Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
A factor-analytic study of the Minnesota multiphasic personality inventory using a transpose matrix (Q-technique)
(USC Thesis Other)
A factor-analytic study of the Minnesota multiphasic personality inventory using a transpose matrix (Q-technique)
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
A FACTOR-ANALYTIC STUDY
O F THE MINNESOTA MULTIPHASIC PERSONALITY INVENTORY
USING A TRANSPOSE MATRIX (Q-TECHNIQUE)
A Dissertation
Presented to
The Faculty of the Psychology Department
The University of Southern California
In Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy
by
Harold orl o
June 1952
P .
Ps
3
This dissertation, 1.vritten by
Harold Borko
under the guidance. of h.!~ ___ Faculty Com1nittee
on Studies, and approved by all its members, has
been presented to and accepted by the Council
on Graduate Study and Research, in partial ful
fillment of requirements for the degree of
DOCTOR OF PHILOSOPHY
Date ........ J µ_n._~ .1 .. .. l 9.5. i. ~ - .......... .
Committee on Studies
I
ACKNOWLEDGMENTS
This study was supported in part by funds pro
vided under contract AF 18 (600) - 76 with the USAF School
of Aviation Medicine, Randolph Field, Texas.
The writer wishes to express hie appreciation to
Dr. J. p. Guilford, who supervised this research for the
School of Aviation Medicine, to Dr. H. s. Grayson, for
1
making available the facilities of the Veterans' Adminis
tration Hospital , and to the Uni ed States Army Clinical
Psychology Trainin Pro ram for he opportuni y to com
nlete the work on this project.
In this expression of his ra 1 ude, h wri er
does not wish to imply hat he sta ements and conclu ions
made herein are endorsed by the s e 1nstitu ion.
TABLE OF CONTENTS
CHAPTER
PAGE
1
l
4
4
I. THE PROBLEM •.••
A. Introduction
• • • • • • • • • • • •
• • • • • • • • • • • •
B.
c.
Statement of purpose
Experimental design
• • • • • • • •
• • • • • • • •
D. Select i on of cases .•.•...•
E. The need for and importance of this
study . . . . . . . . · · • · · • •
7
II. GENERAL I NTRODUCTION TO THE MMPI .....
13
16
A. Description of the inventory and its
development. . . . . . . . . . . • 16
B. Description of the scales. • . . . • 20
1. The clinical scales • • . . • 20
2. The validity scales . . . . . 22
C. Evidence for reliability and validity 24
1. Reliability studies . . • • • 24
2. Validity studies • • • . . . 27
III. BRIEF INTRODUCTION TO THE METHOD OF FACTOR
ANALYSIS AS IT APPLIES TO TIIS INVEST!-
GAT ION • • • • • • • • • • •
• • • • • • • 35
A. Genera 1 orients. t 1 on . . • • . • . . • J 5
B. Review of factorial studies of the MMPI 37
1. Abrams
• • • • • • • • • • •
38
CHAPTER
2. Cook and Wherry
• • • • • • • •
J. Cottle • • • • • • • • • • •
4. Little
• • • • • • • • • • •
5. Stout • • • • • • • • • • •
6. Tyler
• • • • • • • • • • •
Wheeler, Little, and Lehner
• •
8. Winne
c. Summary
• • • • • • • • • • •
• • • • • • • • • • • • •
IV. DESCRIPTION OF THE STATISTICAL PROCEDURES
v.
EMPLOYED
• • • • • • • • • • • • • • •
A. Computation of the tetrachoric corre-
lations
• • • • • • • • • • • • •
B. Use of the Holley adjustment technique
c. Description of Tucker• s machine method
of factor analysis
• •
•
• • • •
•
D. Description of the rotation process
•
INTERPRETATION OF THE FACTORS
..
• • • • • •
A. General introduction • • • • • • •
B. Interpretation of each factor
Factor A. Strength of ego de-
• • •
fenses, bipolar • • • • • •
Factor B. Test-taking attitude,
bipolar • • • • • • • • •
iv
PAGE
39
40
41
42
43
44
45
47
50
50
54
59
67
75
75
79
80
81
CHAPTER
c.
Factor c.
Factor D.
Asocial impulsivity.
Anxiety and tension.
• •
• •
Factor E. Masterful self-assertive-
V
PAGE
83
84
ness versus obsequious
self-disparagement, bipolar 85
Factor F. Hypomania an d dominance. 86
Factor G. Introversion, somatic con
cern, and re gressive be-
havior . . . . . . . • . • 87
Factor H. Narcissism and depend.ency 88
Factor r. Autistic and paranoid ten-
Factor J.
Factor K.
Factor L.
denoies • • • • • • • • •
Depression . .
Not specifiable
• • •
• • •
Emotional reactivity .
• •
• •
• •
Factor M. Level of a spiration, ambi-
89
90
91
92
tion . . . . . . . • . 93
Factor N.
Obse ssional rumination • •
Discussion • • • • • • • • • •
• •
VI.
SUMMARY AND CONCLUSIONS • • • • • • • • • •
94
95
101
101
103
Summary
• •
• • • • • • •
• • • • • •
Conclusions · • • • • • • •
• • • • • •
CHAPTER
BIBLIOGRAPHY • • • • • • • • • • • • • • • • • •
APPENDIX A: THE RELIABILITY AND VALIDITY OF THE
-
• •
FACTOR NAMES • • • • • • • • • • • • • • • • • •
APPENDIX B: A QUALITATIVE STUDY OF THE APPLICA-
vi
PAGE
108
115
BILITY OF THE FACTORIAL TRAITS TO CLINICAL DATA • 125
LIST OF TABLES
TABLE PAGE
I. Reliability Coefficients Reported for the
MMPI • • • • •
• • • • • • • • • • • • • •
26
II. Factors Reported in Two or More of the Factor-
Analytic Studies ~ of the MMPI • • . . . • 48
III. Distribution of the Number of Items Classi-
fied as "Cannot Say" by the Thirty-Five
Subjects . . . . . . . . . . . . . . . • • 51
IV. Computation of the Tetrachorio Correlation
Coefficient from a Four-fold Table • . . . 55
v. Computation of the Adjusted Correlation Coef-
ficients from a Four-fold Table . • . • 57
VI. Distributions of Correlation Coefficients
Computed by bot~ the Original and Modified
Methoda . . •
• • • • • • • • • • • • • • •
VII. Distribution of the Differences between the
Original and Adjusted Correlation Coeffi-
60
ciente ...•...
• • • • • • • •
. . . . 61
VIII. The Correlation Matrix
• • • • • • • •
62-63-64-65
IX. Computation of the Standard Error of a
Zero Tetrachoric Correlation Coefficient
for an N of 373 • • • • • • · · · · • ·
• • 68
TABLE
x. Distribution of the Fourteenth Factor
Residuals (One Diagonal Half of the
Factor Matrix) .•...•..•.
• • •
v111
PAGE
69
XI. Unrotated Factor Loadin e and Communalitiee 70-71
XII. Final Rotated Factor Matrix ...•..•. 73-74
XIII. Summary of Factor Names •......• 96-97
XIV. MMPI Raw Scores Corrected for K . . . . • 117
xv. MMPI Standard Scores Corrected for K
• • •
118
XVI. Comparison of the Factor ames Used by
the Different Judges .......... 122-123
XVII. Rating Scale Showin the Agreement be-
tween the Factor Loadin sand Clinical
Reports for Case No. 10. . . . . . . . • 145
XVIII. Summary of the Rated Agreement between the
Factor Loadings and the Clinical Reports 147
CHAPTER I
THE PROBLEM
A. INTRO DUCTION
Man, from time immemorial , has been concerned with
the tudy of personalitr so that e might better under
stand hi neighbor. Even the pr mitive tribes devised prag
matic tests of personality. The rites of initiation were
such eats, and theoe ceremonie were desi ned to make he
boy rove to the satisfact on of he elders tha he wa
worthy and capable of carr ing on the tribal t radition.
Upon the sa i sfactory com le ion of he e task, which
usually involved tests of stren th, endurance, skill, etc.,
he boy w a s ranted the s atu of anhood.
Today, man 1 still very much in erested in evalu
atin he ersonali y and abilities of his fellow men.
ta ner (60, p . 19-20) points ou hat the modern scien
ific ap roach to the study of ersonality has proceeded
along two more or less discre e lines of development.
There is the cl nical approach hich is concerned ~1th
he a hology of per onali y , the study of abnormal eople
and per onality deviants. Becau e of he neces ity for
e n resul sand of the emp as1 on he hole individual,
theo clinical tudies ere u uall. not ell controlled
2
1
experimentally. The other approach deals with more or le s
I
1
normal people, and the tendency has been to nar ow the prob-
lem and to develop normative, descriptive statistics of the
1Separate aspects of personality. This approach, because of
' its more experimental orientation, has utilized personal-
I
ity tests, statistical norma, and other devices. The re-
,sult has been a split in the field of personality tudy.
'
The Minnesota Multi hasic Personality Inventory
1
is
a device which at empts to utili ze the norma ive a roach
in the study of persona 1 y devian s . A more com lee de
scription of this test will follow in a la er chapter.
2
At this poin it is ufficien t o state th the scale
used to score this te s t are in terms of descriptive psy
chiatric categorie and not psychologically der ed per-
sonality variables . he net re sult is hat one set of vari-
ables is used to descri e the abnor al er onali y and
1
another the normal. I ma be hat t his is a nece ssary
and nroper ap roach, but certainly some wri er of modern
I
,textbook on abnormal sycholo see to ake an o _posite
~ereaf er thi tes will e designated as the r.
2
In orde o fully under tand the roblem w th which
this research is concer ned, consid rable fa iliari y with
the MMPI is presup osed. Should the reader be alto ether
unacquainted with he te , it i u gested ha he read
Chapter II before continuin.
3
view. For example, Cameron (8, p. 141) states:
But the adjustive techniques we find in behavior
pathology are not unique for every individual, and
any of the abnormal reactions can be derived from some
basic adjustive technique that normal children and
adults use in everyday life. Indeed, we shall use
this well-founded relationship as a bridge over which
we may pass, in our analysis of behavior, from normal
everyday conduct to even the most florid and the most
bizarre behavior pathology.
Coleman (14, p. 56), in his listing of fundamental
points of agreement which underlie contemporary psychi
atric and psychological thought, includes the following
statement: "······agreement th t the same fundamental
dynamic principles underlie both normal and abnormal be
havior."
Since these authors, and m any others, clai that
both normal and abnormal individuals use the same adjustive
techniques, it follows that one ought to be able to de
scribe the personalities of both roups in terms of a
single set of basic personality traits .
Turning back to the starting point of this discus
sion, it will be recalled that the 1i1:MPI is scored in terms
of psychiatric diagnostic categories. The uestion arises
as to whether this test can be re-evaluated in terms of
fundamental variables or dimensions of personality, and
if so, wha.t these varia.bles will be . If this can be
done, a more parsimonious solution will be reached and
4
another step taken in the direction of a closer rapproache
ment between the clinical and the statistical approaches
to the study of personality.
B. STATEMENT OF PURPOSE
The aims of this study follow:
1. To determine more exactly what personality
variables are being measured by the items which compose
the MMPI.
2. To determine- whether the PI variables, as
presently scored (Hs, D, Hy , etc.), are supported by
this statistical study .
3. To build a foundation for the future develop ment of new scoring scales. These scales will measure
the variables isolated in this study and will supplement
the present diagnostic scales and presumably increase the
usefulness of the ins t rument.
4. To decide whether a clas sic textbook example of
paranoid schizophrenia or manic depressive sychosis dif fers significantly in hi MMPI performance from the usual
hospital patient who has been similarly diagnosed.
C. E XP RIMENTAL DESIGN
If one were to do a factor analysis of the MMPI
5
by the standard technique wherein each item 1s considered
as a separate test, it would be necessary to administer
the MMPI to a few hundred subJects and then develop a
table of intercorrelations based upon the 550 separate
items. However, this matrix would be Bo large as to be
impracticable, for it would contain 150,975 separate cor
relation coefficients.
Fortunately, it is not necessary to u se the stand
ard or R-techni ue in solvin this problem. Stephenson
(61, 62) has BU ested the inverted factor or Q-technique
which will be utilized in this research design. In the
R-technique, test variables are correlated by means of
a series of ersons; in he -technique, persons are cor
related by means of a series of tests.
3
A third tech
nique is available . This is the P-technique (13), in
which test variables are correlated by mean of a series
3
correlating people means measuring the degree of
similarity of their responses. For a spec1f1e example
of how this is done, the reader is referred to Table IV.
Here the tetrachoric correlation between indivia_uals l
and 2 is computed by mans of a four-fold table. Cell a
contains the number of items both subjects sorted as true.
Cell d contain s the number of items both sorted as false.
Cells band c contain the number of items that the sub
jects answered in opposite directions.
6
4
of occasions for a single person. In interpreting the
factors derived by the Q-technique, a factor can be re
garded as a hypotl1etical "typical person" and the satura
tion of a real person with this factor is merely a measure
of the extent to which he resembles this hypothetical type.
In this present research design , the inverted fac
tor or Q-teohnique was utiliz.ed. It is especially impor
tant to exercise care in selectin the persons to be in
cluded in the study, for what one can ge t out of a fac
tor matrix depends largely on what one puts into it . With
this in mind, it was decided to select only those people
who clearly exemplified such psychiatric categories as
schizophrenia, manic depressive psychosis, anxiety neu-
rosis, etc .
It may be argued that the criterion, namely the
psychiatric diagnostic category, is it self very unreliable
and of uncertain validity . Consequently , if one were to
attempt to make interpretations of the factors on the
basis of this admittedly invalid criterion, the conclu-
4
It is well to state at th point that there is
no complete agreement among the authors, notably Cattell
and Stephenson, as to exactly what c onstitutes a P and Q
technique. The points of disa reement are not clearly
understood by this wr ter; -however, J • • Guilford , in his
new edition of sychometric ~ethod, (to be published)
distinguishes hese techniques n a manner simil r to the
above.
7
sions would necessarily also be invalid. Although it
might appear that we are running around in circles and
getting no place , this is not the case. To pursue the
latter metaphor, it is true that we are running around in
circles, but they are ever narrowing circles, and eventu
ally we will reach a definite point in the center. Trans
lating this metapho1· into the language of scientific ex
perimentation, it becomes apparent that we are employing
the method of successive approximations in this research.
