSUSP)

related to a construct of suspicion (oblique_££ = —.4l). This ARO

93

94
cluster accounted for about 25% of the common domain variance. The
domain variance figures reported in the discussion represent the
largest portion of variance attained by the clusters in the different
analyses.

Mathematically and geometrically the cluster was fairly clear.
Its saliency was underscored by the fact that most of the communality
of each variable included in the cluster was accounted for by the
cluster (Tables 5 and 9, pages 49 and 70). SPAN diagrams l, 4, and
5 (pages 52, 73, and 74) also depicted this cluster as fairly tightly
knit. It appeared to occupy a distinct portion of the domain as
represented on a sphere. The cluster was also theoretically sound
in the sense that it was saturated by measures which were clearly
relevant to acceptance and rejection of others.

Employing the individual scales of the semantic differential
instrument along with its SEM 2 summary score may have artifactually
strengthened the ARO cluster in the analyses. Of ARO's six defining
variables, four were individual scales already predictably related
to the summary score SEM 2. Closer examination, however, suggests
that this is a minor point. Not only are the correlations between
the individual scales and SEM 2 not so high (.59 to .69) as to clearly
indicate an artifact, but the ICLLOV and SEM 2 scores, which form the
core of this cluster, contribute most importantly to its validity.
Thus, it appears that the cluster would not be greatly altered by
methodological refinements.

Finally, the scales and items contributing to this ARO cluster

may not have adequately represented the ARC construct, perhaps due

95
to a restrictive selection of ARO related measures. The ICLLOV items
and its low communality (22 = .49) suggested some uniqueness in this
ARO cluster, as did its modest relationships with both POI Nature of
MAN (NC) and Rotter's Trust Scale (TRUST). While the semantic

differential items also had low communalities (h2 .41,

-—WARM =
h2 - 35 h2 = 34 h2 = 38 preset) the relia-
° ’ -ACCOTH ' ’ ° ’ ’

-— ACCEPT ’ —- GENTLE

bility of these one-item scales could not be estimated. Thus, the
unique variance of these items cannot be determined either. In

terms of scale scores, then, ARO was defined only by SEM 2, ICLLOV,
and a limited relationship with MMSUSP, with both of the latter having

low domain communalities. A more satisfactory delineation of ARO

requires further research.

Clusters Representing Acceptance versus Rejection of Self (SAR)
Two distinct, but conceptually sound, SAR clusters were found

which jointly accounted for up to 45% of the communality. Their

saliency in the domain seems undisputed. These clusters were

labelled SAR (SARSOC) and SAR (SAR ), conforming

social emotional ' emot

with Foa's (1961) description of facets. Thus, Foa suggested that
within a facet structure one must take into account that there are
both social and emotional modes of acceptance and rejection. The
SAR clusters identified in this study appeared to embody this
distinction.

SARsoc contained the Extraverted vs. Introverted, and Dominant
vs. Submissive semantic differential items, the reflected MMPI
Introversion scale, and the EPI Extraversion scale. This cluster

was interpreted as representing the individual's perception or

96
definition of his own action or behavior. SARemot’ on the other
hand, consisted of measures more related to the individual's feelings
about and evaluation of himself. These measures, POI Self-Regard,
Accepts self vs. Rejects self, I'm OK vs. I'm not OK, and Likes
self vs. Dislikes self, were conceptually divergent from the
elements of SARSOC.

The SAR clusters, while highly valid and reliable, were some-
what diffuse. This was reflected by lower intercorrelations of their
definers as compared with the ARO and MAL clusters. Also the MMINT
and EPIEXT measures shared significant communality with other
clusters (MMINT with MAL, and EPIEXT with ARO). Additionally, two
important SAR variables, the Interpersonal Checklist Dominance factor
(ICLDOM) and the SAR semantic differential summary score (SEM 1)
were about equally dependent on each SAR cluster. ICLDOM had empiri-

cal oblique fg's of .64 and .66 on SARSOC and SARem , respectively.

0t
The comparable SEM 1 values were .79 and .82. SEM 1's content seemed

slightly biased toward SARem (five items) and less toward SARSOC

0t
(three items). ICL items appeared about equally divided, but were
difficult to apportion between clusters because of that instrument's
complex scoring system.

The nearly equal associations of ICLDOM and SEM l with both SAR
clusters presented a problem in interpretation. The preset cluster
analysis demonstrated that the addition of these two measures to
SARemot strengthened that cluster but weakened SARsoc’ and also

enhanced the intercluster linkage. A comparison of SPAN diagrams

2 and 5 (pages 55 and 74) revealed that the SAR clusters moved closer

97
together on the preset. While this solution also lessened relation—
ships of ARO with the SAR clusters, a solution which strengthened
one SAR cluster at the expense of another seemed arbitrary. The
decision to add ICLDOM and SEM l to SARSOC would likely have produced
the opposite result. For this reason an interpretation which allowed
ICLDOM and SEM 1 to be partially subsumed under each SAR cluster and
which also provided a psychological basis for the relationship of
the two SAR clusters was chosen and is presented in Figure 4. The
two clusters are depicted as representing ends of a self-acceptance-
rejection (SAR) gradient with ICLDOM and SEM l centrally located

between SAR and SAR .
soc emot

The Maladjustment Cluster (MAL)

 

The MAL cluster was an unexpected finding. It was defined by
the MMPI scales for Tension, Body Symptoms, reflected Social Desira-
bility, and EPI Neuroticism. MMPI Suspicion and Introversion were
also highly related (:9 = .50 and .59, respectively, on the empirical
analysis). MAL plainly contained a heavy loading of measures usually
used to assess pathology or maladjustment. The cluster accounted for
up to 28% of the communality in the analyses.

In retrospect, this cluster's appearance seems less surprising.
Since dimensionality is determined not only by the similarity of
relationships with some referents, but by a minimal interdependence
on others, the sheer similarity of the MMPI and EPI pathology items
was strongly suggestive of a common body of referents. Two of these
MMPI scales (MMINT and MMSUSP) appeared to importantly contribute to

the relationships of MAL with SARSOC and ARO. MAL and ARO were only

98

 

 

 

 

PAISR

| STABLE
AgSSEL
I K

VbKSEL

STABLE

 

 

 

 

 

emotional

 

Figure 4. Diagrammatic representation of the SAR gradient.

99

slightly linked (E = -.25, empirical, -.21, preset). MMSUSP, which
loaded —.41 on ARO, and shared equal, but modest, communality with
ARO (.17) and MAL (.17), largely accounted for their linkage. MMINT
partly accounted for the MAL and SARSOC (E = .35, empirical) linkage,
sharing nearly equal communality with each cluster (.28, .34,
respectively). The oblique factor coefficients of MMINT on MAL and
the SAR clusters suggested a strong negative relationship between
SAR and MAL. Table 17, below, abstracted from Table 6, provides a

review of the relevant factor coefficients.

 

 

 

 

 

 

TABLE 17. Oblique factor coefficients relevant to the relationship
of SAR and MAL (empirical analysis)
MAL SARsocial SARemotional
POISR -.57 .49 .68
EPINEU .86 -.3O -.52
MMTENS .89 -.20 -.48
MMINT .59 -.79 -.S6
MMSD -.78 .49 .57
ICLDOM -.46 .64 .65
ACCSEL -.52 .43 .79
SEM 1 -045 .79 .82

 

polar opposite of some portion of the SAR gradient.

Examination of these relationships suggests that MAL may be a

It was earlier

 

100

noted that MAL's interdomain correlations with SAR and SAR
soc emot
were —.35 and -.59, and were largely unchanged by the preset analysis.
Furthermore, on the empirical analysis the negative relationship
between SARemot and MAL (g = -.59) was somewhat stronger than the

linkage of SARSOC and SARem (5 = .49) (Table 7, page 61). SAR

ot emot’

then, appeared strongly negatively linked to MAL. In addition, scru-
tiny of Table 17 disclosed that negative MAL coefficients consistently
accompanied positive coefficients on at least one of the SAR clusters,
and vice versa. SPAN diagram 2 (page 55) facilitates visualization

of the evidence for polar opposition. In this diagram, if both SAR
clusters were grouped closer to ICLDOM and SEM l, and if these
variables were moved to the right (180° from MAL), the polarity case
would be strengthened.

Although it would have facilitated confirmation of the two
salient dimensions hypothesized, the evidence did not fully warrant
viewing MAL as a polar opposite of SAR. Too much variance was left
"unanchored" to support that interpretation. Furthermore, MAL seemed
more highly linked to the emotional mode of SAR than to the social
mode. What can be stated more confidently is that, within this
domain of tests, measures of pathology appear highly biased in the
direction of measuring a rejection of self. The author concluded,
then, that an MAL cluster did exist; it was highly negatively
dependent on the SAR dimension, but in addition it did suggest a
real dimension of maladjustment, not satisfactorily accounted for by

the present SAR and ARO dimensions.

101

A Minor Cluster: Transparency (TR)

 

Because it accounted for only 6% of the communality on both the
empirical and preset analyses, this doublet cluster of Open vs.
Guarded (OPEN) and Shows feelings vs. Hides feelings (SHOFEE) appeared
peripheral to the search for major dimensions. The cluster also con-
tained POIIN, a measure of independence and self-supportiveness,
tangential to the domain. The cluster was retained as much for its
possible importance as a behavioral construct as for mathematical
reasons. Openness and the showing of one's feelings were viewed as
highly important behaviors in the development of interpersonal
competence.

While both variables were expected to cluster with SAR (Table
18, page 110), they showed ties to both SAR and ARO. The relation—
ship between Transparency and both SAR clusters, however, was stronger
than its correlation with ARO (Table 7, page 61). A review of
partial communalities and oblique factor coefficients (Tables 5 and
6, pages 49 and 51) also suggested that Transparency was related to
both dimensions. OPEN largely accounted for Transparency's link with
ARO, sharing a sizable portion of its communality (.15) with that
cluster. Openness and showing feelings as a Transparency cluster,
then, were viewed as related to both SAR and ARO, accounting for some

of the latter's interdependency.

Other Groupings

 

Two other groupings of variables reported from the principal
axis analysis deserve note: POINC and TRUST, and ACTIVE, WARM and

STRONG. The factors which loaded on either POINC and TRUST, namely,

102

factors 10 (unrotated, Appendix K), and factor 9 (rotated, Appendix
L), strongly supported confirmation of the originally expected
relationship between these variables and ARO. Factor 9 demonstrated
the shared loadings of these variables with ARO variables. ‘Factor

9 also loaded -.32 on MMSUSP, the one MAL variable related to ARO.
SPAN diagram 1 (page 52) illustrated this relationship. POINC and
TRUST lie off the sphere but in the general area of ARO. In terms
of content, the POINC and TRUST items appeared to deal with a kind
of social-political orientation toward others. The temptation to
view this factor as an important adjunct of ARO was counteracted by
the low correlation between POINC and TRUST (E = .14).

Another grouping of variables that appeared to overlap SAR

and ARO showed up as factor 7 (rotated, Appendix L). This factor
loaded heavily on ACTIVE, WARM, and STRONG, all semantic differential
items. On the BCTRY, ACTIVE was generally part of SARSOC while
STRONG fell toward the center of the SAR gradient. WARM was a
definer of ARO. However, these variables all had low communalities:

2 2 2

h 20, h .25, pWARM =

._ ACTIVE = - __ STRONG = .36. Since it is impossible

to predict the unique variance represented by these one-item scales,
it is also impossible to determine whether these variables account
for a dimension linking SAR and ARO. In addition, because this
factor accounted for less than 5% of the variance, it was clearly

not a salient dimension.

Relationship of the Dimensions

 

SARSoc and SARem were only mdldly associated with ARO, thereby

0t
confirming a second hypothesis (page 44). The most direct estimates

103

of this relationship were the correlations between oblique cluster
domains (Table 7, page 61). The correlations between ARO and SARSOC
were .34 (empirical) and .29 (preset). Thus, not more than 11.5%

of the cluster's variances were shared between them. Likewise, for

ARO and SARem , quite comparable correlations, r = .39 (empirical)

0t
and .30 (preset) accounted for not more than 15% of their covariances.
In addition, generally for all the analyses, those variables that
loaded highly on ARO loaded very minimally on the SAR clusters, and
vice versa (Tables 6 and 10 and Appendices K and L). SPAN diagram

5 (page 74) further supports this view. If each cluster were right

at the ends of the dimension defining axes, then all clusters would

be completely independent. As the clusters move away from these
points they become related. It can be seen that the ARC cluster was
centered on the dimension definer Z. The emotional SAR cluster
centered to the left of the dimension definer X, along the are from

Z to X. The distance between the clusters represents their relation-
ship. The same view can be used to see that SARsoc and ARO were not
precisely orthogonal, and therefore slightly interdependent. This
minimal interdependence of SAR and ARO conflicts with the theoretical

position of Rogers, cited earlier (page 17) and has serious implica-

tions for personality theory and psychotherapy (to be discussed later).

Validity of the Item Sample

 

The validity issue concerns whether the fallible, real scores on
the measures are comparable to "true" scores derived from an infinitely

large battery of personality measures. The domain validity coefficient,

104
separately determined for each cluster (Tables 5 and 9, pages 49 and
70), estimates this relationship from knowledge of the intercorrela-
tions of the clusters. Domain validity coefficients for the four
major clusters ranged from .91 to .95 (empirical), very close to the
preset findings (.92 to .96). These high coefficients provided ample
evidence for validity of the test sample to represent the domain,
and thus for accepting the third hypothesis that the item sample in
this study has a high validity coefficient with respect to the per-
sonality questionnaire domain.

Additional validational evidence was provided by the non-
specific groups cross-validation study (page 87). That study deter-
mined the relationship (cosine 6) between groups for each cluster.
The cosine values for SARSOC (.77), SARemot (.80), and MAL (.83) were
quite high. For ARO the cosine value was slightly lower (.68), but
still adequate. These values demonstrate substantial similarity of
clusters across groups which further supports the view that the
clusters represented the relevant domain adequately.

A serious concern.was whether the experimenter's choice of
instruments unduly influenced selection of clusters which confirmed
the hypotheses. Tryon and Bailey (1970) noted that the clusterings
sometimes result from

"forces in the thinking of the investigator that cause

him to measure some attributes and not others, or social

and biological forces currently at work generating

Special correlations between some attributes...in a

particular sample."

The next step, then, must be an expansion of the number and variety

of measures in studies similar to this one to further test the

saliency of SAR and ARO (see Implications, page 114).

 

105

Eysenck's "Super Factors"

 

Controversy over whether Eysenck's Extraversion-Introversion
factor (EPIEXT) as measured by the Eysenck Personality Inventory
(EPI) would cluster more closely with SAR or ARO was described
earlier (page 8). Eysenck's description of extraversion clearly
contained the suggestion of its relation to ARO, while conceptually
it seemed more sensible to put it with SAR. In the clustering
process EPIEXT had a domain communality of only .47, making its
clear interpretation difficult. Its communality was unequally
divided between SARSOC (.30) and ARO (.15), while its oblique factor
coefficient (empirical) was .64 with SARSOC and .38 with ARO. Thus,
EPIEXT clustered mainly with SARsoc’ but also maintained some tie
to ARO. It seemed reasonable to conclude that EPIEXT was primarily
a measure of SAR, permitting acceptance of the initial part of
hypothesis 4 (page 44).

The second part of the hypothesis, that Eysenck's neuroticismr
stability (EPINEU) will cluster with ARO, was rejected, as EPINEU
had practically no relationship to ARO (£9 = -.l6)° It was, however,
a major definer of MAL (£9 = .86) and highly negatively related to
SARSOC (£3 = -.52). Thus, it appeared that the Eysenck Personality
Inventory (EPI) was heavily biased toward measuring both maladjust-

ment and normality related to acceptance and rejection of self, and

had only minimal linkage with the acceptance and rejection of others.

106

Response Set Measures

 

Two measures of response set, AIN (the average intensity level
of the items checked) and NIC (number of items checked), both taken
from the ICL, clustered together and were minimally dependent on
other clusters and variables. Although highly intercorrelated
(r = 067), they were also largely independent of ICLDOM and
ICLLOV (5's of —.25, .Ol, .07, and .17), as suggested by LaForge
(1973). If NIC and AIN were actually measuring response sets, then
ICLLOV and ICLDOM appeared largely free of these. In addition,
for this measure at least, the two response sets, social desirability
and acquiescence, were highly related. Were this conclusion
generalizable, it would contradict the generally accepted implication
that these are relatively independent response sets (Kassebaum at
al., 1959). However, there is strong evidence that this conclusion
may not generalize to other data, as AIN correlated only -.37 with
the MMPI social desirability measure. Apparently these separate
measures of "social desirability" were not assessing the same thing.
The "social desirability" response set needs further exploration.

When the NIC-AIN cluster was dropped from the preset analyses,
the two variables linked with maladjustment, as did MMSD, as one of
its reflected definers. In addition, MMSD loaded highly on both
SAR clusters but low on ARO (£3 = .28, empirical, .23, preset). These
data support the generally accepted contention that, as scores on
maladjustment measures go up, scores on social desirability measures

go down, and vice versa. In addition, the data were confirmatory

107
to Wiggins (1968) and Crowne et al. (Hurley, 1972) in that social
desirability was interpreted as a facet of SAR.

The fourth response set measure (EPILIE), measuring a tendency
to lie, fell far outside the domain. It also appeared isolated in
the factor analysis. This made any interpretation relative to its
effect difficult, aside from noting that it neither occupied pivotal
positions nor was its variance explained by the clusters.

Thus, neither AIN, NIC, nor EPILIE occupied pivotal positions
in the major clusters or had their variance explained by these
clusters, confirming hypothesis 5 (page 44) for these measures of
response set. However, this did not hold for "social desirability"
as measured on the MMPI. For MMPI items a tendency to "look good"
moderately influenced scores on the MAL and SAR clusters. While
this overlap could be interpreted as detracting from the earlier
interpretation of SAR, it is equally plausible to view this tendency
as a legitimate characteristic of highly self-accepting individuals.
In this sense it did not seem surprising that a social desirability

measure related to SAR.

Subsample Comparisons

 

The two SAR clusters and MAL appeared essentially identical
across sexes (Table 12, page 83). This was confirmed in SPAN diagram
6, and also in Table 14 (page 88). ARO, while reasonably similar
for men and women (cosine = .68), showed more instability than the
other major clusters. The ARC sex differences were strongly linked

to differences in Transparency (TR). SPAN diagram 6 (page 86)

108

shows that TR clusters for males and females were widely separate,
with TR for females being linked to both ARO and SAR clusters. The
only significant difference here was the relationship of the variable
OPEN to the ARC and SAR clusters. For women, Openness appeared more
indicative of ARO, while for men it was closer to SAR.

A more minor sex difference appeared in the clustering of the
response set variables AIN and NIC. AIN (ICL social desirability)
and NIC (ICL acquiescence) were more highly linked with maladjustment
for men than for women.

For the two non-specific groups, the four major clusters were
essentially identical across groups. Additional evidence was provided
by the higher order V-analysis reported in Table 16 (page 90).
Cluster 1 contained both SAR clusters for both groups, MAL for both
groups, and TR for Group 1. SARSOC for Group 1 was the only non-
definer included in the cluster. This clustering accentuated the
SAR—MAL linkage. Cluster 2 contained the ARC clusters for both groups
and TR for Group 2. The two ARO clusters were strongly linked (SPAN
diagram 7, page 92). The TR clusters were again spread between the
major clusters in much the same way as in the male-female comparison.
Thus, these data essentially confirmed the results of the male-
female analysis, and hypothesis 6, that subsamples of the total

sample reveal identical clusters in the domain.

Measurement of ARO and SAR (Clusteringgof Variables)

 

The adequacy of an item, scale or test to measure a construct

is based on its ability to discriminate between constructs. This was

109
operationalized by choosing variables which had high empirical factor
loadings on one cluster and low loadings on all others (Table 6,
page 51). Highest factor loadings (clustering) for the variables on
each of three analyses, in addition to predicted a priori clustering,
are given in Table 18.

The semantic differential scales were apparently adequate measures
of SAR and ARO. They distinctly differentiated between SAR and ARO,
and between SARSOC and SARemot better than other measures. Those
scales that loaded highly on ARO, GENTLE (vs. harsh), WARM (vs. cold),
ACCEPT (accepting vs. rejecting), and ACCOTH (accepts others vs.
rejects others) all discriminated well. SEM 2, the ARO summary score,
discriminated less adequately between these dimensions. All of these
variables had been predicted to measure ARO. STABLE, also a predicted
measure of ARO, clustered with SARemot'

EXTRAV (extraverted vs. introverted) and DOMIN (dominant vs.
submissive) both discriminated SARSOC from other dimensions. Dominant
was especially unlinked to ARO, and neither R component related

highly to SARem The best SARe measures were LIKSEL (likes

ot' mot

self vs. dislikes self), IMOK (vs. I'm not OK), STABLE (vs. unstable)
and ACCSEL (accepts self vs. rejects self). All except STABLE were
predicted measurescfl SAR. SEM l, the SAR summary score, was only a
fair discriminator of SAR from ARO and contained elements of both SAR
clusters. The remaining predicted semantic differential SAR indi—
cators, STRONG (vs. weak) and ACTIVE (vs. passive), related to SAR
but not to a differentiating degree. Contrary to expectation, SHOFEE

(shows vs. hides feelings) and OPEN (vs. guarded) formed a separate

llO

 

mm. A42 am. Aav he: mm. Ans has new mamas:

on. he: on. max on. as: om< mmsmzz

ow. a4: om. Ans q<z ma. Ans q<=fisau0fieouana seems:
00m 00m 00m

as. m<m ss.: Anv _m<m ma.l Ans mam m<m aszz

«a. m “mumaao o: qumHm

ow. sex om. Ans a<z om. Anv maxi om< smszm
com com 00m

no. m<m so. has m<m so. Ans m<m m<m amemm

we. a oz< amass
u 08w uofiw ”.050

an. m<m «a. Ans m<m we. Ans _m<m m¢m mmHom

es. s om< ozHom

mm. me as. me as. my smashemuaas zHHom

«m. m am. he: mm. Any mu Houmsau o: on

as. m we. 44: mm. Ans mu “mumsau on 2H4

om. om< mo. Ame om< no. omaf ox< >ougoH
u 08 uoam

mm. mam Haw Ans m<m so. m<m m<m zoagoH

mm. arduououv .mM ummoum mm. Hmuaufi mumuowmkm mamz
umfixm .uafium meson meson «you manmaums