D. SELECTION OF CASES
The cases used in this study were selected from
among the clinical records at the Veterans• Administra
tion Neuroueychiatric Hospi al in Los An eles, California.
In this hospita, the ·roup for of the 1PI is adminis
tered routinely ~B part of he admissions battery. How ever, in order to cu do~rn 1"1e ad.mini st a tion time, a
shortened f orm of the test is used which includes only
~?3 item rather than 550 . The items which were excluded
were those wh c do no enter into any of the existlng
scales. Since the purpo e of this study is o analyze
the tes as it is now bein u ed, it is felt that the
exclusion of the unsoored item is no reat handica.
At the europsychiatric Hos ital, under the direc-
8
tion of Dr. Harry Graysonj a research project was under
taken which was known ao the "Clear-Cut Case Study.
11
In
this study an attempt was made to isolate, from amon the
general hospital patients, some clear-cut examples of vari
ous clinical diagnostic categories . The criterion for a
clear-cut ca e, as defined for Dr. Grayson ' s tudy, in volved t hree inde endent diagnose made by psychiatrist
who knew he pa ient . This s udy had been oing on for
some im, u un or una ely , only a few case had aken
he • From th clear-cu ca udy , six case
we r e inclu ed in he re ent factor-analytic udy . Theoe
ca e cons e of hree tien s uffer n from ma ic-
depres i e cho e , and hre ara oid chi ophrenic .
In he d scu on of he problem wi Dr. Grayson,
an inte estin ques ion a e ; nam 1, 1s a er~on
ho i s clear-cu ca e or ex boo example of aranoid
ch zo hre ia o ewha diffe en from he u ual paranoid
sch zophrenic who con ti u e he ulk of he hospi al
population? It ia hoped ha the fac or analysis to fol
low will provide the ans r to his que tion insofar s
he patien • beha or on he I is concerned. If he
clear-cu paranoid chizophrenicr. ha e h i h loadin son
he a fac or a he oh r par noid c i ophrenics,
i ca ea urned ha he e is no s1 nif1can difference
9
between the two group .
..
The rest of the cases, twenty-eight in number ,
which were elected for inclusion in this study were ob-
tained by carefully checki the clinical records of pa-
tients in the Neuropsychiatric Hospital . In order to ex
plain the criteria for selection , it will be necessary to
discuss some of he hosnital rocedures . men a patient
is f i rst admi ed to the hospital , he is seen by a psy-
chiatrist ho makes a tenta e dia no
•
After he pa-
tient ha been in the hospital for some ime , and this
may vary from three weeks to three months, he is een by
a board of three p ychia rists ho make another iagnos
based upon h e available informa ion a ou the patien .
This will include he so ia worker
1
·s report. A third
diagnosis , based upon the re ul of ps cholo ical te ts ,
i s made by the psychologi t. Finally , there is a new
staff dia no 1~ made a e 1 e hep tient requests
an extended pa
char e .
From the
each patient ~
record folder:
nosi
'
and he
f a ien n
from he hospi al or at the time of dis-
a o e s a emen s ,
a lea three dia
th admi ssion dia no
p s cholo i dia no
th tud ..
w 11
no e
1
'
i r.! .
e e ident that
in his clinical
h s aff dia -
The inclu ion
up n all three
10
diagnoses being in agreement . If more than three diag
noses have been made , all were required to be in agree-
' ment.
5
Thio was the basic .criterion . In addition, only
those patient s were chosen who were veterans. male, white,
varying in age from 25 to 35 , and in I. Q. ratings (Ship
ley-Hartford usually) from 90 to 120 . Some exceptions
were made for the patients included in the clear-cut case
study because very few of these people fell within the
stipulated age range .
A uestion wa rais ed as o whe her another in-
vestigator , usin he sa me criter on would have chosen
the identical patients . In o her ords , is there a high
inter-rater reliabi l ty for election of pa ients o be
included in hiss udy ? hile no fi re can be pre-
sented, the na u e of selec ion should insure ah h re
liabil ty . The first step consisted of reviewin the
records of all pa ients, o er a hou and , who had been
referred o he clin cal p ycholo sec i on for e in .
Upon en er n the neuropsychiatric branch of the hos
pital, he patients had b en referred o he psychology
section fo · routi e tes in . In addition , specia l
5rt is o be noted hat this criterion is somewhat
differ n from Dr. Grayson e criterion for inclu ion in
the clear-cu cases ~ .
11
referrals were made for specific purposes . The files
in the psychology section go back to July, 1946, and
these records were collected in the summer of 1950. The
patients who were not male, white, in the age range of
25 to 35, of normal r. Q., and who had not taken the MMPI
were eliminated. The second step involved comparison of
the psychologist's dia nosis with the hospital staff's
diagnosis. All cases showin , obvious disagreement were
eliminated. U p to thi point, the only records used were
those available in he peycholo ection. In step three ,
the com lete clinical record folder was obtained and check
ed to make sure that all the dia noses a reed. Somewha
less than fifty ca e appeared to meet the criterion.
These were reviewed by Dr. s . M . Wesley, a V. A. consult
ant, and any cases still conside ed doubtful were elimin
ated.
All in all, 34 p tients were selec ed for partici
pation in this study . They were diagnosed in the following
categorie s :
Schizophrenic, paranoid type:
Clear-cut oases ......... 3
Regular patients .......• 5
Schizophrenics, unclassified ...• 3
Manic Depressive Psychosis, Manic
• • • •
Clear-cut ca es
Regular patients
• • • • •
Character Disorders
Psychoneurotic
• • • • • •
Anxie y Reac ion
Depressive Reaction
Hypochondria sis
• • • •
• • •
• • •
Total
• •
. 3
. . . 3
• •
• 6
. . . .5
• •
• •
. 3
. 3
34
In addition, a thirty-fl th case was included.
This case is not a real person , but is the enera l
12
key or the
11
normal
11
wa of clas fyin the statements .
This key wa s determined accordin o he way the corners
on the cards were cu in the individual form of the test .
The thirt y-fifth ca e wa in luded in the hope that it
would pull out a eneral normal factor , ifs ch exists
in the MMPI .
In order to prevent a pas ible misunderstanding,
i t is perha s desirable to sta e that although the num
ber of cases incl ded in this s udy is onl thirty-five ,
lJ 7
the actual Nin the Q-technique design is equal to 373,
for this is the number of items used.
E. THE NEED FOR AND IMPORTANCE OF THIS STtJDY
The MMPI is one of the most widely used psychologi
cal tests in hospitals and clinics dealing with abnormal
personalities. It is also used in schools and industries
as a screening device. There is little doubt that an
experienced examiner can make a diagnosis based upon the
MMPI that will, in the lar e majority of cases, agree with
the psychiatric diagnosis (50). However, it is the con
viction of this investigator th t ince the MMPI is, as
the name implies, a personality inventory, its Job is to
measure personality variables and not sim ly to confirm
the diagnosis of the p ychiatri t. This po~nt is increas
ingly recognized, and some inve tigators, notably Gough,
have developed experimental scale which urport o mea
sure such variables as Dominance (28), Prejudi e (29),
1 Status (22, 23, 24), Social Responsibility (30), etc.
These scales have been developed empirically in much the
same way as the original scales.
No one, to the bet of th wri er•s knowledge,
has attempted to analyze the internal structure of the
test. This sta ement needs to be qual fied, fo there
14
have been a few factor analyses of the MMPI (2 , 15, 17,
47 , 63, 67, 70 , 73), but all of these studies start with
the intercorrelations of the twelve existing scales. This
method is not the most desirable one, because it starts
with the assumption that the items, which at present make
up these scales, are equally valid measures of the syn-
drome . We know that this is probably no the case be
cause of the diversity of the content of the items belong
ing to the same scale. Litle (47) com le ed an item
analysis aimed a de erminin hoe item wh ch are highly
predictive of the scale and those tha con ribute little
or nothin to what the scale as a hole measu es. In the
course of hi invest i ation he di covered that there was
considerable va iation from cale o scale in both the
number and de ree of "pure " items. The Hypochondriasis
scale, for example, a pear to have a number of relatively
"pure*' item wi h unusually hi h redic ive alue, whe eas
the Hypomania scale ha no 1 ems with an n ernal con-
' sietency correlation coeffici n hi her than .48.
The po n be n made is si ply ha he starting
with a ma rix of sc le in ercor elation, i is impossi
le ode 1ve basic er onalit ar ales, be au e h
scale hem elves are com o ed of i em which me ure
differ nt thin a . From hi typ e of anal s , all h
15
can be expected is to obtain factors related to larger
syndromes or groups of traits, such as neurotic and psy
chotic tendencies, which in fact were found in many of
these analyses.
For the above reasons, the present study ignores
the scales and attempts to deal with the individual items.
The difficulty of dealin directly 1th the items wa
indicated earlier, but this aim is achieved indirectly
through the use of the Q-technique. It is hoped that,
by utilizing this technique, it will be po sible to deter
mine more exactly the p ersonality variables being measured
by th · I.
CHAPTER II
GENERAL INTRODUCTION TO THE MMPI
A. DESCRIPTION OF THE INVENTORY AND ITS DEVELOPMENT
In discus sing the purpose of their inventory,
the authoro state:
The MMPI is a psychometric in rument designed
ultimately to .provide, in a sin le test, scores on
all the more im ortant pha ee of personality. The
point of view determinin the importance of a trait
in this ca e is hat of the clinical or per onnel
worker who wi es to assay tho e rai that are
commonl y characteri ic of di a ling psychological
abnormalitie . Te in ru en itself co prise 550
statemen s co er n a wide an e of ubjec matter -
from the phy ~cal cond1 ion to the morale and the
social attitude of the individual bein te ted
( 38, p . 5) •
To t ake h e est , h e nubject or all of these
550 statements into one of hree ca egorie: true, fal e,
' and carno say . The eci ion a s to which categor o
use is based upon whe hr he fe ls tha a particular item
i s true as 1 applies o hi or whe her he con id rs it
to be fal e . The re pones are all ied o a to yield
cores on nine cl n cal and four validi y cale. Thee
allie , or ra scores , are then ransla ed i to T- cores
( tandard score 1 he ean se a 50 and he andard
de iation at 10) b mean of a le . The - scores for
he diff rent scale are entered
are directly co arable . I
a p r ofi l e char and
o e tre
00
ed hat,
17
although each individual scale has meaning by itself , the
personality descrintion and diagnosis are based upon the
entire rofile or configuration of cales and not upon
1
isolated peaks .
---,
The MMPI has bo h an individual (card) form and a
group (booklet) form . In this study a shortened version
of the booklet form was used. The problem of the equiva
lence of thee for swill be taken up in the ection on
the reliab111 y of he tes ins ru ent . Howe er, it may
be men ·oned her e tha ·he available evidence sup ors he
notion of equivalence .
The history and me hodo o of he development of
the l1MPI ha ee adequately covered elsewhere (38) .
Many of he de a ls cancer ed w h he de elo men of the
eparate cale are qui e ariable , and o attempt a com-
plete description wo d be beyond he scope of h s di -
ser a tion . The aj m ha is of his re earc is upon
he inter-relation h of he tem
,
and the cale as
1
The necessity fo r in e re in he MMPI by a pro-
file analys i ra her han epara e hi h _oin 1s on e of
the reasons h nves 1 tor bel eves ha a factor an-
alysis a sed he in ercorrela io s of the cale
scor canno o sola ion o ba 1c e sonali y ar -
a le ut only o more inclu e rai a erns . To et
a ba ic er onal y vari bes, becom necessary to
deal w h he in ercorrel tion s of h e i es , and tis
hoed ha he - echnique will a roach his o jective .
18
I
such are not used.
Briefly then, the general methodology employed in
the development of the MMPI was to contrast a normal (con
trol) group with carefully selected homogeneous groups of
clinical cases. Each scale was empirically derived by
studying the responses of these psychiatrically diagnosed
groups to the separate items, com arin hem wi h the
responses of the con rol group, and then combining those
items which sho ed con is en and s1 nifican difference
into a scale. The 11n1cal grou, of cour e, vari d for
each new scale, but in its entirety it con isted of about
800 subject who ere a n ea he ni er y of i ne
sota Hospital. Thej were carefully selected o include
1
only relatively ure and uncom lica ed exa les of each
psychiatric dia nosis. s a uch fac or a oc al
and marital sta u, age, sex, and in 111 ence were neg-
lee ed, bu care was a en to exclude h x remes and
to avoid obviou 1. dis ur in nfluence .
The normal or contra cases included a ou 700
subjec • A are or io of hi s group were 1 1 or
to patien in th Uni ers ty o 1 al who ta ed tha
they were not under a phy ician' car and considered
themselve to e in ood heal h. The. were reason by
representa ive of he occu~at anal and oc al level
19
of the University Hospital patien s . In addition , abou
200 MMPI records were obtained from students entering
college , and about the same number of records were ob
tained from WPA workers .
The method used to de elop the MMPI scales has a
number of advan ages . One of he e is that it encour
ages the discovery of new factor which are related to he
syndrome~ being con idered. These factor ma be re re-
sented by subtle 1 ems which s atis icall erve o dif-
ferentiate the experimen al from he con rol group. Bu
perhaps the most importan advan a e of his methodology
is ht it fr es he in e or , for the mot ar, fro
any dependency on he accurac of h
report . The meanin fu l relationship 1
ubject•s self etween hep -
chiatric ca egory and the subjec • ver al ehavior. The
importan - thin is ha he subject sa s ,
11
I feel eak
all over muc of he i e ,
11
and n he er ci of the
nta emen. Ina~much
he andardiza ion of he es
wa('.!
no based on the objec ive accurac of he res on es ,
no as u
ion('!
a.s 0 he eraci of he self-re ort need
be ade by the u er of he n rumen
•
hi i no in-
tended 0 m 1 ha he exa iner i 0 ...1.n er sted in
l'1he her he u j
C is on ciousl or unconsciou 1 ly-
in
•
0 h c n rar. , he F, K , and L SC le r e ure
20
of one ' s test taking attitude and are used to determine
the validity of the individual ' s handlin t of the test
items . The lat er is a unique contribution of the MMPI
and something tha o her structured ersonality inv ntor i es do not have . The oint bein made is im ly that the
examiner need not make any ass tions a o the accuracy
of the subject's s a emen about him elf. It is enou h
to know that a par icular cl ss of ps, chiatric pa ients
said the a e hi s a ou themselves , and in hi respect
th su Ject nder con idera on is similar o he cri-
terion group . Becau , there is no de endency upon he ac-
curac of he elf-re or
'
and ecau here are check
on the te t-takin at i ude, he I is one of he mos
used and mos u eful of he e onalit in e ories (7 ,
18) .