 

 

 

 

 

 

 

nwdaumumsao Hmauo<

 

 

 

moanmfium> mo wcfiumumsau Hmduom van umuummxm

.mH mqm<H

 

lll

 

00m 00m 00m L
ma. m<m as. Ans m<m so. Ans m<m m<m szo
mm. oomm<m mm. AQV com com

m<m an. on m<m «4m ><maxm
HOBO OBO OBO
a. as a. so u an .3. as u as .5 ss.:
HOBO HOBO HOBO
mm. m<m ma. Ans mam «m. Anv m<m mam ammuHu
mo.- om< me. Any qm< mm. Aav om< om< 2:2;4
ma. my oh. Ans as ea. Age as m<m zmmo
HOBO HOBO HOBO
ms. m<m mm. m<m ss. m<m m<m uzomsm
ss.- om< am. Ame om< am. Ans om< om< emuou<
00m 00m
ma. a ss. m<m ss. m<m m<m m>Hso<
o~.- om< em. om< as. om< om< mqszmo
ma. me no. Aav ma me. Any my m<m mmmomm
mm.- om< mo.H Ans om< HH.H Ans om< om< N sum
HOBO
an. m<m Ha. Anv m<m mm. as m<m H saw
No.- m mm. om< souofivmuaa: mum
as.. as: ss.- on he: ss.- Amy has “mumsau on nmzzL
mm. Awmumuouv mm. ammoum mm. Hmowuwmafi mvmuuwmxm mamz
omfixm .ucaum Mason Mason «coo oHanum>

 

 

 

 

 

 

 

Awaaumumaau Hmsuo<

 

 

 

 

As.uao0v ma mamas

wwsmaas you can: mpmnasz

112

wmuowvmum mus ua umnu momma :umumaao oz:

.mHOHUMM

.m<m mm wmumaa mdoamamafln m<m soon so swfi: mHHmsvo wcawmoa mOHQMHum>U

.oanwaum>

mo wcfiuwumSHolcoc mucmmmuamu woman xGMHm .mcabaou MMHUm Ga NoumSHo mo nmcfimmv you vaMHm Aavn

.HMumaao uoa vases mHAmHum> «no umnu

.wafiumumSHo mo :oauuuvoum on uammmuaou mmomam xamamw

 

 

 

 

 

 

 

 

 

 

 

ss.- om< am. Anv om< mm. Anv om<. oma msooo<
uosm uoam uoam
ea. mam an. Aav .m<m ma. Ans m<m m<m ammoo<
uoam uoam uoam
cm. Mdm me. m<m Nm. M<m QM< mqm<Hm
mm. Avmuwuouv mm. ummmum mm. Hmofiufiasm muouuonxm mawz
umfixm .ucfium Mason Mason «woo mflnmfiums
nmeuwumsau Hm3u0<

 

As.uaoov ma mamaa

 

 

 

II! 45.!1‘

113
cluster, although for females OPEN related more closely to ARO than
to SAR.

The Interpersonal Checklist also discriminated well between SAR
and ARO, but not between SAR clusters. ICLDOM, a predicted SAR
measure, loaded .64 and .65 on the SAR clusters and only .18 on ARO.
However, ICLDOM also loaded -.46 on MAL. ICLLOV, a predicted ARO
measure, loaded .62 on ARO and only .04 and -.02 on SAR. As general
measures of SAR and ARO, the ICL summary scores appeared sufficient,
although a more detailed examination of the instrument is necessary
to assess its adequacy to account for social and emotional facets
of SAR. As expected, the ICL response style scores (AIN, NIC) did
not cluster with either SAR or ARO.

MAL was not adequately assessed by either semantic differential
scales or ICL measures. The Table 18 data suggested that a non—
semantic scale, like MMPI Body Symptoms (MMBODY), might assess
MAL best. MMBODY loaded .79 on MAL and only mildly on ARO (-.l9),

SAR (--l8) and SARem (- 39), respectively. Of the MMPI scales

SOC 0t

predicted to measure SAR, Introversion (MMINT) was a definer of
SARSOC but Tension (MMTENS) defined MAL. Suspicion (MMSUSP), a
predicted measure of ARO, was closer to MAL, while Social Desirability
(MMSD) also defined MAL. EPI Neuroticism did not link with ARO as
predicted, while EPI Extraversion clustered as expected with SARsoc'
Except for MMBODY, the MMPI and EPI scales were not the best short
measures of their respective dimensions.

Several other variables were predicted to cluster with either

SAR or ARO (Table 18). POISR clustered as expected with SARemot'

114

However, both POINC and Rotter's Trust scale (TRUST) failed to

relate to ARO as anticipated.

Hypothesis 7. The hypothesis (page 44) regarding measurement

 

of the major dimensions was accepted, with some modifications. Either
a series of semantic differential items, as described above, or the
Interpersonal Checklist appeared adequate to measure SAR and ARO,

although ICLDOM also included a modest MAL component.

Implications for Research and Psychotherapy

 

Research

The review of the literature and the present findings revealed
several overlapping areas in need of further research. These
include: (1) selection of measures to represent the domain, (2)
clarification of the dimensions, (3) technique, (4) subjects of

study, and (5) instrumentation.

Selection of measures. It is possible, as in any study which

 

samples a domain, that the sampling was responsible for the results;
i.e., that the major dimensions were artifactual, based solely on

the selection of items. The domain sampling theory of Tryon resulted
in a mathematical index which purports to measure valid sampling of

a domain. The delineation of the domain in terms of content, however,
needs further study. Following further clarification of SAR and

ARO, the shorter measures of these clusters, and of MAL, suggested
above, could be applied, in studies similar to this one, to an

expanding number of measures (Edwards' Personal Preference Scale,

115
Tennessee Self-Concept Scale, Gordon Study of Values, to name a few).
At this stage the importance of choosing measures whose clustering
is difficult to predict needs emphasis. The continued saliency
and reliability of the major dimensions would further confirm their

ubiquity in the personality questionnaire domain.

Clarification. While two facets of SAR were presently identi-

 

fied, ARO appeared as a single dimension. The inclusion of so few
measures apparently relevant to ARO may have been responsible for
this. It remains unclear whether ARO might also have social and
emotional facets, as suggested by Foa (1961). An emotional mode
might include feelings like warmth and love, while a social mode
might include response-related constructs like accepting and affili-
ating, and perhaps related notions of values.

Neither ARO's inner structure nor its boundaries have been
sufficiently charted. Two variables thought likely to relate to ARO
(Rotter's Interpersonal Trust Scale and the POI Nature of Man scale)
fell outside the domain represented by the other variables in this
study, although they related somewhat distantly to the ARC cluster.
The items of these measures seem more related to social and political
aspects of the personality domain. This represents a likely facet
of ARO, rarely tapped by personality inventories. A study employing
value and political preference measures, in addition to a wider
variety of ARC-specific measures, seems likely to further clarify

this important dimension.

116

Technique. The literature reviewed did not support a high
level of correspondence between self-report measures and reports by
others. It is important, therefore, that the SAR and ARO dimensions
be confirmed using techniques involving reporting by others. Because
studies involving objective raters of group behavior have generally
been the province of social psychology, that extensive literature
was not reviewed in this paper. However, research studies concerned
with highly involved others are uncommon, in contrast to arbitrarily
constituted laboratory groups. Earlier cited examples of the former
include the encounter group study of Hurley and Force (1971) and
the study by Borgatta, Cottrell and Mann (1958). Schaefer's (1961)
studies of child and maternal attitudes or behaviors were also
encouraging but too often were heavily biased toward self-reports
of attitudes. Schaefer, too, suggested that more effort be placed
on discovering the major dimensions through techniques beyond self—

reports.

Subjects. A large proportion of the research literature in this

"captive" subjects

area, as well as this study, has been done on
such as students, convicts, mental patients, and military personnel.
This restricted focus raises questions about the generality of the
dimensions. Given that this research grows out of an interactional
theory of personality, more studies of naturally occurring and
functioning interactive groups should be made. Families, work groups,

social and religious groups, political groups, all provide arenas for

the study of both dimensions. Research in psychoterhapy and encounter

117
groups is also useful, as it is there that the masks of everyday life
often seem laid aside, if only for a short time. Ratings of self
and others could be obtained in these groups, and in different situa-
tions for the same subjects, on a variety of behavioral and personality
measures. The clustering technique would be satisfactory to reveal
dimensions of behavior and feeling in these groups. The relevancy
of major dimensions across situations could be explored by comparing
clusters obtained in different natural groups for the same individuals.
While it is unlikely that a given individual would appear the same
across situations, it seems likely that the major dimensions would.
In addition, the natural group setting also provides the opportunity
to study ARO and SAR in relation to situation-specific measures of
creativity and performance. One might then predict, on the basis of

scores on SAR and ARO, what people do best on what kind of tasks.

Instrumentation. Issues of instrumentation relate to all of

 

these noted problems. The author earlier suggested that a semantic
differential measure might be adequate to measure the SAR and ARO
dimensions. In addition to further description and clarification of‘
dimensions, what is necessary to design a comprehensive semantic
differential scale is a thorough study of possible scales relevant

to this type of instrument. One needs, then, to begin with a broad
spectrum of scales and work toward a shorter instrument which amply
Ineasures the dimensions. Translation of the ICL objectives into

a series of semantic scales is one obvious possibility.

118

The ICL as it stands now may also be an adequate measure of
these dimensions. What is necessary is a series of good factor
analytic studies of the items. LaForge (1973) reported on one such
study which involved a factor analysis of the sixteen subdivisions
of the instrument, in addition to the four summary scores. The
factors were promising, but this study represented a small beginning.
It is perplexing that more research studies on the ICL have not
become available.

Shostrom's (1964) Personal Orientation Inventory is also a
promising instrument. Factor analytic studies of the POI have
yielded dimensions similar to SAR and ARO, but may have been greatly
influenced by the overlap of items from scale to scale (Silverstein
and Fisher, 1972). The author understands that the instrument is
currently under revision to eliminate this problem. Given the many
aspects of personality purportedly measured by the P01 (page 30),
it seems likely that factor analytic studies of the revised instru-
ment would reveal SAR and ARO, and that a small number of P01 scales

could be employed as shorter measures of the dimensions.

Psychotherapy_

 

Thoughts about the possible facet structure of the SAR and ARO
dimensions, as suggested by Foa (1961), provide some bridge from a
purely research setting to the clinical field. The author is
interested in how a dissonance between the emotional and social
modes of either dimension may create and maintain anxiety. The
clinician sees evidence, for example, that while an individual may

feel self-rejecting (emotional mode), he may act in ways that convey

 

119

self-acceptance (social mode), or vice versa. Likewise an indi-
vidual may feel accepting of others (emotional mode) while acting
in ways which convey rejection of others (social mode), or vice
versa. Such dissonances may contribute significantly to the pro-
duction of anxiety, and their resolution would then be an important
goal of psychotherapy.