B. DESCRI IO TE SC LE
2
1 . The Cl ical cale
a . Hypochondr asis (Hs) . Thi sea e measure e
a aunt of abnor al concerr bou bod ly func ions . A
hi h ~core nl o ndic . e an i a ure approach to life '
2
he e de er ion s r a ed upon ho se found in
the I -anual , and ome unpu 1 hed ma erial iven to
the wr:ter b • G . Gou h .
21
problems.
b. Depressiop (D). A high
0
core in the D scale
is indicative of poor morale coupled with a feeling of
uselessness, inability to assume a normal optimism with
regard to the future, lack of self-confidence , tendency
to worry, narrowness of interests, i troversion, despon
dency, and subjective dis re s~ .
c. H~ster a (Hy). The H cale mea ure the degree
to which he u J ee s ike pa ient who v e de eloped
conver~ion-t pe er a p o • , uch ind vidual are
also likely to be immature, narcissi tic, 11nreal stic,
and nin htful . Te _ and c le s are co rela ed
both stati call nd n eanin .
d. ychop De __ Thi s ale mea-
ure the imila i y of he b j ec to a r ou of per on
who e chief diffic ty lie in he ab ence of deep e o
tional res onai ene , na ility to p r ofit fro exper -
ence , and di regard of ocial ore . eople with 1 h
loadin on thi scale are sually char cter z d ao ein
irres o sible, i pul e , and actl o .
e .
-Femin
core nd i a e - de ia
in he d i c on of the o
c o fi or evid , ho
on of he bao c ·n eres pat ern
o i e sex. W o ou sid
...
al y cannot e aso ed.
22
f. Paranoia (Pa). Patients scoring high on this
scale are characterized by suspiciousness, over-sensi
tivity, and delusions of persecution, with or without ex
pansive egotism. On a more normal continuum, they m a y be
described as being aggressive , criti cal , irritable, stub
born, and moody.
g. Psychasthenia (Ptl. This scale measures the
similarity of the subject to psychia.tric patients who are
troubled by phobias or compulsive behavior . There are ac companying generalized feelin s of inadequac , apprehen siveness, tenseness, insecurity, and agitation.
h. Schizophrenia ( S _c_l • Su j ec ts coring high on
this scale tend to manifest bizarre and unusual thoughts
or behavior and are usually described as ein bashful ;
withdrawn, and oversensi tive .
1. Eypoma_nia (Ma ). Thi scale mea ures the per
sonality factor characteristic of persons with marked
overproductivity in tho ht and action. These people
are usually confident , distractible , irritable, impatient ,
and aggressive .
2. The Validity Scales
a. The ~estion Score(?) . This consists of
the total numter of "cannot sa
II
or unan ered i ems .
23
The size of this score affects the significance of he
other scores. Obviou ly, should there be too many items
of this kind, the remaining ones might be too few to be
revealing . Accordingly , should a patient place more than
fifty items in the ''cannot say " category, he is required
to deal with these items again in order to reduce the
total to fifty or less .
b. The Lie Score ill· It is a measure of the de
gree to which he ubject may be attemnting to falsify
his scores by al ays choosin he response that places
hi in the mos acceptable l i ght socially .
c. The Validity core ffi. This scale consists
of a set of items very infrequently answered in the
s c ored direction by he standardization roup , an hence
is indicative of a sort of ross ec entricity, unconven
tional behavior, carelessne sin responding , or deliberate
deception .
d . The Su or V ___ cale ill_. K s essen-
tially a correction fac or and is used to improve the dis
crimina or y n o er of he clinical scale . It identifies
test-takin a titu es, an . affords an index of the de ree
to which as Jee ha been guarded or eva ive (hi h K) ,
or too frank and self-critical (low K) in respondin .
In addit on o he bor scales , a num er of authors
24
have developed tentative scales for measuring such traits
as social introversion (18), social responsibility (30),
academic achievement (26), social status (23, 24, 25),
prejudice (29), dominance (28), neuroticism (74), low oack
pain (35), parietal-frontal brain damage (3), and subtle
and obvious keys for some of the clinical scales (72).
C. EVIDE CE FOR RELIABILITY AND VALIDITY
1. Reliabi 1 ies tudies
In his book on fundamental statistics, J •• Guil
ford emphasi zes he oin ha no absolu e coefflcien of
reliability can be i ven for any test (32) . Amon he
reasons he ives which are especiall y relevant to the
M~ tPI, is the fact tha he reliabi l ity coerficient will
vary from one population to anot her . Here we are dealing
with various categorie s of sychiatric patients as well
as with normals. A second reason for the variability of
the test reliability is its dependence upon he met _od
of computation -- whether test-retest or alternate form
reliabilities are used . Since it is i m possible (becau e
of the nature of he i ems ) to divide the est into two
strictly comparable halves, the spli -half method has
not been u ed. A third diff cul yin achievin high re
liability with which he au hor of any ersonality test
25
is confronted is that the rea er the responsiveness of
the test item to the mood changes of the individual, the
less stable, and hence the less reliable the test . In
spite of all these difficulties , the data reported on the
reliability of the ~MPI appear to be very satisfactory.
Three different groups of authors, all using
slightly diff erent m ethods and populations, have reported
on the reliability of the M1PI. Hathaway and cKinley
(38) comput ed the test-retest reliability for six clinical
scale s using he indivi ual form of he test on a popu
lation consistin · of approxinately 40 normals . The time
between t est and retest varied fro hree days to slightly
over one year. Cottle (16) computed the alternate form
reliability, usin both the individual and r oup forms of
the test. His population consis ed of 100 normal s , and
both testin s took lace within one week . Holzberg and
Alessi (40) al so used the alternate form method of com putin the reliability, but differed from Cottle in that
they used the full and shortened version of the individual
form of the te t . Their subjects ere 30 psychiatric
patient s repre en a ive of six diagnostic categorie .
Both testing took nlace wi hin three days .
The results of all hre studies are shown in he
accompanyin Table I. It will be seen hat, ins ite of
26
TABLE I*
RELIABILITY COEFFICIENTS REPORTED FOR THE MMPI
Hathaway Holzberg
Scale and Cottle and
i cKinley Alassi
Question . 75
Lie
. 46 . 85
F .75 . 93
K
. 76
Hypochondria sis . 80 . 81 . 67
Depression . 77
.66 . 80
Hyster ia . 57 . 72 . 87
Psychopathic Deviate . 71 . 80
. 52
Masculinity Femininity . 91 . 76
Paranoia . 56 . 78
Psychasthenia . 74 . 90 . 72
Schizoohrenia . 86 . 89
Hypomania . 83 .76 . 59
*This table appeared ori inall in the revised
manual for the PI (38, . ?) .
-.
27
the different methods and opulations used, the results
are similar.
2. Validity Studies
It will be recalled that the scales of the MPI
were developed by contrastin a group of normals with a
group of carefully selected psychiatric patients. The
chief criterion of excellence was the valid prediction
of clinical cases a ainst the neuropsychiatric staff
diagnosis, and it is with this criterion that the test
must be validated. However, the problem is not this
straightforward. In discu sin he validation of the
MPI test scores with the psychiatric diagnosis in the
Burros Yearbook, Rotter (6) states:
Its reliability and validity are dependent upon
the reliability and validity of diagnosis of disease
entities themselves •.• not only must it be true hat
such disease entities exist, ut also that they can
be reliably determined from one psychiatrist to an
other.
This dilemma is reminiscent of the early problems
concernin the validation of in elli ence tests . The cri
terion against which the early es s were validated was
the teacher's estimate of in elligence, which w s an ad-
mittedly inaccurate cri erion . ow, after constan
improvement, it is the intelli ence test that becomes the
28
. criterion, and teachers• ratings are validated against
that instrument. Because of successive improvements, the
experimental test has become the criterion . One might say
that we have pulled ourselves up by the bootstraps, but
this is the way science develops . An attempt is being
made to do the same with the MMPI .
At the present time, at the University of Minne
sota Hospital , Dr . Hathaway and h i s staff are doin a
study involvin dia nos ic "misses," cases in which the
ps chiatric dia nosis and he J .1 I dia nosis disa ree.
The s e cases are bein , follo ed u to etermine which
dia nosis ra correct a nd, if possible , o find ou why
a mistake 1as made . This study and o hers in pro re s s
are expected o improve th . validi of the inventory.
. eanwhile, it is nece ~ s r o work Nith an imperfect
criterion, and thi will necessarily reduce the validity
co efficien .
A search throu h e 11 erature reveals fourteen
studiee3in which the uthors discuss the validity or
diagnost·c dependabi l ity of the I . In order to sum-
marize these tudies, a briefs atement about each will
be iven . This will be folloved by a ratin which will
3
Hathawa and c i ley reported fou separate valid
ity tud1es on different scale , here considered as one .
29
indicate whether the results are positive, essentially
positive, questionably positive, or negative. A similar
system was used by Ellis (19) in his article on the val
idity of personality questionnaires.
a. Aaronson and Welsh examined 98 MMPI records.
They had twelve judges sort the data into the proper diag
nostic categories using profile analysi~ technique (21).
The authors state: "Identifications were made correctly
at far above the .01 level of confidence for the psycho pathic, psychoneurotic, and psychotic profiles and between
• 01 and • 02 levels for the alcoholic group" ( 1, p. 324) •
The results reported here differ from those obtained by Ru
bin on the same data (Infra, p. 33). Evaluation: Positive.
b. Benton gave the _MPI to a gr oup of 85 male
neuropsychiatric oatients suf ering from known disorders
in order to evaluate the instrument. He found that:
(a) Five of the 10 schizophrenic oatients ave positive
results, i. e. T-score of 70 or over, on the
11
Sc
11
scale.
(b) Five of the nine hysterical patien s ave positive
results on the "Hy" ·scale. (c) Thirteen of the 16 delin
quents gave positive results on the
11
Pd
11
scale . (d) Nine
of the ten confessed homosexuals gave positive results
on the "Mf" scale, but only i f they were hon est in taking
the test. (e) Of the 17 organic cases , no eneral state-
30
ment of the influence of physical disease on "neurotic"
trends can b e made {4). Evaluation: Questionably posi
tive.
c. Benton and Probst asked a group of four aval
psychia rists to rate 70 patients on a two-point scale ,
normal and abnorma.l, with respect "o each of the nine
clinical trends investigated by the MMPI . The AMPI wa
then administered and the ean est
0
cores of the normal
and abnormal groups were c omputed and compared. They
concluded:
There is a significan de re of a reement be
tween he psych atric ra ings and he e s t sc res
wi respect o the sychopa hie Deviate, Paranoia,
and Schizophrenia rends ... There wa s no a
significan degree of a reeme t between the psy chi atric ratin ~s and the est scores wit respect o the
Hypochondriasis, Depre sion, Hys eria, Feminini y,
and Psychasthenia trends (5, p . 78).
Evaluation: uestionab- ly positive .
d . Capwell gave the MMPI to groups of delin
quent and non-delinquent girls . She concluded hat,
"Each ·cale except the Hy or Hys eria cales shows a
clear differentiatio be we en h two groups . The grea -
es differ nee appears in h scores for Pd (Ps cho
pathic Devia e)" (9, p . 297) . Evaluation: Positive .
e . Gou h , in hi particular
O
udy , adm ni s er d
the 1M I o 136 neuro ;s c hia ric army cases and found
31
that the scores of normals differed basical l y from those
of the psychiatric patien s, and furthermore, distinct
profiles could be distinguished for the mild, moderate,
and severe psychoneurosis, psychosis, and psychopathic
personality groups. However, the diagnosis of the psy
chotics was not as accurate as that of the neurotics (21,
p. 37). Evalua ion: Positive.
f. Hathaway and M :cKinley , in their new manual
state:
As for validi y, a high score on scale has een
found to predict posi i el the corresponding final
clinical diagnosis or es imate in more than 60 per
cent of new psychia ric ad ~sion. This percentage
is derived from different a ion among various inds
of clinical cases , whic is considerably more diffi
cult than mere differentiation of abnormal from nor
mal rou s. Even n cases in which a hig score is
not followed by a corres ondin diagnosis, the pre -
ence of the trait o an abnor al de ·re in he
0
ymp
tomatic picture will nearl al ay s be noted (38, p. 6).
Evaluation: Ess ntially po~itive.
g. Hunt and hi co-workerR studied h e effic ency
of some standard Jv!PI nrofi e sins in differentiating
betwe en sycho ic and non-J rchotic adul, m ale psychi
atric pa.tient • They fou d hat he 1'-1MPI dia nosis
agreed. with he criter on ·n 61 ner cent of the 90 ca es
studied. Howe er, h rela ion h i p was no statis ically
significant ( x
2
= 2.69, >-05) (44). Evaluation:
Questionably po iv.
32
h. Leverenz used the MPI for a year at the Fort
Snellin ~ Station Hos ital, ana found that in the majority
of cases the MMPI confirmed the clinical impression. He
further stated that even in those cases where it did not,
the test was of defi ite value in revealin unsusuected
,L
trends, evaluatin borderline conditions and mixtures
of neurotic and psychotic elements, and ivin a prog
nosis (46). Evaluation: Es entie.l l y positive .
i. Meehl e ployed a bli d dia ~ no stic roce ure
usin only the .J _ PI nrofile to cla.ss fy his atien s
as psycho ics, neuro ics, and conduct disorders. He
found ha a nrox ma el wo-third of ac al a.bnormal
were iden ified, but he still obtained JO per cent false
negatives. Of he bnormal ca es, .bout we-thirds ,v-ere
la.ced in he a p: nropriate dia nos ic cate ory. He con
cluded that ·he discrimination were much better than
chance in h e statis ical sens, bu al o i ndicated that
the pro ortion of false cla . sifica ions was consider ble
(50). Evaluation: Es entially positive.
j. ichae l nd Buhler gave he 1~PI to 90 neuro-
p s ch atric a ien sin a public gene ~l hos i al and
found it o be only 45 per cent ucces sful . The sat
th .t s ran of ucce f 1 d i no is wa l imi d , and
o 11 1 help in discrimina inu p s chopa
33
from psychoneurotice (52). Evaluation : Ne ative.
k. Morris studied the -u I scores for 320 naval
personnel. These included controls , psychoneurotics ,
psychopaths, and schizophrenics . He found that "the
inventory does differentiate border l ine norals from
serious pathological states but does not aid in the dif ferential diagnosis among the athological roups " (54,
p . 374). Evaluation: Es en ially positive.
1 . Ribin used he analysis of variance method
to evaluate the nin cl n cal cale of the MPI by check ing it against the psychia ric diagnosis of 98 V. A. pa
tients who were classified as psycho ath , neuro ics ,
psychotics , an alcoholics . His conclusion was that the
Pd and Sc scales were useful , but in eneral, the test as
a whole wa found oh ve lit le dia nostic value (56) .
Evaluation: Ne a ve . It should be recal ed, however ,
that Aaronson and Welsh r worked these same data and came
out with distinctly positive results .
rn . Schmidt adm.nis ered he M!PI to both normal
and abnormal Army Air Force personnel and fou a. that it
did
11
dist1ngui h gr phical y and •with statistical igni
ficance between nor al oldiers and hose di nosea as
consti utional p ychopaths ; ild and severe neurosis; and
psychosis" ( 58 , pp . 130-131) . ~va l uation: Posi ive .
34
n. Van Vorst used the J ~I with a grou of psy chopathic delinquents, and reported that the ob ained re
sults did not anpear to support the claims of the test
"to an extent which would justify defining any character-
4
istic response pattern for the psychopa hie personality"
(6 8, p. 583) . Evaluation: Negative .
In summary, of the 14 studies reported here, the
validit results were positive or essentially posi ive
in eight cases, quest onably positive in three, and nega
tive in hree. Of the three ne ative case, he data of
one was reworked and positive result s obtained. From the
above summary, one m a y cone lude ha. the l\ I 1 s a most
promising instrument for personality diagnosis. The fo~e
going i alt o eth r con ruent wi h previous tudies ade
by Ellis (19) and Burton (7) in indicatin the value of
the ,!PI for certai kinds of personality appraisal .
The ·
1
1 PI 1-1as chosen for use in this research be
cause it is area onably reliable and valid instrument
with which o asses n important specs of personality.
Factor anal sis ec iques, roperly applied o this test,
should reveal he basic p rsonality rai s tha it mea
sures.
4
This s udy wa printed only a an a stract and no
enou h m eria,l vTas re or ed o e lua.te it pro rly.
CHAPTER III
BRIEF INTRODUCTION TO THE ~ 1 ETHOD OF FACTOR ANALYSIS
AS IT APPLIES TO THIS INVESTIGATION
A. GENERAL ORIEN'l
1
ATION
Factor analysis has become a very specialized
branch of psychometrics, an the literature on the sub
ject is both voluminous and highly technical . Any attempt
to summarize this 11 erature is bound to be unsatisfactory ,
since it will be too uperficial o satisfy he exper and
too detailed to in erest the eneral reader. Consequently,
instead of ttempting to urvey he field, only a brief
statement regardi g the na ure of factor analysis and the
general procedures used in this study wil l be made . Ref
erences to the details of he compu ational process will
be found in the bibliography .
Holzin er (41, . J) ives a ood escription of
the nature of factor analysis in the opening paragraph of
his book. He states:
Factor a na,lysis is a branc of statistical theory
c oncerned with he resoluti on of a s e of descriptive
variable s in terms of a. mall number of categorie s or
factors. This resolution is acc omplished by the an
aly sis of the intercorrela ions of the variables. A
satisfactory sol tion will yield f ctor whi ch c onvey
all the esRential infor ation of he ori inal set of
variables. The chief a is hus o . t ai scien i ic
parsimony or economy of descrip ion .
36
In the present study, the methodolo ical orienta
tion is guided in the ma.in by Thurstone
I
s M:u.ltiple Factor
Analysis (65) and Chapter XVI of the second edition of
Guilford's Psyohometri~ Methods (31). Two important modi
fications of Thurstone•s basic desi n were made. They
are:
{l) The use of the Q-technique.
A brief description of this method and the
reason for its use has been given in a previous chapter
under the general headin of exoerimental design. For
a more extensive discussion the reader is directed to the
•
following references in the biblio raphy: (12, 61, 62).
(2) The use of IBM punched card machines.
Instead of using the complete centroid method
to extract the factors from the original correlation mat
rix, or to reduce the correlation to a smaller factor
matrix, IBM punched card m ach nee were used. This pro
cedure was develoned by Ledyard R. Tucker (66), and is
based upon an adaptation of Hotell1ng
1
s iterative process
for determining princi le components (42).
In order to ive psychological meanin to the ob
tained factors, a comprehensive stldy of other factor
analytic investigations of er onality variables 1r a made.
Again, it is impossible to do Justice to th extensive
37
literature in a paragraph or two . The major source book
used for this purpose was Cattell's Description and Mea
surement of Personaliti (11). The excellent Job done by
this author in surveying and cataloging the various syn
dromes, factors, traits, etc. provided a very important
aid in lnterpreting he factors found in he present
anal:ysis.
B. REVIEW OF FACTO RIAL STUDIES OF THE MMPI
A dili en search of he 11 erature revealed
ei ht st di e s in w h ch the factor-analy ic echnique was
applied o the] PI. Four of these studies were unpub
lished ma ters or doctoral dis ertations . Four of he
e ht authors included in their matrices only the Il4PI
scales (47, 63, 67, 70), while three other tudies in-
cluded. o her da a ch as abili (15 ), a itua.e (2),
a nd in erest tes s (17). The ei hth study, by Winne
( 73) , utilized a s l i h ly differen des n in whic he
fac or analy ed hirt selec ed items from th H, D,
and H. cal s of ,he PI. Because no two studies are
exactl alike, i nece ry o review each one separ
ately efore co p rin h ir res 1 . To acili a e he
cop risons, as a n ard e hod of resen in the data
will be u ed, and on y r s 1 s r l a ed o he fac-
38
torial composition will be outlined.
1. Abrams (2)
{a) Purpose: To compare neui'otics and normals in
order to disclose differences in their in ellectual func
tioning and emotional adjustment wit the basic aim of
discovering the fundamental factors that underlie these
differences.
(b) Subject_s: World War II ve erans divid_ed into
a neurotic group of 201 individ als receivin disability
pensions for psychoneurosis and a nor al or control group
of 117 men.
(o) Procedure: Two matrices were derived, one
composed of the in vercorrelations of the test scores ob
tained from the neurotic subjects, and the other from the
normal subjects. Seventeen tes t scores were included in
each matrix. They consis ed of he ?, L, F, and the
nine clinical scales of the PI, plus the five Primary
Mental Ability Scores. For each m a rix, four factors were
extracted by the complete cen r oid e hod and thee were
rotated accor ing to the criterion of simple structure.
The two sets of factors were s omewhat different a nd the
author concluded ha there was a real diff rence be
tween normals and neurotics which app ared in heir
intellectual functioning as well as in their emotional
adjustment.
(d) Resulting factors:
(1) Normal group
39
(A) Not interpreted but highest loadings
are on Hs , D, a.nd Hy. It m ay be a
neuroti c factor.
(B) Adjustment
(C) Emo ional re ctivity
{D) Externalization
(2) Neurotic group
(A) Verbal rea onin
(B) Intellectual
(C) aladjustment
(D) Severit y of s mptoms or a tendency
to1r1ar ns.rchosis .
2. Coolr and Wherry ( 1 .5 )
(a) Purpose: To determi e he relationship be
tween personality variable a me ~ ure by the M MPI,
a ptitude, and psycho-motor es s by means of a factor
analysis.
{b) Subject~: 111 naval enlis ed sub arine can
didates.
----,
J
40
(c) Procedure: Twenty test variables were inter
correlated, including the nine clinical scales of the
1 -~MPI plus the L and F scales. These were factor analyzed
by a modified Thurstone group centroid method. Six inde
pendent factors were obtained.
(d) Resulting factors:
(A) Tendency toward personality maladjustment
(B) umerical-verbal intelli ·ence
{C) Tendency to ard overactivity, reckless
ness, lack of self-consciousness and
self-criticism
(D) Tendency to ~ ard p ranoia
(E) Mechanical coordination
(F) Tendencv toward a feminine interest pat
tern coupled with compulsivity
3. Cottle (17)
(a) Pu!:Q,oee: To determine the co on factor e
ing measured by the 1 rnPI, Stron, Kuder, ad Be l l inven . tories.
(b) Sub_J_eQ._t_~: 400 ma .,le ve erans of Wo1.,ld l ar II
who were referred for te sting a nd counselin a he Psy
chological Services Center of Syracuse n verity .
(c) Procedure: Pearson product- o e correlation
-
----,
41
were computed using the raw scores of the 400 cases on
all 34 variables. These variables included the nine
clinical and the L and F scales of the lif:MPI. The Thurstone
centroid method was used in factor analyzing this matrix.
Eight factors were obtained, and ley were rotated ob
liquely.
(d) B_e§lti!}g_ £~_tors: The )1MP I ha,d si nificant
loadings (.33) on only three factors. These were inter
preted as:
(A) endency oward serious maladjust ent
i n the psychotic direction
(B) Tende_cy toward neurotic adju tment
(C) Ma cul ne-feminine 14 terests
4. Litle (47)
{a) PurJ>.o~: To subject he tJ n PI to an explora
tory stat tical a nalysi in an effort to obtain a bet-
ter understa.ndin 0 the co osi io of the e and of
the rela ionship bet,veen the scales.
( b)
Su2_Jects: 250 male fre h en and s o ho ore
s udents at ·he Uni e sit of North Carol na.
( C) Procedure: Then ne cl nical scales of the
---
I were intercorrela ed and h earson correlat·on
co f icient computed. Then he a t rix as factor anal .ed
42
using Tl1urstone • s centroid metnod. Four factors were ob t ained , but only three had sufficiently high loadin .\~ to
warrant interpretation . An ort ogonal rotation was per-
f orme with the aim· of ac:t11eving simple structure and pos-
1 tive manifold. Neither criterion was completely achieved.
(d) Resulting fa~ors:
(A) A bi-polar contin1um with passivity or
inactiv .ty at one e and ag ressiveness
or the de
other
e for o er activity on the
{B) endency tovard psychotic adjustment
(C) General ne .rotic · -icator
5 • St OU , ( 6 J )
(a) Purpo~e: To e ermine how many variab es are
being measured by the 1 •'
1
• I scale and tivhether he factor
patterns are alike for oth men and women .
(b) S __ ubJec~t-, ~~= 172 1-vomen and 153 en
Pennsylvania State Colle e .
udents of
(c) Procedur: Men and women were plac di spa-
___ , __
rate matrices made up of h e nine clinical cales of the
P TMPI and the scale . Thurstone ' s centroid m ethod was
used o factor a 1 ze
factors accoun ed or a
e
a, n i eac case h ee
very sma _ ercenta e of
43
the common variance. The test confi uration was the same
for both nen and women. The factors ere not rotated
since the author was inter sted only in deter ning how
many factors were m easured by the 1MPI scales and not in
interpreting these factors; corseq ently, it is impossi ble to com are the re ults of thi study wit the others.
6. Tyler (67)
(a) Purpose : To re ort the resul s o a fac or
analysin of h I \ ·PI ~rh ich ~ as scored for fifteen scales
exclusive of the L, F, and K variables .
(b) s~bjec s: 107 fe ale radu ~ e student pr -
paring to become elem en ary sc ool .eachers at the ni
ver ity of Cal . fornia, Berkeley. Al l ~1bj cts l · d passed
a, physica examination , a nd none had been referred to a.
psychiatri t for fur here a ni nat on .
(c) Procedure : Th e su jects answered all que~-
ti on s int e ookle form of h PI . The tes s were
scored f or the n~ne orig nal clinical scale lus ix of
Gough's experimental s cales. Thee six cal ~ were:
(1) c ial n r over on , (2) social tus , (3 J rejudice,
(4) academic achi vement riv, (5) s ocial es~onsibil ty,
ard (6) oci;.l d om n nee . The L, F, K sc les were
not u e n he na ysis . The i ntercorr 1 t i ons were
computed from the raw scores.
(d) Result!_ng factors:
(A) General aladjustment
(B) Conflic and lac~ of insight
(C) Social aggrecsiveness
(D) Self-conf dence
( E) Depende1 cy
7. Wheeler, Litle, and Lehner (70)
44
(a) PurE~se: To de ermine he ature and extent
of the interrelationslips amon the var ou scales of
t h e ~ Jv!PI by utiliiing a factor a na.lysis a roach.
(b) SubJects: 110 neuropsycniatric ma.le patients
in a, veterans' ~idm· nistra ion 11ospi al a nd 112 male col
lege studen at CLP .•
(c) Procedure: The hor ( 73 i em) form of the
ir3pr was ad inisterea. o ach of he subjects . Three
correlation matr ce rere for ea., one for he p . tients ,
' one for t he s udents, and a nother for the wo roup s com
bined. Each matrix con s s ed of welve sea es , the nine
clinica cale plus L, K, aid F. These were intercor-
related the asi of he ra scores t e Pearson
oroduct-mom n coe ·cent .
rs:
(a)
(1) Normal group
( 2)
(3)
(A) Psychotic adjustment
(B) Neurotic adjustment
(C) Masculinity-femi .nity interests
and paranoid endencieo
(D) Bi-polar mood dimension
Neuropsychiatric r oup
(A) Psychotic adju tment
(B) Neurotic ac1just 1 ent
(C ) Paranoid adju ment
( D) I O S ,nifican load ng
0omb1ned group
(A) Psyc otic
adjust ent
(B) Neuro ic ad.Ju
ment
( C) Paranoid adju rnent
(.D) Psychopathic adju ent
8. \Alinne (73)
Purpose : To factor ana y ze e responses of
a normal and a neur tic r o p 0 th arne e t items and
y so doin ~
clB~ ri v the ifferences
between th roups .