Another vital issue for the clinician is whether growth in
therapy is prompted merely by working toward an enhancement of
self-acceptance. The present findings strongly suggest a negative
answer. Additionally, the Hurley-Force study (1971) supported this
view. Therapists, then, must be aware of the dimension of acceptance-
rejection of others as a separate function. Therapeutic methods
which facilitate the growth of acceptance of others as well as
acceptance of self need attention. This task may become easier

when SAR and ARO are better defined.

Summary_and Conclusions

A review of personality research literature suggested that
personality could be adequately described by a small number of
dimensions. The author conducted the present study to assess
evidence for a two-dimensional theory based on constructs of
acceptance and rejection of self (SAR) and acceptance and rejection
of others (ARO). The SAR and ARO dimensions were expected to be
only minimally interdependent.

The present study was limited to self-report, paper-and-pencil

‘measures of personality. A mixed battery of instruments which

120

included the Eysenck Personality Inventory (EPI), the Interpersonal
Checklist (ICL), a set of semantic differential scales, the Rotter
Interpersonal Trust Scale, and selected scales from the MMPI and
Personal Orientation Inventory (POI) was administered to 182 college
students in elementary psychology courses. The data were analyzed
by the BCTRY key-cluster method of cluster analysis, and by a non-
communality principal axis factor analysis. The BCTRY method employs
communality as a measure of the sufficiency of dimensions to account
for the domain in question.

The findings generally supported the expectation of two major
and minimally interdependent dimensions of acceptance and rejection
of self and acceptance and rejection of others. More specifically,
the cluster and factor analyses revealed four major clusters and two
or three minor ones. Most salient of these was an ARO cluster which
was described by behaviors of warmth, gentleness, and acceptance of
others, and by adjectives related to the affection-hostility, or
love-hate, dimension of the Interpersonal Checklist. SAR was
described by two major clusters, representing "social" and "emotional"
modes of self-acceptance (SARSOC and SARemot’ respectively). The
social mode was defined by measures representing extraversion and
assertiveness, while the emotional mode was defined by measures
linked with feelings about oneself. The relationship between the
two SAR clusters was depicted as a gradient with SARsoc and SARemot
at its ends and two broad measuresof SAR centrally located between
the two specific clusters. Another major cluster, MAL, apparently

measuring more severe maladjustment than accounted for by SAR and

121
ARO, was at least partly representative of the rejection pole of
SAR. It was suggested that popular measures of maladjustment appear
heavily biased toward SAR. Interpretations of the minor clusters
were offered in the text.

Measures of extraversion—introversion and anxiety, as well as
a measure of a social desirability response set, tended to relate
highly to SAR and MAL, but not to ARO. The data suggested that a
tendency to "look good" might be part of the SAR dimension. Males
and females were nearly identical on all dimensions, except that an
openness measure was related to ARO for females but to SAR for males.
It was suggested that a set of semantic differential scales could
adequately measure SAR and ARO.

The data supported the main allegations of the present study
for college students on self-report, paper-and-pencil measures. The
continual recurrence of dimensions similar to SAR and ARO in studies
different from the present one provides the best evidence for
generalization of the findings. The author suggested that studies
employing an expanding number and variety of personality questionnaires
was an important next step in further clarifying and generalizing

the SAR and ARO dimensions.

APPENDICES

APPENDIX A

AN ARBITRARY CLASSIFICATION OF MMPI ITEMS BY CONTENT

 

 

Number of
Category Content area items
1 General health 9
2 General neurologic symptoms 19
3 Cranial nerves 11
4 Motility and coordination 6
5 Sensibility 5
6 Vasomotor, trophic, speech, secretory problems 10
7 Cardiorespiratory system 5
8 Gastrointestinal system 11
9 Genitourinary system 5
10 Habits 19
11 Family and marital relations 26
12 Occupational problems 18
13 Educational problems 12
14 Sexual attitudes 16
15 Religious attitudes 19
16 Political attitudes — law and order 46
17 Social attitudes 72
18 Affect, depressive 32
19 Affect, manic 24
20 Obsessive and compulsive states 15
21 Delusions, hallucinations, illusions, ideas
of reference 31
22 Phobias 29
23 Sadistic, masochistic trends 7
24 Morale 33
25 Items primarily related to masculinity-
femininity 55
26 Items to indicate whether the individual is
trying to place himself in an improbably
acceptable light 15

 

(Dahlstrom and Welsh, 1960)

122

U

m
U

HWUJP-BWUJU'JW
GDVCth-l-‘wNH

10.
11.
12.
l3.
14.
15.

WWHUUHWW

l6.
l7.

Hm

SD 18.
B 19.

I 20.

APPENDIX B

*
MMPI ITEMS

For the following questions use the #6 answer sheet.
If the question is TRUE or MOSTLY TRUE as it applies
to you, then blacken the space on your answer sheet
marked "T". If a statement is FALSE or NOT USUALLY
TRUE as applies to you, blacken the space marked
"F". Please be careful to answer next to the cor-
rect number.

I have a good appetite.

I wake up fresh and rested most mornings.

My hands and feet are usually warm enough.

There seems to be a lump in my throat much of the time.
I work under a great deal of tension.

I have diarrhea once a month or more.

I am very seldom troubled by constipation.

At times I have fits of laughing and crying that I
cannOt control.

I am troubled by attacks of nausea and vomiting.

I am bothered by acid stomach several times a week.

I find it hard to keep my mind on a task or job.

I seldom worry about my health.

My sleep is fitful and disturbed.

Much of the time my head seems to hurt all over.

Once a week or oftener I feel suddenly hot all over,
without apparent cause.

I am in just as good physical health as most of my friends.
I prefer to pass by school friends, or people I know but
have not seen for a long time, unless they speak to me
first.

I am liked by most people who know me.

I am almost never bothered by pains over the heart or in
my chest.

I am a good mixer.

 

*
Keyed by scale: B = body symptoms, S = suspicion, I = intro-

version, T

= tension, SD = social desirability.

123

(I)

(0wa

Hmawwwa
U ..
m
U

m
U

H

21.

22.
23.

24.

25.
26.
27.

28.
29.

30.
31.

32.

33.
34.
35.
36.

37.
38.

39.
40.

41.
42.
43.
44.
45.
46.
47.

48.

49.
50.
51.
52.

53.
54.
55.

56.

124

Parts of my body often have feelings like burning,
tingling, crawling, or like "going to sleep."

I hardly ever feel pain in the back of the neck.

I think a great many people exaggerate their misfor-
tunes in order to gain the sympathy and help of others.
I am troubled by discomfort in the pit of my stomach
every few days or oftener.

My feelings are not easily hurt.

I am certainly lacking in self—confidence.

It takes a lot of argument to convince most people of
the truth.

My hardest battles are with myself.

I have little or no trouble with my muscles twitching
or jumping.

I am happy most of the time.

There seems to be a fullness in my head or nose most of
the time.

I frequently find it necessary to stand up for what I
think is right.

Often I feel as if there were a tight band about my head.
I have a great deal of stomach trouble.

I do not worry about catching diseases.

1 commonly wonder what hidden reason another person may
have for doing something nice for me.

Criticism or scolding hurts me terribly.

Most nights I go to sleep without thoughts or ideas
bothering me.

During the past few years I have been well most of the time.
I have had periods in which I carried on activities
without knowing later what I had been doing.

I cry easily.

I have never felt better in my life than I do now.

The top of my head sometimes feels tender.

I do not tire quickly.

I am afraid when I look down from a high place.

I am not afraid to handle money.

It makes me uncomfortable to put on a stunt at a party
even when others are doing the same sort of things.

I frequently have to fight against showing that I am
bashful.

I seldom or never have dizzy spells.

I find it hard to make talk when I meet new people.

I am afraid of losing my mind.

I frequently notice my hand shakes when I try to do
something.

I feel weak all over much of the time.

I have very few headaches.

Sometimes, when embarrassed, I break out in a sweat which
annoys me greatly.

I have had no difficulty in keeping my balance in walking.

SD

57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.

78.
79.

80.
81.
82.

83.
84.

85.

86.

87.

88.

89.

125

I wish I were not so shy.

I frequently find myself worrying about something.

I hardly ever notice my heart pounding and I am seldom
short of breath.

I have periods of such great restlessness that I

cannot sit long in a chair.

I believe I am no more nervous than most others.

I have few or no pains.

My way of doing things is apt to be misunderstood by
others.

I have reason for feeling jealous of one or more members
of my family.

No one cares much what happens to you.

I sweat very easily even on cool days.

I am entirely self-confident.

It is safer to trust nobody.

When in a group of people I have trouble thinking of

the right things to talk about.

I have often felt that strangers were looking at me
critically.

Most people make friends because friends are likely to
be useful to them.

I am sure I am being talked about.

I am likely not to speak to people until they speak to me.
I am so touchy on some subjects that I can't talk about
them.

In school I found it very hard to talk before the class.
I seem to make friends about as quickly as others do.

I think nearly anyone would tell a lie to keep out of
trouble.

I am more sensitive than most other people.

Most people inwardly dislike putting themselves out to
help other people.

I am easily embarrassed.

I worry over money and business.

I have never been paralyzed or had any unusual weakness
of any of my muscles.

I cannot keep my mind on one thing.

I feel anxiety about something or someone almost all the
time.

I have certainly had more than my share of things to
worry about.

Sometimes I become so excited that I find it hard to
get to sleep.

I tend to be on my guard with people who are somewhat
more friendly than I had expected.

I get anxious and upset when I have to make a short
trip away from home.

I have been afraid of things or people that I knew could
not hurt me.

I: 1.3“

 

 

HHHHI-l

._]

(Dr-J

90.

91.
92.
93.

94.
95.
96.
97.
98.
99.
100.
101.
102.
103.

104 O
105.

106.
107.
108.
109.
110.

111.
112.

113 O

114.
115.
116.
117.
118.

119 O

120.
121 O

126

I have no dread of going into a room by myself where other
people have already gathered and are talking.

Almost every day something happens to frighten me.

I feel uneasy indoors.

I have sometimes stayed away from another person because
I feared doing or saying something that I might regret
afterwards.

I am not unusually self-conscious.

At parties I am more likely to sit by myself or with just
one other person than to join in with the crowd.

People often disappoint me.

I am afraid to be alone in the dark.

I have often felt badly over being misunderstood when
trying to keep someone from making a mistake.

People have often misunderstood my intentions when I was
trying to put them right and be helpful.

I have often met people who were supposed to be experts
who were no better than I.

I am usually calm and not easily upset.

If given the chance I would make a good leader of people.
I have at times had to be rough with people who were rude
or annoying.

I worry quite a bit over possible misfortunes.

PeOple generally demand more respect for their own rights
than they are willing to allow for others.

There are certain people whom I dislike so much that I

am inwardly pleased when they are catching it for some-
thing they have done.

It makes me nervous to have to wait.

I have had periods in which I lost sleep over worry.

I am often inclined to go out of my way to win a point
with someone who has opposed me.

I am bothered by people outside, on streetcars, in stores,
etc., watching me.