The M:MPI wa
used only s a source of i ems n d
did not , in itself , e ter into the tud
•
Th re earch
46
is reported here, because to this investigator's knowledge,
it is the only time that a number of M.PI items were fac
tor analyzed, and it is oped tha the resul of Winne's
study will be useful in interpreting the factors obta ned
in the present investigation.
( b) SubJ ~ct.~: 560 World War II ve erans divided
into two groups of 280 each . The me ber of one gr oup
were diagnosed as neurotic by the Veterans• Administ ation
and the other group was composed of ormal individuals.
(c) Procedure: Of the 117 1 e s aki p the Hs,
D, and Hy scales of the ~M I, tnir y were foun d to dif-
ferentiate et een the ormals an d
1% level of confidence as de erm
These were combined into a scale
he neur t cs at the
ed by a Chi-square test .
f neurotici m (74).
Tetrachoric coefficients were us d to measure the inter
correlation of these items, an wo matric were devel
oped, one for the nor mal r o a t d the ot er for the neur
otics. These matrices were factor a nalyzed y he Thurs
tone method and the factors rot ed or ho· nally.
(d) ,Besulting factor: Five fac ors~ ere able to
account for the r soons s of the norm al
factors 1vere ne e ssar .., to account for
the neurotic ~rou o th ame ems.
r ou wile six
he re · po se of
The fir t two f c-
tors were s1m·1ar for bot rou·o and only hes e -,~o '\\ e 1 ·e
.--------------------
interpreted. They are:
(A) Hypoohondriasis
(B) Bi-polar continuu of anxiety and de
pression
C. SUM1v1ARY
1
Upon examining ~ hese factor analytic studies, it
become [ app .rent ha the are milar wt _ recpect to
both met od lo y nd r .sult • By eferri ~ to Table II
it will be seen ta n . e ine matrice analyzed, . ix
eurotic and ix sychotic fac ors ere interpre ed. A
paranoid factor a ~ e red ree es and a p ychopathic
4?
deviate factor twice. ~ culine-fe in ne interest pattern,
enera.l maladju , en , and emotional reactivity were all
repo ed as factor in four studies . Of al of these, emo
tional reactivit co es closest o e ng a basic erson
alit y variable. For he r est, h e factor were inter preted in erm of c orbi a l o s of ,raits, syndromes , or
diagnos ic ca egories .
The re ul so ,hese studies are significant .
From hem one may v
1
• dly rea on -ha t ('\ i ce th factors
1
The re o ts f S out and w nn
in hes mmar or in Table II ecau se
1ot compara le wi e others.
re not ncluded
heir re ults are
TABLE II*
FACTORS REPORTED IN TWO OR M ORE OF THE FACTOR-ANALYTIC ST1JDIES
OF THE Ml'1PI
- - ----------------· ---------------------------------------------------------------------------------------------------
Investigator
Abrams
Normal Matrix
Neurotic l-1atrix
Cook and Wherry
Cottle
Little
Tyler
Wheeler
Normal
N europ sy c l1 lat ric
Combined
Adjustment
Maladjust
ment
X
X
X
X
Emo
tional
Reac
tivity
X
X
X
X
Neu
rotic
X
X
X
X
X
X
Psy
chotic
X
X
X
X
X
X
Para-,
noid
X
X
X
Mascu
linity
. Femin
inity
X
X
X
X
*In m aking the above comparisons a certain amount of liberty had to be
taken with the factor names of the quoted studies. For the authors• precise
viewpoints t he r eader is referred to the original articles.
P'sycho
pathic
X
X
~
ex,
49
measured by the MMP1 are dia~nostic categori s, then the
test can be used t make a differen ial diagnosis. How-
' ever, if one's aim is to URe a factor-analy ic approach
------,
to study basic personality variables, then an experimental
design that has as its starting point he int ercorrelations
of the ~-1J1PI scales has limited usefulne a s, becau s e 1 t does
not get at the separate traits which are fundamental to
the syndrome.
CHAPTER IV
DESCRIPTION OF TH STATISTICAL P OCEDURES EMPLOYED
A. COMPUTATION OF THE TETRACHORIC CORRELATI ONS
In setting up a fa.ctor analys , the first step
in the ste.tistica. proce dure is to obtain a matrix of
intercorrelations. Usually the Pearson product-mo ent
cor:r·elation coefficient is computed. Howev .r, in the
present study 1 t 'tie. 1mposqible t co u e _ Pe rson r,
becauge there v.rere ins ic n cla in ervals. The
statements t,11Jhich make u the 11J1 ,v1PI are classified by tl1e
examinees into three cate ories. tru , fa se, an cannot
say. Since th d a a e in t his ri hoto~ i ea form, i
was decided o compu e the era h o ic corr l ation co f
flcient by elimina in l_e "canno , ay" ca egory and ar
ranging the data in a four-fold a le . Actually, there
er very few items class fed in h
shown in Table III.
ca ory a s
Twenty subject di no
gory at all. Of t he re m .i er,
to six times, and only wo use _
times.
se h "ca no sa y " cate i t fro one
fr o twen y o hi t y
B cau s i wa ~ con 1 er d 1 por ant o inc _ de a 1
of t he i es int e co p ~ ~ on of ,h , corr 1 i on co ffi -
TABLE III
DISTRIBUTION OF T HE NUMBER OF I TE S CLASSIFIED
AS "CANNOT SAY" BY THE THIRTY-FIVE SUBJECTS
Tumber
of Item s
0
1
2
3
.5
6
21
30
Tot al
Subjects
20
8
1
1
1
2
l
1
35
I
51
.52
cients, it was decided. to treat each
11
cannot say" answer
as a deviant response. The rationale was that if a per
son cannot answer
11
true
11
to a statemen like the follow
ing, "I am liked by most people who know me ,
11
then most
probably he tends toward being not liked; that is, he
tends to respond. in the deviant direction. Consequently ,
the response can wit some justification be placed in the
"false" category. In this fashion all items which were
pla.ced by the examine e s into the "cannot say" category
were recorded in the devian or unhealthy direction .
It is to be emphasized hat · his rocedure is not
Justified in clinica.l practice aince the e t was not
standard.17..ed this way. It is used in the present study
only to avoid eliroinatin any items from the analysis.
1
Also, in the case s under c onsideration, ti hird cate
gory was used infr•equently, and he re corin had no ef
fect on the test profile . This proced~re had them rit
of enabling the experimenter to eliminate the "cannot say"
classification , while still keep ng all of he 373 items
in the study.
Beforl oing on o a discu
putational procedure emplo y d, 1
viel'T the assumptions un _ erlyin ·1'1e
make ure that the f ~ I d a et
on of he actual c o -
wold be we to ·e-
etr ch r c r,
a sum t·o
•
nd to
T e
53
tetrachorio r requires that both the X and Y variables
be continuous but artificially dichotomi zed, normally dis
tributed, and linearly relat ed. To mee t these require
ments, it is neces ary and logical to a ssume that the
different subjects responded to the MMPI s atements with
different degrees of a greement and disa reement. Guil
ford (32, p. 333), in discussing the use of the tetrachor
ic r with personality que tionnaires, sta es:
The as Gumption of con inuity and normality of
distribution can e defended as follows: It is un
likely that all who resp ond
1
Yes' o either ques
tion do so with equal degree of affirmation. It ls
similarly unlikely that those who respond' To ' do so
with equal degree of negation. It is mot likely
that either question represent a continuum of behavior
extending from strong affirmation at the one extreme
to strong negation at the other. Continuity is thus
the probable state of affairs, not a real dichotomy.
If a continuity is thus _ ranted, the eneral law of
unimodal distribution approach ng normality in psy
chological traits m ay be cited in defense of the other
requirement.
The equation used in the calculation of the tetra
choric r is known as the cosine-pi apuroxi ation formula
(32, 336).
The cell fre uencies re a, b , c, and d, w th band o
repre senting the unlike-sine ca es nd a and d he like-
si ned cas e s . The varia 1 be1n carrel ted re the
subjects, individuals 1 and 2, for example. Cell a con
tains the number of statements which both subjects classi
fied as true; cells band c contain the number of state
ments classified in opposing ways , true-false or false
true. In this way , every subject was correlated with
every other subject until all 595 correlation coefficients
were computed. An illustrative example is worked out in
Table IV.
B. USE OF THE HOLLEY ADJUS MET TEC NIQUE
In the R-technique method of factor analysis, the
Gcores of different te s are corr elated by means of the
Pearson roduct-moment correlation coefficient. 'When
this is done the means and variabilities of the different
tests are automat ically equated . In the case of the in-
-
verted or Q-technique factor analysis, he means and vari-
ab lities of the questionnaire items are not equated.
This addit ional dif f erence be ween the two factor a na lysis
designs was pointed out by J . W . Holley (39) in an un-
published paper . e also outlined a method of obtaining
equal means and variabilities for all individual answer
ing a roup of ersonality te t items . The s eps to be
follo ed are:
,-f
M
cd
,a
.,...
I>
...-i
rd
s::
H
TABLE IV
COMPUTATION OF THE TETRACHORIC CORRELATION
COEFFICIENT FROM A FOUR-FOLD TABLE
Individual 2
- Total
+ a b
8 180
C d
120 1
Total 160 21
.__, _______________ ___... ___ ___._.___ --..i ____ __,
r = cos
r = cos
r = cos
r = cos
r = cos
180° X '(Ee'
------
fbc + '{ad
180° X , ... 9 _3_X _ 7 _ J
f93 X 73 ♦~87 X 120
180° X f6789
·-------
0
180 X 82 . 4o
-----------1
82. 40 • 102.18
0
180 X 82 . 4o
- -------·
184 . .58
r = cos (180°x .4464)
r: COB 80.4
0
r = . 167
55
56
(1) A second paired stat emen for each item
is inferred. This alternate item ia the logical
opposite of he statement from w1ich it is inferred.
(For example: The opposite of the statement, 'You
get tired easily ' wo1ld be •vou do not e t tired
easily.
1
The opposite of the statement , 'You are
a nervous uerson' wo1l e, 'You re no a nervous
person. ')
(2) Eac inferred s atemen ls cored in a
manner opposite to iJs paired al ernate. (For ex a ple: If the answer to a i ven item s •yes,' th
ap ropriate
0
core for logical opp o 1 e wo d be
, no. •)
(3) he resultin co. e r om both Lhe ori · al
and . e in~err d sta e en t a r ro u.&: e to ther
in he co pu a ion ,f he f n 1 co r e l on c effi
cients.
The r su ting data ield an e ual pro~ortion of
1
yes
1
and • o ' re p on s ce each s a e et an
swe •ye
1
1
as a redo poi e c ns ere ~ ' no .
1
Conversely , everv sta e -ent an wered
1
no
1
las a
paired opn o site a 1,•. ered
I
yes .
1
u
Th e s ep s nvol ed
tetrachoric rare a s follow s :
co pu ng aajuste
(1) Prepare a four- fold table ba ed on the
o iginal data. (See llus r iv
(2) Av r
togeth rand ~ivid g y two . The ea o for .ve
t 1 values , r ,her · -olv ad
..._ V
· t he v lue of
IV . )
i n
inferre
l V 1 e t ne
o he or g n 1 _ t , i
d no OU l it . T
o ke p he f n 1
us ative
ex m le n Tab e V wil m te h l ear .
(3) 0
he e corre atio1 c oeff cient us n
T ABLE V
CO!-~PUTATION OF THE ADJUSTED COP~ELATION C OEFFICIENTS FROM A FOUR-FOLD
TA BLEl
,--4
Origi nal Data Adjusted D a t a
r-1
I ndividual 2
~
...
-
Total
~
.
....
I
91 II
~
180
-
~c
d
~
'
71 120 II 1
,-f
Individua l 2
rl
...
-
Total
cu
I
:::s
+
ro
I
103. 5 r 83 . 0 ~ 18 §.!.2._
-rl
t
-rl
-
l C
81 . 0 ~d' 101 . 5~ 186.
rd .
H
otal I
160
I 213 tf 37.J _
s::
H
Total
I
186 .• .2 _ I_ 186. 5Ll1U •O
80°
'
X b
I
r = co
I
-
a
• I
+
180°
X 8J . 0
I
r = cos
103 . .5 + 83 . 0 -
0
r = cos 180 X . 445
J
180 c)
X 83. 0
-
r = cos 80 . 1
0
r = cos I
I
r = . 172
186 . 5
1
For the compu on of the origina l correlation coefficient refer
0 Table IV.
\..n
-'1
58
a modified cosi e-pi formula.
180° X b
a + b
The Holley Adjustment Tee nique is a ery satis f ct or method of o taining eqtlal means and varlabili ties
fo r all individuals an werin or sorting he items in a
personal.ty · nventory such as the ... .fi'I .
lat ion coe f c en re co pu ed on the
~rom thees andardi zed distribu ions .
The new corre
asis of the da a
Sine actor analy-
si ha tra · t anally been a sed · on a matrix of Pears on
correla on coeffi en a n wh ich t e me ,n a _ v r ab 1-
itie of he et be·n. corre a ed were eq ated, it mar
be th t tis s a neces ary condi ion n al factor a naly s1 ro l ems. Certain y s em f"" aesira le for e t rans
ose techni u e o mee as any oft e conditio sand assu p tion of the R- echnique fa or a.nalysl a : ossi le .
Holley •.., cor ec ion a s another a.dvan a e . It i
o,vn he. e rachor · c r a pproaches i n alue the P rson
r mos closely ·when v e s n even pl t in h su of
he cell alue ; tha is, wen he da. a are die otomized
a h med· n p o n • Fo 1 s rea son, 1he adjusted c rre-
la 1o a e pro a l y clo s r n ~1ue ·o the Pears n r's
a l correla ion .
.59
In order to compare the amoun t of difference be-
tween the original and adjusted correlation coefficien s,
three freque ncy di str·bution tales were prepared. These
are: (1) distribution of the original correlation coef
ficients, Table VI; (2) distrib tion of he adjusted cor relation coefficien·s , Table VI; a nd (J) dl tribution of
the ifference s be ween the two, Table VII.