I enjoy social gatherings just to be with people.

I am quite often not in on the gossip and talk of the
group I belong to.

I have often found people jealous of my good ideas, just
because they had not thought of them first.

Whenever possible I avoid being in a crowd.

I do not mind meeting strangers.

While in trains, buses, etc., I often talk to strangers.
I dread the thought of an earthquake.

I am afraid of finding myself in a closet or small closed
place.

I must admit that I have at times been.worried beyond
reason over something that really did not matter.

I am a high-strung person.

I have frequently worked under people who seem to have
things arranged so that they get credit for good work but
are able to pass off mistakes onto those under them.

I 122.
I 123.
SD 124.
T 125.
B 126.
I 127.
S 129.

127

I sometimes find it hard to stick up for my rights because
I am so reserved.

In a group of people I would not be embarrassed to be
called upon to start a discussion or give an opinion

about something I know well.

The future seems hopeless to me.

Several times a week I feel as if something dreadful is
about to happen.

I feel tired a good deal of the time.

I like parties and socials.

I sometimes feel that I am about to go to pieces.

A large number of people are guilty of bad sexual conduct.

 

I.l Jill l(.l1' ll’.

 

 

APPENDIX C

EPI ITEMS

The following questions are to be answered on your #5 answer sheet.
For these questions try and decide whether a "Yes" answer or a

"No" answer represents your usual way of acting or feeling. If
your answer is "Yes" blacken the space numbered "1" on your answer
sheet. If your answer is "No" blacken the space numbered "2" on
your answer sheet. Remember:

YES -- 1

NO -- 2

1. Do you often long for excitement?

2. Do you often need understanding friends to cheer you up?

3. Are you usually carefree?

4. Do you find it very hard to take no for an answer?

5. Do you stop and think things over before doing anything?

6. If you say you will do something, do you always keep your
promise, no matter how inconvenient it might be to do so?

7. Does your mood often go up and down?

8. Do you generally do and say things quickly without stopping
to think?
9. Do you ever feel "just miserable" for no good reason?
10. Would you do almost anything for a dare?
11. Do you suddenly feel shy when you want to talk to an
attractive stranger?
12. Once in a while do you lose your temper and get angry?
13. Do you often do things on the spur of the moment?
14. Do you often worry about things you should not have done or
said?
15. Generally do you prefer reading to meeting people?
16. Are your feelings rather easily hurt?
17. Do you like going out a lot?
18. Do you occasionally have thoughts and ideas that you would not
like other people to know about?
19. Are you sometimes bubbling over with energy and sometimes
very sluggish?
20. Do you prefer to have few but special friends?

128

21.
22.
23.
24.
25.

26.
27.
28.

29.
30.
31.
32.

33.
34.

35.
36.

37.
38.
39.
40.
41.
42.
43.
44.

45.
46.

47.
48.

49.
50.
51.
52.
53.
54.
55.
56.
57.

129

Do you daydream a lot?

When people shout at you, do you shout back?

Are you often troubled about feelings of guilt?

Are all your habits good and desirable ones?

Can you usually let yourself go and enjoy yourself a lot at

a gay party?

Would you call yourself tense or "highly-strung"?

Do other people think of you as being very lively?

After you have done something important, do you often come
away feeling you could have done better?

Are you mostly quiet when you are with other people?

Do you sometimes gossip?

Do ideas run through your head so that you cannot sleep?

If there is something you want to know about, would you rather
look it up in a book than talk to someone about it?

Do you get palpitations or thumping in your heart?

Do you like the kind of work that you need to pay close
attention to?

Do you get attacks of shaking or trembling?

Would you always declare everything at customs, even if you
knew that you could never be found out?

Do you hate being with a crowd who play jokes on one another?
Are you an irritable person?

Do you like doing things in which you have to act quickly?

Do you worry about awful things that might happen?

Are you slow and unhurried in the way you move?

Have you ever been late for an appointment or work?

Do you have many nightmares?

Do you like talking to people so much that you would never miss
a chance of talking to a stranger?

Are you troubled by aches and pains?

Would you be very unhappy if you could not see lots of people
most of the time?

Would you call yourself a nervous person?

Of all the people you know are there some whom you definitely
do not like?

Would you say you were fairly self-confident?

Are you easily hurt when people find fault with you or your work?
Do you find it hard to really enjoy yourself at a lively party?
Are you troubled with feelings of inferiority?

Can you easily get some life into a rather dull party?

Do you sometimes talk about things you know nothing about?

Do you worry about your health?

Do you like playing pranks on others?

Do you suffer from sleeplessness?

 

APPENDIX D

ICL ITEMS

The following questions are to be answered on your #1 answer sheet.

Following is a list of words and phrases which describe ways people
may behave in relation to one another. Go through the list. If an
item describes YOU, place a mark in column 1 of your answer sheet.

If an item does not describe you, leave the space on the answer sheet
blank. Be sure to mark the right numbered item.

 

l. Able to give orders 31. Dependent
2. Appreciative 32. Cruel and unkind
3. Apologetic 33. Dictatorial
4. Able to take care of self 34. Distrusts everybody
5. Accepts advice readily 35. Dominating
6. Able to doubt others 36. Easily embarrassed
7. Affectionate and understanding 37. Eager to get along with
8. Acts important others
9. Able to criticize self 38. Easily fooled
10. Admires and imitates others 39. Egotistical and conceited
11. Agrees with everyone 40. Easily led
12. Always ashamed of self 41. Encouraging others
13. Very anxious to be approved of 42, Enjoys taking care of others
14. Always giving advice 43. Expects everyone to admire
15. Bitter him
16. Bighearted and unselfish 44. Faithful follower
l7. Boastful 45. Frequently disappointed
18. Businesslike 46. Firm but just
19. Bossy 47. Fond of everyone
20. Can be frank and honest 48. Forceful
21. Clinging vine 49. Friendly
22. Can be strict if necessary 50. Forgives anything
23. Considerate 51. Frequently angry
24. Cold and unfeeling 52. Friendly all the time
25. Can complain if necessary 53. Generous to a fault
26. Cooperative 54. Gives freely of self
27. Complaining 55. Good leader
28. Can be indifferent to others 56. Grateful
29. Critical of others 57. Hard-boiled when necessary
30. Can be obedient 58. Helpful

130

 

LI

 

59.
60.
61.
62.
63.

64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.

131

Hard-hearted

Hard to convince
Hot-tempered

Hard to impress

Impatient with others'
mistakes

Independent

Irritable

Jealous

Kind and reassuring

Likes responsibility
Lakes self-confidence
Likes to compete with others
Lets others make decisions
Likes everybody

Likes to be taken care of
Loves everyone

Makes a good impression
Manages others

Meek

Modest

Hardly ever talks back
Often admired

Obeys too willingly
Often gloomy

Outspoken

Overprotective of others
Often unfriendly
Oversympathetic

Often helped by others
Passive and unaggressive
Proud and self—satisfied
Always pleasant and agreeable
Resentful

Respected by others
Rebels against everything
Resents being bossed
Self-reliant and assertive
Sarcastic

Self-punishing
Self-confident
Self-seeking

Shrewd and calculating
Self-respecting

Shy

Sincere and devoted to friends
Selfish

Skeptical '

Sociable and neighborly
Slow to forgive a wrong
Somewhat snobbish

109.
110.
111.
112.

113.

114.
115.

116.
117.
118.
119.
120.
121.

122.
123.
124.
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.

Spineless

Stern but fair

Spoils people with kindness
Straightforward and direct
Stubborn

Suspicious

Too easily influenced by
friends

Thinks only of self

Tender and soft-hearted
Timid

Too lenient with others
Touchy and easily hurt

Too willing to give to
others

Tries to be too successful
Trusting and eager to please
Tries to comfort everyone
Usually gives in

Very respectful to authority
Wants everyone's love

Well thought of

Wants to be led

Will confide in anyone
Warm

Wants everyone to like him
Will believe anyone
Well-behaved

 

APPENDIX E

*
POI ITEMS

This inventory consists of pairs of statements. Read each statement
and decide which of the two paired statements most consistently

applies to you.
first statement of a pair is TRUE or MOSTLY TRUE as applied to you,
blacken between the lines marked "A" on your answer sheet. If the
second answer is TRUE or MOSTLY TRUE as applied to you, blacken the

space marked "B”.

Mark your answers on the #2 answer sheet. If the

am afraid to be myself.
am not afraid to be myself.

sometimes feel embarrassed by compliments.
am not embarrassed by compliments.

do what others expect of me.
feel free to not do what others expect of me.

4. A. It is possible to live life in terms of what I want
to do.
B. It is not possible to live life in terms of what I
want to do.

can cope with the ups and downs of life.
cannot cope with the ups and downs of life.

believe the pursuit of self—interest is opposed to

interest in others.

believe the pursuit of self-interest is not opposed

to interest in others.

live in terms of my wants, likes, dislikes and values.
do not live in terms of my wants, likes, dislikes

and values.

 

*
Keyed by scale: SR = self-regard, NC = nature of man,

SR 1 A. I
B. I
SR 2 A. I
B. I
IN 3. A. I
B. I
SR
SR 5. A. I
B. I
NC 6. A. I
B. I
SR 7. A. I
B. I
IN = inner-directedness.

132

SR 8.
SR,NC 9.
IN 10.
NC 11.
IN 12.
IN 13.
SR 14.
SR 15.
IN 16.
SR 17.
IN 18.
NC 19.
SR,NC 20.

133

I trust my ability to size up a situation.
I do not trust my ability to size up a situation.

I believe I have an innate capacity to cope with life.
I do not believe I have an innate capacity to cope
with life.

I must justify my actions in the pursuit of my own
interests.

I need not justify my actions in the pursuit of my
own interests.

I believe that man is essentially good and can be
trusted.

I believe that man is essentially evil and cannot
be trusted.

I live by the rules and standards of society.
I do not always need to live by the rules and
standards of society.

Reasons are needed to justify my feelings.
Reasons are not needed to justify my feelings.

I often feel it necessary to defend my past actions.
I do not feel it necessary to defend my past actions.

It is important that others accept my point of view.
It is not necessary for others to accept my point
of view.

I only feel free to express warm feelings to my
friends.

I feel free to express both warm and hostile feelings
to my friends.

I feel free to be myself and bear the consequences.
I do not feel free to be myself and bear the
consequences.

I will continue to grow only by setting my sights on
a high—level, socially approved goal.
1 will continue to grow best by being myself.

Man is naturally cooperative.
Man is naturally antagonistic.

Self-interest is natural.
Self-interest is unnatural.

NC

IN

NC

NC

NC

NC

SR

NC

SR

NC

NC

SR

SR

NC

NC

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

134

I like only masculine men and feminine women.
I like men and women who show masculinity as well
as femininity.

People should always control their anger.
People should express honestly-felt anger.

The truly spiritual man is sometimes sensual.
The truly spiritual man is never sensual.

I have a problem in fusing sex and love.
I have no problem in fusing sex and love.

Evil is the result of frustration in trying to be good.
Evil is an intrinsic part of human nature which
fights good.

A person can completely change his essential nature.
A person can never change his essential nature.

I am assertive and affirming.
I am not assertive and affirming.

Women should be trusting and yielding.
Women should not be trusting and yielding.

It is a good idea to think about your greatest
potential.

A person who thinks about his greatest potential
gets conceited.

Men should be assertive and affirming.
Men should not be assertive and affirming.

Men and women must be both yielding and assertive.
Men and women must not be both yielding and assertive.

I am self-sufficient.
I am not self-sufficient.

I feel certain and secure in my relationships with
others.

I feel uncertain and insecure in my relationships
with others.

People have an instinct for evil.
People do not have an instinct for evil.

People are both good and evil.
People are not both good and evil.

 

NC

SR

36. A.
B.
37. A.
B.

135

People are basically good.
People are not basically good.

I can feel comfortable with less than a perfect
performance

I feel uncomfortable with anything less than a
perfect performance.

 

APPENDIX F

ROTTER INTERPERSONAL TRUST SCALE

For the following questions use the #3 answer sheet. These questions
attempt to measure the attitudes and beliefs of people on a variety
of questions. If you strongly agree with an item, fill in the space
numbered one. Mark the space numbered two ifgyou mildly agree with
the item. That is, mark number two if you think the item is generally
more true than untrue, according to your beliefs. Fill in the space
numbered three iijou feel the item is about equally true as untrue.
Fill in the space numbered four iijou mildly disagree with the item.
That is, mark number four if you feel that the item is more untrue
than true. Ifgyou strongly disagree with an item, fill in the space
numbered five.

 

 

 

 

 

 

 

Strongly agree

Mildly agree

Agree and disagree equally
Mildly disagree

Strongly disagree

. O

UIJ-‘UJNH

1. Hypocrisy is on the increase in our society.

2. In dealing with strangers one is better off to be cautious
until they have provided evidence that they are trustworthy.

3. This country has a dark future unless we can attract better
people into politics.

4. Fear of social disgrace or punishment rather than conscience
prevents most people from breaking the law.

5. Parents usually can be relied upon to keep their promises.

6. Using the Honor System of per having a teacher present during
exams would probably result in increased cheating.

7. The United Nations will never be an effective force in keeping
world peace.

8. Most people can be counted on to do what they say they will do.

9. The judiciary is a place where we can all get unbiased treatment.

0. It is safe to believe that in spite of what people say, most
people are primarily interested in their own welfare.

11. The future seems very promising.

12. Most people would be horrified if they knew how much news the
public hears and sees is distorted.

136

 

137

13. Most elected public officials are really sincere in their
campaign promises.

14. Even though we have reports in newspapers, radio and television,
it is hard to get objective accounts of public events.

15. Most experts can be relied upon to tell the truth about the
limits of their knowledge.

16. Most parents can be relied upon to carry out their threats
of punishment.

17. In these competitive times one has to be alert or someone is
likely to take advantage of you.

18. Many major national sport contests are fixed in one way or
another.

19. Most idealists are sincere and usually practice what they preach.

20. Most salesmen are honest in describing their products.

21. Most students in school would BEE cheat even if they were sure
of getting away with it.

22. Most repairmen will not overcharge even if they think you are
ignorant of their specialty.

23. A large share of accident claims filed against insurance
companies are phony.

24. Most people answer public opinion polls honestly.

25. If we really knew what was going on in international politics,
the public would have more reason to be frightened than they
now seem to be.

II. mmcmH " . . . . H
II. Hpo mmmoemm “ . . . . "
II. m>fimmfiauwm " . . . . .
1|.mamm mmqumHn “ . . . . "
.II Bums u . . . . H
II. ommm<au “ . . . . H
.II mcmuqu " . . . . u
I ME»? u o o o o u
.II mafiuummmm " . . . . u
.I. m>Hso< " . . . . "
II. amumm " . . . . "
ul.mquummm mzomm " . . . . “

GOHmmmumaH Mach mucmmmummu

 

moans: .m crow up .Hh\m~\aa .BOHmuo> Hmucmsfiummxm

NMOHZMHVZH AEOmmmmMMHZH gonzaAHnndE

U NHszmm<

ummn scans uawom msu wHopHoam
“mama

O O O O O O O O O O O O
O O

O O O O I O O O O O O O
O O

O O O I O O O O O C O O
O O

O O O O O O O O O O O O
O I

O O O 0 O O O O O O O O
0 O

O O O O O O O O O O O O
O O

O O O I O O O O O O O O
O O

O O O O O O O O O O O O
O 0

I O O O O O O O O O O O
O O

O I O O O O O 0 O O O O
O O

O O O O O O 0 O O O O O
O O

O

aawo
Mo<m wagon
wGHHHouuaoo
mqmm mmMHA
taco
zmmo

use mouse

ozomam
wafiumooo<
m>Hmm<m
mauawu

mquqmmm mmaHmv

 

mmawom wcw3oaaom mnu mo

some do

HO

“ZOHBUDMHmzH

“mama .50»

138

APPENDIX H

SEMANTIC DIFFERENTIAL SCALES

‘* Ax ls
<5“? A)» €35. .5} ‘6,»
'f 4} ‘§%yé$ ‘$$'€§ é§
$ At?) (by $2" 9’ <9

1. Hides feelings Shows feelings
2. Gentle Harsh
3. Passive Active
4. Accepting Rejecting
5. Strong Weak
6. Open Guarded
7. Cold Warm

8. Likes self Dislikes self

9. I'm OK I'm not OK
10. Extraverted Introverted
ll. Dominant Submissive
12. Unstable Stable

13. Accepts self Rejects self

[ed [*4 [rd [#4 [rd [*4 [#4 [rd [*4 [*4 [*4 [ti [*4 [ti
[Nd [kfl [hfl [NJ [“4 [Nfl [“3 [Nfl [“8 [ti [“8 [hi [hi [hi
hi [Ofl [OJ [OJ [OJ [Ofl [OJ [OJ [Oi [Oi [Oi [Oi [Oi [Oi
Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi
Oivfl [Ufl [Ufl [Oi [Oi [Ufl [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi
[ofl [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [Oi [O9 [01,
[\0 [\fl [\0 [\0 [\d [\0 [‘fl [‘0 [\fl [‘4 [‘4 [\fl [‘4 [‘4

l4. Accepts others _‘ Rejects others
fish 03-10

0-8309 Michigan State University Printing Service

139

“1‘...

APPENDIX I

 

RAW CORRELATION MATRIX OF ALL VARIABLES

0CD‘JO‘J.I¢JKIH

<DO‘JO\I.4dth

POISP
POIVC
POIIN
TRUST
FPILIE
FPIEXT
FDINEU
HHYEWS
unsusp
MMBODY
HHINY
MMSD
NIC
AIN
ICLDO"
ICLLOV
SHOFEF
GENYLE
ACTIVE
ACCEPT
SYDOUG
OPEN
HARM
LIKSEL
IMOK
EXYRAV
"0“!“
STABLE
ACCSEL
ACCOTH
SEH1
SE"?
86!

9015"
POINC
POIIN
YRUSY
EPILIE
EPIEXT
EPINEU
HHYENS
MMSUSP
HHRODV
MH1NT
MMSD
NIC
4|~
ICLDOM
ICLLOV
SHorEE
GENYLE
ACYIVE
ACCEPT
SYPONG
nPEN
HAHN
L1KSEL
IHOK
EXYRAV
DOMIN
SYABLE
ACCSEL
ACCOTH
SENL
SENZ
SEN

pOISP
1.000

.10
(.27?)
.115

'.0‘1
.370
.-.52.:
-.IS?
’.171
‘.396
"'605
cfilb
-.cop
c.57fi
.661
'.00“
.506
v.056
.31‘
.13d
.38?
.348
.193
($004
(a 33’
.273
.261
.374
'569;

0112

fiPEBJ
a
.063

-,367
9.212
v.349

- 3.

61:55
.,.72
..2°‘
-.44.
-,2go
'. 12

..°38

2
pOINC

.101
1.000
.009
.137
‘.007
.130
-,n4o
-,115
-.?26
-.171
'.100
.051
.109
‘.011
.076
.137
.161
.165
0.01!
.246
“.066
.052
.130
“.010
'.021

. 1
0.095

90114

.277

.009
1.000
'003,
.0191

0051
‘0103
'0237
'0125
.0170

'02“

'v170

13
MIC

'00,:
010'
'02“
.030

4
TRUSY

.110
i137)
.7v36
1.000
.013
.077
c.125
9105
0.28
-.130
0.110
.244
.030
.0070
.162
.166
I09,
.057
[052
.113
.028
.150
I191
.059
.060
9160
.190
.0048
.177
.192
.103
.198
'11,!

s
EPILIE

'.041
0,007
.9191

.033
10000
.0246
0,177
0.139
0.064
’g103

.073

.009
.4120

in¢4
0,010

.177
9.102
0.019
'0109
ogoo1
'00,!
-;oo:
'QOSO
'gU}!
.,004
'g152
'o090
.a056
'3051
“.007
.Q191
0.098
.0003

as
ICLOON

“E!“
.356
.107
élQ?

0.034
3283

.0996

99000

'9000

o ilg

. .

5-
907

”390

1.000

03140
0395

.8118
,’°9
9102
.Q 7
;3 1
.009
0909
:05,

9
.no.

9 0
.0099

140

'[020
. 014

7
EPINEU

0.046
.0106
'p125
0.177
"066

1 000
<121§>
.1.

 

17
Snorer

.300
.161
.200
.097
'Q102
.199
'Q100
0.100
’Q196
-.170
.9290
.102
0.017
'g215
.266
.099
11000
“.110
.24.
.074

£33
4;
.166
.157
.139
.213
.179
.109
.130

iii

53)
.111
303‘

O
HHIIUI

o."'
.011,
0.237
.000!
0.330

.3 00
<17731
.000
I92,
(769
'1!
(£12213
.297
.310
.0...
-.I21
‘91..
O...’
“0191
.0122
.910.
.9107
.0075
'0'01
.0300
9.009
n.014
'0...
0.3.7
0.107
'0...
'0'0!
0'0.

ll
OINVLI

.g.’.
01:.
0.1 Q
.UOV
'.'1’
010’
.109
0...,
“.10.
.990
0....
.000
'1’.
.105
.001.

0
ungu'r

I01?!
0.226
0.02,
I...’
I 0 4
I:I:0
.‘16M~
.979~'
1.000
03"
.1.2
0.3.5
.336
4259
Ig010
..229
0.1.5
I'160
0.062
0.30,
.024
0.173
'026’
'01}?
0.12.
.015
.103
0.152
Igto7
0.2.7
I O

I. 0

1.
get!!!

031‘
I..‘.
090’
.09!
I01.’
025‘
0.210
Ig0‘1
I006!
I3I97
I.’ 3
03 7
.007

8.
union!

0‘896
0H!!!
'|1'0
190
n‘.3
‘ole

I‘m.
0

p
C
It
C

I CI
r.-(d
CI

.|.¢d

I
r.“
O“.
.‘i

.3
.l
i

C O 0 L: I C C I. I Cl
'Rfibva'75 «an

non-p .. 4..-

a-c.o \n 1...

 

30
occur:

11:.

'2 o
'01!
L113
on:
{21?
099

I I 0. 0
'h
i.
u

‘33
'21.