In umm ry, one a y conclude tha y dju t ng the
tetrachoric correla i .n coeffic en s i n accord n e e wit
Holley's tee iqu
J
he correl i n a rix 1sed i n e
Q- nay s ecomes or a, alo
0
ous 0 e ma ri 1 A d 1
t ✓ R-analy i
•
e di _ference a e no reat en o gh 0
ju
·ry
don 0 epara e s udie~ • C nsequ ly,
a -
jug ed cor elat on coef icien ,s wee ar an ed in a matr x ,
Table VIII, prepa r .tor y t the factor ~naly is.
C. ESCR nTIO OF U C KER
1
S _ C TINE ~ E-3O D
o·· FACTOR A ALYSIS ( 66 )
Rese rcr1 orke who u s e actor a a l
de of grat i u ~e o Dr. T cker for eli in ting much
ro ine c o ~ut tional 1, or evelo .J.:
us i punche c rds ·n tne calc at on of
Tucker• as sed in his 8 U
.,
•
a te for
ctor loadin
he m achines
•
r qu r ed o
od
h roe ~u e a, e a unch, a o r, and
60
TABLE VI
DISTRIBUTIONS OF CORRELATION COEFFICIENTS
CO .,fp-LJ _ED BY BOT HE ORIGINAL AND
l~ODIFIED -1ET. ODS
--
Frequencie s
Size of
Correlation Original Adjusted
Coefficient Coeffic ents Coefficients
-
.900 to
.999 2 1
.800 to . 899
3
4
. 700 to
. 799 15
9
.600 to . 699
39 31
. 500 to
. 599 62
55
. 4oo to . 499
91 85
.300 to
. 399 121 113
. 200 to
. 299 102
113
.100 to . 199 74 84
.ooo 0 . 099
33 33
-. 100 t -. 001 12 12
-.200 to -. 101
7
10
-. JOO to - . ?01
13
9
- . 400 to -. ,01
15 17
-.500 to - . 401 6
5
-. 600 0 -. 501 0 4
---------·---------
I
I
TABLE VII
DISmRIBUTION OF THE DI!c., E N CES BETWEEN
TTE ORIGI AL A iID ADJUSTED CORRELATION
COEFFICIENTS
-
-
·-
Differences
Between Coefficients Freauencies
.200 up 2
. 180 to . 199 2
• 160 to
• 79 3
. 140 to . 1.59
.5
. 120 to . 139
5
.100 t . 119 10
.080 to .099 13
.060 to . 079 26
.040 to . 05 9 43
.020 0 . 039 82
.o 0 t o . 019 164
-.020 to -. 001 154
-.040 to -. 021 48
-. 060 to -.041
25
-. 080 to -.061 6
-.100 to -. 081 4
-.120 to -.101 1
-.140 to -. 121 2
61
62
TABLE VIII *
THE CORRELATION MATRIX
, .
1 2
3 5 7
1 1000
172 327 -24.5
204 287 294
350
2 172
1000
279 -196 003 172 163 172
J
327 279
1000
-336 -139
40.5
155
421
4
-~64
-19·
1
6 -336
1000 003
-358 -187 -311
5
003
--139
003 1000
139
262 -063
6 287 172 405
-358 139
1000 146 2.54
7
294 163 1.55 -187
262 146 1000
279
8
350 172 421
-311 -063 2.54 279
1000
9
366 155 327 -327 139 2.54 230 366
10 213 279
~§
-516 -413 43.5
104 466
11 089 -012 -146 336 302 294 139
12 163 196 .531 -502
-104 311
104 480
1:3 327 163 427 -427 -172 262 204 466
14 163 113 155
-122 080
-038 187 262
15
421 -021 271 -287 196 311 271 342
16 271 254 633 - 509 -342 286 146
627
17 271
104 294 -342 -230 172
114
319
18 427
139 435
-46i
222 443
29 397
19
286 204
524 -40 -311 187 012
503
20 24.5 030 254 -238 196 405 413 294
21 311 163 271 -350 -072 294 366 279
22
375
230 .503
-381 -286 196 187 496
23
496 24.5 54.5 -531
-222 421
155
496
24
3.50 15.5 294 -389
089 54.5 294 381
2.5
405 279 473 -302 -021 327
24a
466
26 204 -012 113 -181 271 443 3 2 139
27 279
04.5
397 -350
146
389
122 435
28
358 -089 2.54
-350
40.5 509
204 230
29 503 238 327 -163 319
286
279 381
JO 37.5
113 319 -238 213
421
397 279
31
421 003 286
-319 327
262 488 40.5
32 3.58 397
413 -222 030 294 163 435
5~
3.58 155 139 -336
181 294 302 327
427 196 621
-389 -163 319 245 531
35
311 254 702 -459 -397 43.5 139
607
*Decimal points omitted
9
10
1 J66 213
2
155 279
3
327 653
4
-327 -516
5 139 -413
6 254 435
7
230 104
8
366 466
9
1000
358
10
358
1000
11 OJ8 163
12 311 727
13 552 737
14 080 122
15 358
286
16
397
887
17 413 566
18
531 .5 .52
19 413 743
20 230 254
21 146
336
22 496 743
23 496 831
24
350
488
25
466
665
26 254 113
27 342
531
28 421 163
29 350
181
30 196
375
31
375
122
32 4 5
559
33 509 238
34
559
665
35
480
929
TABLE VIII (continued)
THE CORRELATION MATRIX
11 12
13
14
089 163 327 163
-012 196 163 113
045
531
427
155
-146 -502 -427 -122
3J6 -104 -172 080
302 311 262 -038
294 104 204
187
139
480 466 262
038 311
552
080
16J 727 737
122
1000 222 204 -206
222 1000 702 146
204 702 1000 146
-206 146 146 1000
089 271
397
096
038 702 691 245
146
539 539
254
262 581 672 204
-047 545 678 130
413 271 366 045
146
213 342 054
113.
566 647 - 012
187 691 770
146
254 427 459 196
1.55 588 640 063
294 080 -00J 089
254 427 413 113
196 279
262 122
131 113 294 222
294 342 279 -0J0
054 254 350
045
089 427 545
262
122
319 381 113
054 588 665 196
080 708 776 155
63
15
16
17
421 271 271
-021 2.54 104
271 633 294
-287 -509 -342
196 -342 -230
311 286 172
271 146
113
342 627
319
358 397
413
286 887 566
089 038 146
271 702
539
397
691
539
096 24.5 254
1000
J42 279
342 1000 678
279 678 1000
389 516
4_58
294 760 516
271 245
358
336 405
327
381 787 607
473 787
684
421 443 302
413 621
389
180 003 072
336
466
311
381 196 146
451 254 254
286 230 113
574 163 131
254
593
451
397
262 311
327 760
503
336 879 607
18
19
1 42? 286
2
139
204
3
435 524
9- -465
-404
5
222
-311
6 44J 187
7
294 012
8
397 503
9 531 413
10
.552 743
11 262 -047
12 581 .545
13 672 678
14 204
130
15 389 294
16 516 760
17 458 516
18 1000
503
19 503
1000
20 480
230
21 294 405
22 466
633
2J
684 766
24
574 381
25 607 607
26 381 054
27 509 389
28 581 245
29
44J 286
30 389
146
31
496
279
32 496 451
33
509 230
34 588
640
35
566 743
TABLE VIII (continued)
THE CORRELATION MATRIX
20 21 22
2J
24.5 311
37.5
496
030 163 230 245
2.54 271 503 545
-238 -350 -381 -531
196 -072 -286 -222
40.5 294 196 421
413 366 18?
15.5
294 279 496 496
230 146 496 496
254
336 743 831
413 146 113 187
271 213 566 691
366 342 647
770
045 054 -012 146
271
a36
381 473
245
0.5 787 787
3.58 327 607 684
480
294 466 684
230 405
633 766
1000
559
204
350
559
1000
336 381
204
36
1000 678
350 381 678 1000
24
350
1.55
294
-389
089
545
294
381
350
488
2.54
427
459
196
421
443
302
574
381
405
405
405
451
405 4o5 405 451 1000
204
319 588 665 496
342 096 080
213 366
286 1 2 443 516 451
336 139 139 J81 509
327
405
327 75
286
381 381 271 350
524
516
545 271 427 496
319 054 4.59 574 389
413 443 254
397
581
342 262
633
696 366
302 366 727
822 566
2.5 26
405 2o4
279 -012
473 113
-302 -181
-021 271
327 443
238 342
466
139
466
254
665 113
155 294
.588 080
640
-003
063 089
413 180
621 003
389 0?2
607 381
607 0.54
204
342
319 096
588 080
665 213
496 366
1000 213
213 1000
4.59 327
381 .566
358 366
204 480
350 389
503 196
516
421
633
122
727 113
1
2
J
4
.5
6
7
8
9
10
11
12
lJ
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Jl
32
33
J4
J.5
2? 28 29
279 3.58 503
045 -089 238
397
2.54 327
TABLE VIII (continued)
THE CORRELATION MATRIX
JO Jl 32
33
J4
37.5
421
358 358 427
113 003
397 155 196
319
286 413
139
621
35
311
254
702
-350 -350 -163 -238 -319 -222 -336 -389 -459
146 405
319 213 327 030 181 -163 -307
389 509 286 421 262 294 294
319 435
122 204
279 397
488
163 302 245 139
435 230 381 279 40.5 435 327 531 607
342 421
350 196
375 435 509 559
480
531 163 181
375
122
559
238 665 929
254 196 131 294 054 089 122 054 080
427
279
114
342 254 427 319 588 708
413 262 29 279 350 545 381 665 776
113 122 222 -030 045 262
113 196
155
336 381 451 286
574 254
397 327 336
466
196 254 230 163
593
262 760
879
311 146
254 113 131 451 311
503 607
509 581 443 389 496 496
509 588 566
389 245 286 146
279 451 230 640
743
286
336 327 381 516 319 413 342 302
122
139 405 J81 545 054 443 262 366
443
139 327 271 271 459 254
633 727
516 381
375 350 427 .574
397
696 822
4.51
509
2 6 524 496 389 581 366 566
459 381 358
204
350 503 516
633 727
327 566 366 480
389 196 421 122 113
1000 427
358 381
397
381 381
435 588
427 1000 480
381
531 381 459 230 336
3.58
480 1000 421
509 311
480
271 327
381 381 421 1000
588
139
427 262
336
397 531 509 588 1000 089 588 230 271
381 381 311 139
089 1000 271 6 O
653
381 459
480 427 588 271 1000
389 319
435 230 271 262 230
660 389 1000 721
588 336 327 336 271
53 319
721 1000
6.5
..
Sum
11.162
5.,41
12.318
-9-875
1.826
10.810
8.405
12.650
12.066
14 . .555
5.792
lJ.218
14.754
4.796
11.122
14.716
11.655
15.623
13.021
11.048
10.079
13.562
16.396
13.710
14.695
8.126
12.506
11.189
11.750
11.025
11.913
12.809
12.150
14.584
16.025
66
a tabulator, preferably with a summary punch.
Tucker's article describing this method is divided
into five sections: (1) description of the method ; (2)
illustrative example ; (J) operational directions; (4)
card lay-out ; and ( 5 ) machine set-up . The section head ings alone are suff icient to indicate th.A.t this machine
method of factor analysis is described very completely
and in sufficient detail to enable a trained IBM operator
to work out a similar roblem.
In th s cha ter no at empt will be made to describe
the metnod itself. Instead , some points of theoretical
int er st will e di cussed. The most important point is
that instead of computing centroid factors , t his IBM pro
cedure is desi ned to det ermin e the principle components
by an adaptation of Hotellin
1
s iterative proces s . A solu tion in terms of rinci le components utili es the mathe
matically more defensible least- square method for comput ing factor loadings . Ho·ever, for hand computation , Thur-
tone ' s centroid method is m uch more feasible and has
ther efore b en more enerall used .
The de in of the present research differs from
Thurstone's method in t 11 another way . It will be re
called that in he centroid method, the values to be used
in the dia onal cells ~ re suall the hi hest correlation
67
coefficient in that column. This is considered to be the
beet estimate of the communality. Tucker, in his method,
suggests that in the original table of correlations, the
diagonal value b e unity an this value be corrected with
each successive table of residuals leading finally to a
relatively accurate estimate.
After the loadings on a factor have been computed
by the iterative process, the table of residual correla
tions is found by subtracting t he product of t he factor
loadings for each pair of variabl e s from t heir correlation.
First, a table of products, and t hen a table of residuals
is found. Extraction was continued, in this case, until
no residual correlation was greater than .08, which is
the approximate value of t h e standard error of a zero
1
correlation coe f ficient. A frequency distribution of
the final residual values has been prepared in Table x.
D. DESCRIPTION OF THE ROTATIO N PR OCESS
As a result of the factor extractions, an arbitrary
orthogonal reference frame was obtained; the projections
on these arbitrary ax s are shown in Table XI. The next
problem wa to ro t ate t h i frame , wh ch consist s of four-
1
The computation of he standard error of a zero
tetrachorio correlation c oeffi cient is found in Table IX.
68
TABLE IX
COMPUTATION OF THE STANDARD ERROR O F A ZERO TETRA CHORIC CORRELATION COEFFIC IENT FOR ANN OF 373
VPP'qq'
rt -
-
YY
1
\fN
\J. 5 x. 5 x •5 X .5
rt
-
-
.3989 x . 3989 xVJ73
.25
r t
-
-
-1591 X 19 . 3132
TABLE X
DISTRIBUTION OF THE FOURTEEN TH FACTOR RESIDUALS
(ONE DIAGONAL HALF OF THE FACTOR MATRIX)
Interval. Frequency
.070 to .079 1
.060 to .069
5
.050 to .059 11
.040 to .049
23
.OJO to
.039 J8
.020 to .029 61
.010 to .019 78
.ooo to .009
105
-.010 to -.001 101
-.020 to -.011 62
-.030 to -.021
49
-.040 to -.031 26
-.050 to -.041 24
-.060 to -.051
7
-.070 to -.061 4
Total
595
69
TABLE XI
UNROTATED FACTOR LOADINGS AND COM MUNALITIES*
FACTORS
CASES A B C D E F G H I J K L M N
h2
1 52
26 -06 -27 15 -10 16 -11 09 05 -13 -01 lJ
02 5296
2 27 -11 -04 -04
37 -13
00
23 37
00 -06 13 -12
14 486:3
3
65 -18 -03
04
17 -30 -07
14
05
00 12 -25 -0?
04 686?
4
-56
00 -06 -10 14 -06 16 07 -29
04 11
14 ·: -05
06 5044
5
-03 69 -27
02 14 16 16
-17
20 -17 17 05 -09
~ 06
7620
6
52 27 -06 36
00 -22 -08 -01 09 22 04
-OJ 17 -01 6194
7 35
38
26 -04 29 12 OJ
11
-0~
-22 02 -12 04
OJ 5142
8 64
-05
02
-13 13 -OJ -07 -08 -1 -22 16
05
00
-06 5582
9
60 08
-23 -20 -08 15 17
16
-05
02 -11 -18 01 12 6046
10 82 -46 • ~07 28
-05
-16 -03 OJ
06 -01 -08 05 -02 -011.0103
11 23
·28 12 47 08 16
36
-10 00 -06 09 11 -02 -16 5880
12 71 -26 01
23
-14 14 -04 -06 16 -22 06 -02 -11 01
7593
13 79
-24 04 -02 -12 26 18 02 0 8 -01 09 05
00 14
8352
14 21 -01 -17 -21 24
19 -43 -13
01
-17
-11 11 01
-03 4669
15 53
26 02
-25
-16 -04 09 -18 OJ -02
04 08 21 01
5326
16 82 -47 05 -02 06 01
-13
- 11 -02 -06 -03
02 06 -11
9499
17
61 -26 08 -04 02
31
01
-13 -09 12 -28 13 -02 -11 6915
18
77
19 -17
08
-09 23 0 3 -03
11 OJ
01 -08 -10 02
7570
19 72 -38
02 -17 -13 -03 -03
-04 00 16 08
-03 -13
-04
7638
20 50
35 33
19
16 28 -01 -09 - 14 26 18
-05 -07
10
7667
*Decimal points omi t ted
-'1
0
I
~
CASES A B
21 49 17
22 74 -32
23
88 -21
24
67
26
25 76 -09
26 34 54
27
62 11
28 51 51
29 52
4o
30 50
42
31 55
56
32
64 -14
33 57
- 40
34 77
-24
35
89 -39
TABLE XI (continued)
UNROTATED FACTOR LOADINGS AND COMMUNALITIES
FACTORS
C D E F G H I J K L
56 -16 05 -02 -10 01 14
27
08 08
10 -10 -04 -09 27
00 -11 -08 -17
02
-01 00 -12 03
08 -14 09 15 -17 01
o4
15 -12 00
-19 17 03 -02 OJ
20
-14 -08 -07 -04 16
25 07 -01 22 07
-13
24 -01 -07 -10 11 -20 03 -20 -09
-19 15 -13 -10 02 -11 -14
-15
06 11
-37
09 -24 01 -15 -12 -04 16 06 -04
-05 -29
20 -15
06
-13 -02 11 -06 19
21 19
00 -26 00 01 -02 -17 -12 02
27 - 32 -25 -07 -07 -06 -02 -11 16 -09
- 36
04
35
14 00 08
-09
11 02
13
·-03 -17 -14
14
-06
35
02 02
-07
12
-12 -08 15
08
05 05 -09
00 03 -26
-03
10 -03 -15 -10 04
-15
06
09 11
r
-01
06
-06
25
-02
-17
-09
-04
-27
-10
-13
07
-10
07
02
N
h2
-12 7410
-01 8041
03 924o
05 7272
-19
8011
-08 6331
-04 5620
05 7963
-05 7·276
19 6645
25 9929
26 8189
-07 7150
01 7868
08 '1.0432
""
.....
teen dimensions, with regard to the configuration of
thirty-five variables, so that the coordinate axes have
meaning. In rotating the axes, an atte mpt was made to
get positive manifold and simple structure While still
retaining a mutually orthogonal relationship.
72
The method of rotation used in this st dy is the
graphical method wherein one plane is rotated at a time.
The actual technique used here was one developed by W. S.
Zimmerman ( 75), which makes use of the principle of pro
J ection and eliminate s the necessity of computln new co
ordinates after each rotation . The final rotated factor
loadings are shovm in Table XII.
2
2
The reader's attention is called to he fact that
in Table XII the factor communality is, in two instances,
above 1.00. Although this a y be illo ical, it does hap
pen occasionally. In the present study , thee abn rmally
high communalitie s occur for cases 10 a nd 35. The corre
lation between these two variables is .929 , and the proba
ble reason for the communaliti s above 1 . 00 is h· t the
correlation coefficient is too hi h . A pparently , the cor
relation incluced error v riance a el l as true variance.
Fortunately, this discrepancy can be i nored in th solu tion of the problem.
TABLE XII
FINAL ROTATED FACTOR MATRIX*
-
F A C T O R S
-- --- --- - -~ ------~--- - - - -- -
CASES
A B C D E F G H I J K L M N
h2
1 -07 33
08 24 18 -08 19
o4 22
23
22 30
01
19 5302
2 00
30 01 -02 16 00 01
38 43 -10 02
-09
18 11 5085
3
50
42 01 18
27 20 05 -04 31 07 02 10
03 15 7107
4
-26
-05 -33
-06 -19 -26 ..-03 -10 -28 -21 -24 -15
... 11
-08 518J
5
-26
-33
-11 18 -26 -05 30 -05 34 26 -01 19
24
31 7567
6 24 12 24 60
-OJ 17 -02 10
25 19 OJ 07 -02 05 6)71
7
-07
00
13
20 19 10 40
05
10
OJ
18
-09
01 43 5068
8 34 23 07 10 30
18 30 12 00 02
05 30 11 24 5728
9
09 37 -03 03
04
17 12 22 00 46
29 15 -01 23 6113
10 63
46
32 09 13 31 -07
28
13 -08 22 13
o4
-01 1.0056
11 11 05 32 2.5 -43
11 40 04 11
-07
14 01 11 11
5990
12 54 15 29 -11 03
40
05
21 18 06 20 21 06
17 7704
13 32 45
35
-14 -02 31
11 28 OJ 18 14 28 -01 29 8495
14 -10 -08 -13 -08 42
35
08 12 11 -01 05
26
23
00 4946
15
07 15
24 20 14 -08 12 03 OJ
28 12 45 -06 26
5337
16 49 47 23 -03 34 38
10 20 06 -06 16
32
-04 -01 9617
17
06 41
34 -08 15
41
13
11 -08 02
JO
28 06 -10 7021
18 21 20 24
13
01
39
17 17
20 45 21 24 12 22
7756
19
43 52
26 -12 26 22 00 11 01 22 00 22
05 01 7649
20 -08
19. 45
32.
00 41 26 -14 01 14.
8
,-09 -0,5 10
J.5 7655
'
-
'
*Deci mal points om i tted
'3
\.,J
CASES
A B C
21 -02 23
66
22
39
58
22
23 34
53
38
24
19
01
35
25 45 36 20
26 04
-09 03
27
42 10 09
28 10
-05
06
29 -08 27 1.5
30
20 01
25
31
08 00
32
32
04
51
-14
33
-02 03 31
34 34 54 -01
35
58 55
21
...
TABLE XII
(continued)
FINAL ROTATED FACTOR MATRIX
F A C T O R S
D E F G H I J
16
39
00 16
-03
08 10
-01 13 05 12 16
-07
01
04 12 29 -02 12 14
23
44 14
19
04 41 00
19
05
08
05 30
41
07
33
53
04 18
13
00 01
35
26 -01 17 12 10 00
19
4.5 -04 24 -02 02 10 60
26 26 -08 25 -03 17 28
44 12 -01 01 01 14
07
19 32 -08 05 -13 -02 47
20 00 44 10 42 12 04
17 21 06 20 36 -07
48
04
19 37
22 18 08 20
18 21 34 -04 30 -05
02
K L
-09 -02
37
24
29 32
10 22
01 18
29 -11
17 32
04
JO
09 25
31 -02
07 15
01 26
19 02
17 15
04
25
M
00
-02
08
-01
06
29
25
19
47
25
16
15
20
-11
09
N h2
16 ?4J6
11 8084
05
9237 1
25 7348
07 7964
08
6517
13 562J
14
7959
14
7252
42 6708
67 996J
16
8331
21
6995
16
7982
10 1.051
-'1
~
CHAPTER V
INTERPRETATION OF THE FACTORS
A. GENERAL INTRODUCTION
The most important and most interesting part of a
factor-analytic study is the in erpretation of the factors
that have been extracted from the correlation matrix. Be
fore going into this aspect of the problem, it is necessary
to delve into some of the back round material, including
the aims and orientation of the investigator. It will be
recalled from the introductory chapter of this disserta
tion that the writer referred to the study of personality
as having proceeded along two relatively discrete lines of
development -- the clinical and the experimental. It was
believed that the MMPI was an instrument which combined
the two, since it utilized a normative approach in study
ing personality devlan s. The test is scored in term of
scales which describe the abnormal personality but which
have only limited applioab lity to the less deviant in-
dividual. On the other hand, many psycholo is s believe
that the normal and the bnormal differ only quantitatively,
which is to say that the -ame trait contin um can e u ed
to describe them both. The qu tion, therefore, is:
11
\fuat are these basic traits?" In the factor analys s ,
76
fourteen such traits were indicated. It seems logical
that these factors should be named in accordance with
hypotheses that will help effect a rapprochement between
the experimental and clinical methods of personality study.
If at all possible, and if consistent with the data, these
factors should be given names which can with equal facil
ity be used to describe all individuals regardless of
diagnosis.
Fortunately, Cattelt, in hi book, Description and
Measurement of Perso~ality (11), summarizes a consider
able amount of research in the field of personality, and
he lists a dozen traits, the existence of which he con
siders to be most convincingly established by the use of
ratings, questionnaires, tests, and clinical practice.
In addition, he ha s a list of nuclear clusters (11, pp .
246-267) and a catalogue of clinical trait factors (11,
pp. 355-368). Becaus e it is important that the names
used are in accord with the previous work in the field
of personality, these lists will be used as a backlog,
or reference list, of hypotheses to be considered in nam
ing the factors derived from the MMPI .
Insofar as the actual mechanics of interpreting
factors are concerned, the first problem that one faces
is to decide what vall of the factor loadin is to be
J
77
considered significant. Thuretone (64, p . 79) offers
the following advice:
A projection or factor loading of .20 accounts
for only four per cent of the total variance of a
test. We have not regarded a projection as signi
ficant in namin factors unless it is as large as
.40. The naming of a factor cannot be made with
confidence unless the projections are as large as
.50 or .60 so that the factor accounts for a fourth
or a third of the variance of a test. Confidence
in naming a factor is also determined by the number of
tests that have significant projections of .40 or
higher on the factor.
This is a somewhat rigorous criterion for use with
personality teats, and in the present research it was
modified slightly. The loadin s were divided in o four
categories: (1) .40 or above, very significant; (2) .30
to .4o, significant; (J} .20 to .JO, sli htly significant;
and (4} .oo to .20, not s1 nificant .
In attempting to interpret a factor, the investi
gator tried to formulate his hypotheses on the basis of
those cases with very si nificant loadin a and then
checked to see whether the hypotheses held for those
cases with significant loadin s. Finally, those cases
with zero loadin s were checked to make sure that they
did not have the trait in u stion.
The en ire stalt i s a en i nto consideration
while lookin for the co m on elements involved un 11 an
idea for a factor name emerges. Thurstone (65, p. 338)
describes the process in these words:
In making interpretations for a new factor that
has been identified factorially we distinguish, in
the spontaneity for our approval, between interpret
ations that seem vague or far-fetched and those oc
casional incisive interpretations that 'click,' as
78
it were, and by which the factorial results snap into
a meaningful whole. Thie is only to say that the
scientific interpretation of experiments is as sub
Jeotive now as it always has been. To demand that fac
torial interpretations shall be objective is as ab
surd as to demand that the interpretation of a physical
experiment shall be objective. It never is .
Guilford (33) discusse the proces of factor
naming as fol l ows:
The namin (of factors) depends upon features
that the clusterin measures seem to h vein common
and that are unique to them . Attaohin a meanin ful
label facilitat s communication and systematic think
ing. Any label or definition of a factor should be
regarded as a hypothesis, in the same manner that any
trait name is a hypothesis .
Both Thurstone and Guilford are referring to the
R-analysis meaning, of course, that test data constitutes
the clusters from wh ch individual factors are derived.
Tests are relatively objective instruments with whi ch to
work; they have definite content and form, and one would
need certain specific abilities to as s these tests.
Even so, both a lthors stress that naming factors is sub
jective, and the name is to ere arded as a hypothesi
and not as a proven concb1sion . These statements are even
more apropos hen in erpretin a -analysis factor. In
79
the latter case, instead of dealing with relatively objec
tive tests, the variables are the individuals who took the
test. The problem of deciding what personality trait a
particular group of human beings have in common is a very
difficult one, and the results must necessarily be re
garded as tentative.
Considerable pains were taken to be as critical
as possible in deciding upon appropriate nomenclature for
the factors. The steps taken will be outlined in detail
and the necessary data made avail ble so that other in
vestigators may be able to check the results. The MMPI
profiles (which may be reconstructed from Tables XIV and
XV) were interpreted so as to arrive at a personality
description by the usual techniques of profile analysis.
In addition, the various profiles were looked up in the
Atlas {37) in order to obtain additional data and to in
crease the number of case s from which the interpretations
were made.
Now that the ·eneral methodology has been dis
cussed, we will turn to the specific problem of naming
each factor.
B. INTERPRETATION OF EAC FACTOR
Factor A - Strength of Ego Defenses, bipolar
Case No .
10
35
12
3
16
25
19
27
Factor
Loading
. 63
. 58
.54
.,o
. 9
. 45
.4J
. 42
Case No .
22
8
23
34
13
4
5
-
Factor
Loading
. 39
. 34
. 34
. 34
. 32
- . 26
-. 26
80
Thirteen of the thirty-five cases have significant
p ositive loadings on this factor , and two cases have
negative loadin s .