033

If?

H 0
CI,
IN.
CD.

I
C>
1-
N0

 

1§_
r-o.u
bEéL.

 

OIIVG’UIOONUO‘

d’O‘dO\IA¢flNI‘

POISR
POINC
POIIN
YRUST
EPILIE
EPIEXT
EPINEU
MHYENS
HHSUQP
MMBODY
MMINT
"HSD
NIC
AIN
ICLDOH
tCLLOV
QHOFFE
GENYLE
ACYIVE
ACCEPT
SYRONG
OPEN
HARM
LIKSEL
[HOK
ExTRAV
DON!”
‘TABLE
ACCSEL
ACCOTH
5E"1
SE"?
SE!

#0159
POINC
POIlN
YRUSY
FPILIE
EPIEXT
EPINEU
MMTENS
NHSUSP
MMBODY
Mnlflf
HHSD
NIC
AIN
ICLDOM
'CLLOV
SHOFEE
GENTLE
ACTIVE
ACCEDY
STRONG
OPEN
HARM
LIKSEL
Inox
EXTRAV
DOHIN
SYABLE
ACCSEL
ACCOTH
SEMI
SEHZ
95x

21
srnouo

.502
.0086
.099
.025
-.032
.171
'021‘
'.182
.02‘

 

22
3PEV

.349
.002

.197

#104

.195
0130
.039
0101
'9050
0210
'0076
“.075
'926°
'0109
.0212
.171
.161
'0095

.473
I166

33
SE!

0063
0069
0102
.0151
.0253
0209
.227
0155
’00‘1
009°
“003°
’00?!
008.
'9119
‘00},
.136
0030
025,
0079
011‘
I011
001’
.257
.073
9110
0100
.9060
021‘
.005
.133
0:01
9 0
1.000

.1l71

25

Inc:

(21;)
'00 1
00 4
0065
“000‘
0191
-.353
0.326

0206
01 7
10000
.179
73;;
(tfgl)
'33?
G 2)
a 2

3110

20
EXYIAV

.273
I121
[045
[169
-|152
1506
-.110
o‘o“
'01!
' 015
931)
72
|0¢0
-5160
1915
.093
I218
.126
.336
0136
I297
.296
.107
019‘
7170
1.0
6\
g 5
.262
.165

4000»

'7
1100

27
00"!"

.263
.029
.000,
.140
”.090
.366
.0127
C'OS‘
.105
.0035
0.472
.215
.006
o 17
.569
I. 0,
,179
'0016
.331
.0002
’370
.151
.011
0210
.101\

Sui-3%”

.053
.277

l.
590

0
'0060

[I
SYAGLE

.37!
'0'1.
.113
.0000
o.|56
.190
'0317
.0300
0.102
.080”
0.2.3
.35.
o..°§
'01,!
.079

29
ACCSEL

3’6,
I.010
.154
.177
I.0.1
937‘
(1 45
7339
I020,
0.367
0.0,0
(4531
I0090
H.302
0‘0‘
.037
0130
0056
.101
9199
.310
.275
0152‘

,{6$5)

I
0005

Go
iccotu

S232
520’
066
192
'g007
:25:
.31 9
. 1 7
'0207
'i116
'1230
£206
.100
-;108
.172
3330
'00!
.153
{091
'356
016
'20!
.357
S115
;202
:149
{020
3099
5350
igfllfl
229

f §§§)

APPENDIX.J

CORRELATIONS BETWEEN DEFINING VARIABLES OF
CLUSTERS IN THE EMPIRICAL ANALYSIS

A. First cluster - ARO

 

SEM 2 ACCEPT ACCOTH WARM
SEM 2 .98 .66 .59 .68
ACCEPT .49 .36 .26
ACCOTH .47 .36
WARM .50

B. Second cluster — MAL

EPINEU MMTENS MMBODY MMSD
EPINEU .74 .75 .66 .60
MMTENS .74 .66 .62
MMBODY .63 .56
MMSD .57
C. Third cluster — SAR ,

scalal

DOMIN EXTRAV MMINT EPIEXT
DOMIN .59 .54 .40 . 35
EXTRAV .58 .43 .41
MMINT .24 . 34
EPIEXT .25
D. Fourth cluster - SARemotional

LIKSEL IMOK ACCSEL POISR
LIKSEL .54 .55 .39 .22
IMOK .50 . 34 .23
‘ACCSEL .27 .17
POISR .22

 

Estimated 2? in diagonals.

142

APPENDIX K

PRINCIPAL AXIS MATRIX

 

Factor Factor Factor Factor Factor Factor Factor Factor Factor

 

1 2 3 4 5 6 7 8 9
POISR .77 .15 .02 .09 .00 .11 -.02 .07 .00
POINC .16 -.30 -.ll -.32 -.05 .05 -.44 .22 .16
POIIN .30 .13 -.O9 .04 -.62 .24 -.08 .09 .11
TRUST .23 -.17 -.05 -.33 .24 .08 -.19 -.31 .55
EPILIE -.05 .02 -.37 -.15 .44 .13 .46 -.00 .13
EPIEXT .45 -.26 .37 -.19 -.O7 -.23 -.17 .07 .31
EPINEU -.65 -.27 .50 .14 -.13 .08 -.03 -.09 .11
MMTENS -.59 -.17 .57 .04 -.08 .13 .05 -.20 .13
MMSUSP -.37 .26 .56 .15 .11 .10 .07 .00 .20
MMBODY -.54 -.12 .55 .12 -.O6 .13 .11 -.17 .00
MMINT -.77 -.O6 -.20 .21 .00 .13 .01 -.11 .24
MMSD .73 .12 -.24 -.08 .10 -.21 -.08 .06 .06
NIC -.15 -.44 .48 .06 .36 .08 -.14 .44 .07
AIN -.48 -.29 .28 .03 .47 .07 -.08 .39 .00
ICLDOM .72 .23 .30 -.O8 .23 .08 -.Ol .06 .04
ICLLOV .11 -.70 -.21 -.O9 .10 .04 .34 -.05 .14
SHOFEE .40 .03 .08 -.26 -.35 .42 .09 .33 .11
GENTLE .03 -.69 -.Ol -.02 .02 -.22 -.Ol -.14 .19
ACTIVE .44 .17 .28 -.05 -.13 -.38 .28 .24 .38
ACCEPT .35 -.60 -.05 .02 .09 .14 -.08 .00 .18
STRONG .49 .18 .30 .16 .13 -.ll .31 -.04 .10
OPEN .48 -.18 .10 -.18 -.16 .58 .25 .01 .08
'WARMI .34 -.55 -.05 .01 -.20 -.15 .33 -.00 .26
LIKSEL .65 .12 .05 .43 .19 .23 -.13 -.13 .02
IMOK .70 .07 .04 .42 .20 .14 -.08 -.11 .00
EXTRAV .48 -.06 .48 -.40 -.06 -.18 .01 -.19 .17
DOMIN .42 .21 .54 -.32 .12 -.17 -.03 -.18 .02
STABLE .48 -.00 -.03 .57 .01 -.17 .06 .26 .10
.ACCSEL .72 .03 -.00 .27 .14 .09 -.19 -.27 .00
ACCOTH .38 -.43 -.09 -.05 .02 .17 -.13 -.19 .12

143

 

144

 

Factor Factor Factor Factor Factor Factor Factor Factor Factor

 

1 2 3 4 5 6 7 8 9
SEM l .88 .09 .35 -.O3 -.02 .14 .11 -.02 .07
SEM 2 .55 -.74 -.O8 .22 -.Ol -.09 .07 .00 .04
SEX .09 -.33 .20 .38 -.50 -.18 -.09 -.O3 -.01
H.LOAD .88 -.74 .57 .57 -.62 .58 .46 .44 .55
PROP.VAR .25 .10 .09 .05 .05 .04 .03 .03 .03
CUM.P.V. .25 .36 .45 .51 .56 .60 .64 .67 .70

 

1'

APPENDIX L

VARIMAX ROTATION ANALYSIS - ROTATED FACTOR LOADINGS

 

Factor Factor Factor Factor Factor Factor Factor Factor Factor

 

 

l 2 3 4 5 6 7 8 9
POISR .27 -.Ol -.36 .56 -.O4 .29 .15 -.05 .01
POINC .06 -.16 -.31 -.15 -.26 .18 -.15 .28 .39
POIIN -.Ol .05 -.17 .09 -.38 .53 -.06 -.31 -.16
TRUST .09 -.11 -.05 .09 .10 -.02 .05 -.Ol .82
EPILIE -.15 -.10 -.15 -.05 .73 -.02 -.02 -.02 -.07
EPIEXT .65 -.30 -.12 .07 -.27 .05 -.02 .15 -.10
EPINEU -.13 -.Ol .79 -.26 -.22 -.09 -.09 .20 -.00
MMTENS -.Ol .04 .85 -.23 -.1O -.06 -.02 .12 .05
MMSUSP .14 .36 .55 .02 .06 -.O7 -.06 .19 -.31
MMBODY -.OO 7 .04 .80 -.15 -.05 -.06 -.04 .10 -.09
MMINT -.66 .10 .40 -.3O .04 -.22 -.11 -.00 .08
MMSD .30 -.O8 -.67 .33 .01 .02 .12 -.10 .05
NIC .06 -.18 .29 .00 -.05 .00 .04 .83 .02
AIN -013 -.02 .30 -.17 .14 -.16 -.08 .76 -.O4
ICLDOM .51 .12 -.22 .52 .10 .21 .19 .11 .12
ICLLOV -.O4 -.79 -.02 -.09 .27 .06 -.02 .02 -.01
SHOFEE .13 .01 -.14 .03 -.06 .78 .14 .03 .07
GENTLE .12 -.70 .04 -.11 -.O9 -.21 -.ll .07 .00
ACTIVE .32 .09 -.15 .07 -.11 .09 .77 -.OO .03
ACCEPT .09 -.63 -.11 .21 -.03 .13 -.19 .18 .05
STRONG .32 .02 .02 .44 .13 .02 .45 -.O7 -.07
OPEN .19 -.28 .04 .24 .18 .72 -.03 -.07 .04
‘WARM, .00 -.65 -.O4 -.00 -.07 .14 .44 -.10 .13
LIKSEL .08 -.02 -.14 .85 -.O3 .08 -.02 -.02 .01
IMOK. .11 -.09 -.19 .83 -.02 .04 .06 -.03 .01
EXTRAV .83 -.14 .01 .05 -.05 .09 .07 -.08 .07
DOMIN .77 .16 .05 .17 .02 -.00 .14 -.Ol .15
STABLE -.12 -.13 -.26 .53 -.20 -.O4 .41 .10 -.23
.ACCSEL .21 -.13 -.23 .75 -.08 .00 -.01 -.14 .15
.ACCOTH .05 -.45 -.09 .24 -.O6 .14 -.10 -.02 .36

145

146

 

Factor Factor Factor Factor Factor Factor Factor Factor Factor

 

l 2 3 4 S 6 7 8 9
SEM 1 .55 -.07 -.14 .59 -.Ol .40 .29 -.O6 .10
SEM 2 .03 -.84 -.17 .32 -.17 .04 .18 .08 .07
SEX -.Ol -.32 .19 .08 -.62 .00 .14 -.11 -.19
HI.LOAD .83 -.84 .85 .85 .73 .73 .77 .83 .82
PROP.VAR .10 .10 .11 .12 .04 .06 .04 .05 .04

CUM.P.V. .10 .20 .32 .45 .50 .56 .61 .66 .70

 

 

APPENDIX M

CORRELATIONS BETWEEN DEFINING VARIABLES OF

CLUSTERS IN THE PRESET ANALYSIS

A. First cluster - ARO

SEM 2 ACCEPT
SEM 2 .98 .66
ACCEPT .49
ACCOTH
WARM

B. Second cluster - MAL

EPINEU MMTENS
EPINEU .74 .75
MMTENS .74
MMBODY
MMSD
C. Third cluster - SAR

social

DOMIN EXTRAV
DOMIN .59 .54
EXTRAV .58
IMMINT
EPIEXT
D. Fourth cluster - SARemotional

LIKSEL IMOK
LIKSEL .54 .55
IMDKL .50
ACCSEL
POISR

 

Estimated h? in diagonals.

ACCOTH

.59
.36
.47

MMBODY

.66
.66
.62

MMINT

.40
.43
.24

ACCSEL

.39
.34
.27

WARM

.68
.26
.36
.50

MMSD

.60
.62
.56
.57

EPIEXT

.35
.41
.34
.25

POISR

.22
.23
.17
.22

REFERENCES

 

REFERENCES

Abbott, R. D., Fry, R. M., and Abbott, S. The R scale of the MMPI
as a measure of acquiescence: Replication for non-pathological

content trait adjectives. Psychological Reports, 1972,_§l,
806.

Abbott, R. D., Fry, R. M., and Abbott, S. Tests of an alternative
explanation of data supporting the R scale of the MMPI as a
measure of acquiescence. Psychological Reports, 1973, gg,
1243-1246.

Adams, H. B. Mental illness or interpersonal behavior. American
PBychologist, 1964,_12, 191-197.

Allport, G. W. Patterns and Growth in Personality. New York: Holt,
Rinehart and Winston, 1937, 593.

Armentrout, J. A. Correspondence of the MMPI and minidmult in a
college population. JOurnal of Clinical Psychology, 1970,
.26, 493—495.

Armentrout, J. A., and Rouser, D. L. Utility of the mini-mult with
delinquents. Journal of Consulting and Clinical Psychology,
1970, 33, 450.

Becker, W. C. The matching of behavior rating and questionnaire
personality factors. Psychological Bulletin, 1960,_§Z,
201—212.

Berne, E. Principles of’Gromp Treatment. New York: Grove Press,
1966, 379.

Borgatta, E. F., Cottrell, L. S., and Mann, J. M. The spectrum of
individual interaction characteristics: An interdimensional
analysis. Psychological Reports, 1958, 4, 279-319.

Buros, O. K. (Ed.). Personality Tests and Reviews. New Jersey:
The Gryphon Press, 1970, 1959.

Cattell, R. B. Personality and social Psychology. San Diego: Robert
R. Knapp, 1964, 799.

148

 

 

149

Dahlstrom, W. G., and Welsh, G. S. An MMPI Handbook. Minneapolis:
University of Minnesota Press, 1960, 559.

Dicken, C., and Wiggins, J. S. The social desirability scale is not
a short form of the MMPI: A reply to Edwards and Walker.
Psychological Reports, 1964, 13, 711-714.

Edwards, A. L. The Social Desirability Variable in Personality
Assessment and.Research. New York: Holt, Rinehart and
Winston, 1957, 108.

Edwards, A. L. Social desirability and performance on the MMPI.
Psychometrika, 1964, 23, 295-308.

Edwards, A. L. The Measurement of Personality Traits by scales and
Inventories. New York: Holt, Rinehart and Winston, 1970,
306.

Edwards, A. L., Diers, C. J., and Walter, J. Response sets and
factor loadings on sixty—one personality scales. Journal
of.Applied Psychology, 1962,_4§ (3), 220-225.

Edwards, A. L., Gocka, E., and Holloway, H. The development of an
MMPI acquiescence scale. Journal of Clinical Psychology,
1964, gg, 148-150.

Eysenck, H. J. The Structure of'Human Personality. London: Methuen
and Co., Ltd., 1953, 348.

Eysenck, H. J. Primaries or second-order factors: A critical con-
sideration of Cattell's 16PF battery. Journal of'Social and
Clinical Psychology, 1972, 11, 265—269.

Eysenck, H. J., and Eysenck, S. B. Manual fbr the Eysenck Personality
Inventory. San Diego: Educational and Industrial Testing
Service, 1968.

Foa, U. G. Convergences in the analysis of the structure of inter-
personal behavior. Psychological Review, 1961,.§§, 341-353.

Fox, J. Social desirability and the prediction of MMPI scores.
Journal of Consulting Psychology, 1966, 30, 460.

Freedman, M. B., Leary, T. F., Ossorio, A. G., and Coffey, H. S.
The interpersonal dimension of personality. Journal of
Personality, 1951, 39, 143—161.

Harford, T., Lubetkin, B., and Alpert, G. Comparison of the standard
MMPI and the mini-mult in psychiatric outpatient clinic.
JOurnal of Consulting and Clinical Psychology, in press.

150

Harris, T. A. I'm 0K - You're OK, A Practical Guide to Transactional
Analysis. New York: Harper and Row, 1967, 278.

Hochriech, D. J., and Rotter, J. R. Have college students become
less trusting? Journal of Personality and Social Psychology,
1970:.l2 (3), 211-214.

Hurley, J. R. Two prepotent interpersonal dimensions and the effects
of trainers on groups. Small Group Behavior, in press, a.

Hurley, J. R. "Some effects of trainers on T-groups" reconsidered.
Journal of Applied Behavioral Psychology, in press, b.

Hurley, J. R. Helpful behaviors in groups of mental health profes-
sionals and undergraduates. International Journal of Group
Psychotherapy, in press, c.

Hurley, J. R., and Force, E. A multiobserver multimethod study of
interpersonal competence in groups. Unpublished, Michigan
State University, 1971.

Kassebaum, G. G., Couch, A. 8., and Slater, P. E. The factorial
dimensions of the MMPI. Journal of Consulting Psychology,
1959, 33, 226—236.

Kincannon, J. C. Prediction of the standard MMPI scale scores from
71 items: The mini-mult. Journal of Consulting and
Clinical Psychology, 1968, gg, 319-325.

Knapp, R. Relationship of a measure of self-actualization to neuroti-
cism and extraversion. Journal of Consulting Psychology,
l965,_g9, 168-172.

Kulik, C. C., and Kulik, J. A. The cluster analysis of the MMPI
item pool of the BCTRY system - A critique and reanalysis.
Unpublished manuscript, University of Michigan, 1972.

LaForge, R. Using the ICL. Mill Valley, California: Self-
published, 1973.

LaForge, R., and Suczek, R. F. The interpersonal dimension of
personality 111. An interpersonal checklist. Journal of
Personality, 1955, 34, 94-112.

Leary, T. Interpersonal Diagnosis of'Personality. New York: The
Ronald Press 00., 1957, 518.

Lieberman, M. A., Yalom, I. D., and Miles, M. B. Encounter Groups:
First Facts. Chapter 7: The Leaders. New York: Basic
Books, Inc., 1973, 495.

151

Lorr, M., and Mc'Nair, D. M. An interpersonal behavior circle.
Journal of Abnormal Social Psychology, 1963,_§Z, 68-75.

Lorr, M., and Mc'Nair, D. M. Expansion of the interpersonal behavior
circle. Journal of Personality and Social Psychology, 1965,
_2, 823-830.

Nunnally, J. C. Psychometric Theory. New York: McGraw-Hill, 1967,
640.

Omwake, K. T. The relation between acceptance of self and acceptance
of others shown by three personality inventories. Journal
of Consulting Psychology, 1954,.18, 443-446.

Peterson, D. R. Scope and generality of verbally defined personality
factors. Psychological Review, 1965, 12, 48-59.

Rogers, C. R. Client-Centered Therapy. Boston: Houghton-Mifflin
Company, 1951, 560.

Rotter, J. B. A new scale for measurement of interpersonal trust.
Journal of Personality, 1967,_§5, 651-665.

Schaefer, E. S. A circumplex model for maternal behavior. Journal
of'Abnormal social Behavior, 1959, 226-235.

Schaefer, E. S. Converging conceptual models for maternal behavior
and for child behavior. In Glidewell, J. C. (Ed.), Parental
Attitudes and Child Behavior. New York: C. C. Thomas, 1961.

Sheerer, E. T. An analysis of the relationship between acceptance
of and respect for self and acceptance of and respect for
others in ten counseling cases. JOurnal of Consulting
Psychology, 1949, 13, 169-175.

Shostrom, E. L. An inventory for the measurement of self—actualization.
Educational and Psychological Measurement, 1964, 24, 207—218.

Shostrom, E. L. EITS Manual for the Personal Orientation Inventory.
San Diego: Educational and Industrial Testing Service, 1966.

Shostrom, E. L., and Knapp, R. R. The relationship of a measure of
self-actualization (POI) to a measure of pathology (MMPI) and
to therapeutic growth. American JOurnal of'PSychotherapy,
1966,.29, 193-202.

Silverstein, A. B., and Fisher, G. Item overlap and the "built—in"

factor structure of the personal orientation inventory.
Peychological Reports, 1972, 31, 492-494.

 

‘ -.._—-; ~—

 

 

 

152

Stein, K. B. The TSC scales: The outcome of a cluster analysis of
the 550 MMPI items. In McReynolds, P. (Ed.), Advances in
Psychological Assessment - Volume 1. Palo Alto, California:
Science and Behavior Books, 1968, 336.

Stock, D. An investigation into the interrelations between the
self-concept and feelings directed toward other persons and
groups. Journal of Consulting Psychology, 1949, 13, 176—180.

Streiner, D. L., Woodward, C. W., Goodman, J. T., and Mc'Lean, A.
Comparisons of the MMPI and the mini-mult. Canadian JOurnal
of Behavioral Science, 1973, 5 (1), 76-82.

Thurstone, L. L. Multiple Factor Analysis. Chicago: University
of Chicago Press, 1961, 535.

Tosi, D. J., and Hoffman, S. A factor analysis of the personal
orientation inventory. Journal of Humanistic Psychology,
1972,_12 (1), 86-93.

Tryon, R. C. Unrestricted cluster and factor analysis, with appli-
cations to the MMPI and Holzinger-Harmon problems. Multi-
variate Behavioral Research, 1966, 1, 229-244.

Tryon, R. C., and Bailey, D. E. Cluster Analysis. New York:
McGraw-Hill Book Co., 1970.

Welsh, G. 8. Factor dimensions A and R. In Welsh, G. S., and
Dahlstrom, W. G., Basic Readings on the MMPI in Psychology
and Medicine. Minneapolis: University of Minnesota Press,
1956, 264-281.

Wiggins, J. 8. Personality structure. .Annual Review of.Psychology,
1968, 1_9, 293-350.

 

"lil'flilll’ll'lllfll