The outstanding common feature of this entire clus te r is that the profiles with high positive loadings in va r iably have high Kand low L scores . On the other hand,
the cases with negative loadings h ve low Kand high F
s cores . The clinical scales are all very varied , so the
i nterpretation mu t be based on the validity s cales .
Clinically, patients with high positive loadings
on this factor have relatively strong ego defenses. They
appear to be fi htin hard to han on to reality a.n~
normalcy . Usually they hav some insight into their dis
orders in that they rec ~nize that theie is omethin wrong
with them. At any rate, they attemp to sort t e te t
•
81
items carefully and with consideration. Although they may
be seriously 111, the prognosis for a temporary remission
is excellent. The · ego may be said to have sufficient re
siliency to snap back to normal in the supporting hospital
a tmoaphere.
Turning to the other portion of the continuum, high
1
negative loadings characterize those subJects with poor
contact with reality, lack of insight, gross ccentricity
and/or carele s ne sin taking the test .
In view of the above, Factor A is considered to
be a bipolar fact or. The positive end of the continuum
is described in term of strong ego defenses, r cuperative
capability, and good judgment. On the negative end are
weak ego defenses , poo r contac with reality , gross ec centr city, lack of Judgment , and/or carelessne sin t aking
the test .
Factor B - Test-takin Attitude , bipolar
Case No .
22
35
34
19
23
32
16
10
Fact or
Loading
. 58
• 5.5
. 54
. 52
. 51
. 51
.47
.46
Factor
Case No . Loading
13 . 45
3
. 42
17
. 41
9 . 37
25 . 36
1 .33
2 . JO
5 --33
82
Sixteen cases have high loadings on this factor.
One has a significant negative loading which means that
a bipolar factor is involved. Over half of these cases
also have high loadings on Factor A. Nevertheless, the
two traits are independent, but they can act together in
a syndrome. All diagnostic categories are represented,
and the clinical scales do not appear to have anything
in com. men.
In terms of validity scales , those profiles that
have a high loadin on Factor B invariably have high K
and low F scor es . There i s also a tendency for the L
scores to be somewhat hi her t han in Factor A. By analyz
ing those cases with a hi h loadin on Factor A and a zero
loading on Factor B, e find that in eve r y instance the
T-score for F 1s hi her than for K. This means that the
patients with h i h loading s on Factor B admit to less bi
zarre material; i . e., they are m ore guarded . Those cases
with a high l oading on Band a zero loadin on A have T
scores on F and K that are a proxima t ely equal.
The conclusion reached is that while Factor A mea
sures the stren th of e o defenses in general, Factor B
measure s the test-takin at 1 ude . Thu we see tl1at
while Factor A and Bare two independent tr its, t hey
can and often do co-v ry.
The subjects who achieve a high positive loading
on this factor have a guarded and evasive test-taking
attitude and/or may be "faking good." A high negative
loading indicates that the subject is lacking in defenses ,
ie "faking bad,
11
or is turning in a completely invalid
test.
Factor C - Asocial Impulsivity
Factor Factor
Case No . Loading Case
1 0 • Loading
21 .66
17 .34
20 .45 10
.32
23 .38 11 .32
13 .35 31 .32
24
. 35
33
. 31
4
-.33
There are ten case s witl1 high positive loadings
on this factor. Case o. 4 has a fairly high negative
loading, but this is probably due more to the mechanics
of the rotation process. Case o . 4 was reflected, and
since it stands alone, it is be ieved that it cannot be
used to deter ine th trait . Factor C will, ther fore,
be considered as unipolar, in pite of the one negative
loading.
A majority of the cases which have high loadings
on thi factor have been diagnosed as manic or impulse
I
84
neurotic. Case No. 21, with a loading of .66, has a Pd
score of 100; Case No. 20, the next highest with a load
ing of .45, has a Pd score of 95. All the other cases
that are highly loaded on this factor have Pd scores
which are among the top three peaks on the profile. In
addition, those cases with zero loadin s invariably have
low Pd scores. Anparently, impulsive activity of an aso
cial nature is the common personality trait shared by
those cases with high loadings on Factor c.
Factor D - Anxiety and ension
Case No.
6
26
28
Factor
Loading
.60
. 53
. 45
Case o.
24
30
20
Factor
Loading
.44
.44
.32
All of the ix cases r th hi h loadings on this
factor have hi h peaks on the D c le. Three of the cases
have been d a nos d a anxiety reactions , and anxiety
ems o be the ou standin common trait .
lelsh ( 69) hr. orked out at chnique for ass ssing
the e ree of an 1 revealed b the I profile . He
label this an anxiet. index . The lat er ind.ex is not
.....,
as
based upon factorial analysis. In order to check on the
present interpretation of Factor D, Welsh ' s anxiety in
dex was computed for the six cases in qu stion . In five
of these six cases, a high Welsh index wa obtained in
dicating the exis ence of pronounced anxiety. Interest
ingly enough, the sixth case, No . 24 , had a low index
in seemin contradiction with the factorial evidence .
However, inspection of he re ords reveals that this par
ticular patien han been dia gnosed as a victi of anxiety
neuro i. Thi su h the fa or a l approach to
the detection of anxiety may be more en i ive t han the
Welsh index. At a ny rate , Factor D wa in e reted in
terms of anxiety and tension .
Factor E - M asterful Self-assertivene s~ Ver u
Ob~equious Self-di~para ement, binolar
Ca e o.
14
21
16
Factor
Loa in
. 42
. 39
• 4
Case o.
Jl
8
11
Factor
Loading
. 32
• 0
-.43
This is a b i olar factor rangi in loadin fr om
~.42 to -.43. In xaminin h refiles , th outs and-
86
ing characteristics appear to be low Hs and Hy scores for
the positive end of the continuum. Diagnostically, the
cases that have significant loadings on this factor, both
positive and negative, are non-neurotic cases. The only
possible exception is case No. 31 which had been diag
nosed as a depressive reaction bu which had no other
neurotic signs.
Clinically, the patients with high positive load
ing on this fac or are characterized y self-assertion,
stubbornness, and -egotis. On the negative end of the
continuum, the patients are described as being over in
hibited and self-dispara in . There is a tendency for
the illness to be acute rather than chronic, and to be
the result of stron environ ntal pressure .
Factor F - Hypomania and Dominance
Case o.
32
17
20
12
18
16
Factor
Loadin
• 4 ,4
. 41
. 41
. 40
.39
. 37
Case No .
34
14
35
10
13
Fae or
Loadi g
.37
. 35
.34
.31
. 31
87
The definition of this factor 1s related to the
diagnostic category of hypomania . Five of the six manic
patient s have high positive loadings , and all three de pressive patients have low negative loading . Also hav
i ng h i gh loadings on this factor are two impulse neurotics
with acting- out tendencies and two conversion hysterics
who have acted out in terms of their own body symbolism.
It is interesting to note that the profiles of the a bove
cases do not all h ve high Ma scores . However , in read
ing the Atlas for er onality descr ption of individ al
with similar profiles , one encounters frequent references
to such phrases as irres onsible, h adstron, hard to dis
cipline , and no frustration tolerance . The factor is not
simply hypomania , but it includes the complem entary trait
of dominance . This is in comple e agreement w th Cattell
who labels hi Factor E, "Dominance (hypomania) vs . Sub missiveness . "
Factor G - Introversion ,
Somatic Concern, and Re ressive Behavior
Factor Factor
Case No . Loading Case o • Loading
7
• 40
5
. 30
11 . 40 8 . 30
25 . 30
88
Four of the five patients with significant loadings
on this factor are schizophrenics; the other has been
diagnosed as an anxiety neurotic.
In interpreting the factor, it is not the schizo
phrenia per~ that is the most important element, be
cause there are other psychotics who do not have high load
ings on thi factor. The common dimension seems to be a
loss of object relationships, accompanied by a turning in-
ward of the libidinal ener y, wi h resultin omatic con-
cern and regressive infantile behavior . Obviously, this is
more than simple introversion; it is the extreme of intro
version with tendencies oward p sychotic with rawal.
Factor H - arcissism and Dependency
Case o.
32
24
25
Factor
Loadi
. 42
. 41
.41
Case o .
2
33
35
Factor
Loading
.38
. 36
.JO
Four of the six case s with significant loading
are neurotics with eith anx ety or conver ion reac-
tion. One of he remainin ca es is the key, and the
other has been diagno sed as hav n an acute psychosis
with somatization reacti on and reality distortion but
89
without hallucinations. From an analysis of the profiles,
one can infer that the common element is some sort of a
neurotic mechanism.
It is interesting to note that the highest peak
on the profile for cq,se No. 3.5 (the key) is on Hy which
may be interpreted as being indicative of narcissism and
immaturity. One may speculate that the psychological
stereotype of a normal person is one who is somewhat
dependent or at least not completely independent.
After referring to the Atlas, it becomes clear that
the trait which these patients have in common is a nar
cissistic personality structure with strong dependency
needs.
Factor I - Autistic and Paranoid Tendencies
Case No.
2
5
3
Factor
Loading
.4J
.34
.31
Case No.
6
1
Factor
Loading
.25
.22
The only cases that have any significant, or
slightly significant, loadings (above .20) on this fac
tor have been diagnosed as paranoid schizophrenics; how
ever, this includes only five of the eight cases that
have been so diagnosed. One of the thre remaining cases,
90
No. 4, cannot be accounted for, except possibly to dis
count its importance because the loadings were reflected
during the rotation process, or becau e of the very high
F score which may have invalidated the record. At any
rate, case No. 4 will be considered a miss. The other
two cases, N os. 7 and 8, also have zero loadings on this
factor, but they both show a decided dip on the Pa scale.
One may assume that they were not quite as paranoid as
the other cases . Further substantiation of this a ssump tion can be had by referring o cases os . 9, 10, and 11.
These ca ses were all dia nosed as unclassified s chizo
phrenics (withou t paranoid elements), and hey too have
zero loadings on this factor. The obvious conclusion
is that Factor I must be rela ed to the paranoid tenden
cies in the p ychosis and not to the dia nosis of para
noid schizophrenia per se .
Factor J - Depression
Factor
Case o . Loadin
28
13
31
9
.60
.48
. 47
. 46
Case o.
18
26
25
Factor
Loading
.45
. 5
. 33
91
The profiles of the patients having high loadings
on this factor are varied. The interpretation was based
solely upon the descriptions of simtlar profiles found
in the Atlas. In most of these descriptions there is a
recurring theme of fatigue, depression, and some hypo
chondriacal concern. In the realm of interpersonal re
lationships, one finds evidence of strong feelin s of
inferiority and some withdrawal. Many patients with
similar profiles obtained from the Atlas were uinida
and gave histories of self-mutilation. The factor is
named "depression " since thi word seems to carry with
it a connotation that is co pa ible with the total con
stellation of the above-mentioned ehavioral traits.
Factor K - ot Specifiable
Factor
Case No . Loading
22 .37
JO .31
17 .JO
There are only three cases with significant load ins on this factor, nd he hi h t loadin i below
.40. I is mos l ble ton m h fac or wi h any de r ee
of confidence, so for all practical pur os sit must be
considered a residual.
Factor L - Emotional Reactivity
Factor
Case No. Loading
15 .45
16 .32
23 .32
27 .32
Factor
Case No. Loading
1 .JO
8 .JO
28 .JO
92
Of the seven cases with significant loadings on this
factor, cases Nos. 15 and 16 have been diagnosed as manic,
No. 23 as an impulse neurotic, and Nos. 27 and 28 as anx
iety reactions. From the profiles, one gets the impres
sion that this factor seems to combine certain elements
in the D, Pd, and Ma scales. The com mon elem ent seems
to be a demoralizing general emotionality . he case
descriptions in the Atlas help to confirm this hypothesis.
In spite of being in different diagnostic categories, all
the patient show a lack of emotional control. Their be
havior is described in terms of irritability, crying
spells, temper tantrums, and violent and abusive activity.
The factor name i s emotional reactivity, or simply emo
tionality.
Factor M - Level of Aspiration, Ambition
Case No .
29
Factor
Loading
Factor
Case No . Loading
26 • 29
27 • 25
JO • 25
5 . 24
13 . 23
93
Case No . 29 is the only one that has a very signi
ficant loading on this factor. The other oases have only
slightly significant loadin s, and these are not much help
in making an interpretation. On the other hand, more than
one case is needed in order to determine co mon elements .
Actually, case No . 29 refers to a rather unusual
individual, and this may nrovide a clue to the trai be
ing represented by Factor M . From his social history ,
we learn that he wa extremely ambitious . In th army
he "'rorked hi ra.r u from a private to a c ,ptain, and
felt that he could hold bigger and bett r Jobs than he ~ras
offered. In th four years of post-war civilian life
preceding hospitali zation, he had from fifteen to twenty
jobs. He always uit a jo when he was doing vell in
his work, and the reason h e gave was that he felt he
94
wasn't getting ahead. When he was unable to achieve his
ambitious goals, he felt depressed, and this led to
drink and eventually hospitalization.
Because this record is unique among the 35 cases
included in the study, it is believed that Factor M may
be a factor having to do with level of aspiration and am
bition. Thie hypothesis must be regarded as being even
more tentative than the others, because it is based upon
one case.
Factor - Obsessional
Case No.
31
7
30
Rumination
Factor
Loading
.67
.43
.42
Factor
Case No. Loading
20 .35
5 .31
Although the diagnoses on these cases are differ
ent, the profiles have certain similarities; in every
profile D, Pt, and Sc are above .70 T-score. The Atlas
provides additional clues, for it fr equently describes
patients with similar profiles as being rigid, sullen,
withdrawn, restless, nervous, constantly worried, etc.
The factor name that best seem s to describe this group
of traits is obsessional rumination. This is not meant
95
to imply a compulsive neurosis. As a matter of fact, the
trait comes closest to describing a schizoid personality
and is compatible with the high Sc score on the profiles.
C. DISCUSSION
Table XIII lists and describes each factor in sum
mary form. Ordinarily, a factor-analytic study ends with
the interpretation of the factors and a statement about
the relationship of these factors to previous studies.
The present study is somewhat of a ione er work. It is
a Q-analysie, and the factor names differ from those
found by the previously mentioned investigators. It is
desirable, therefore, to present material showing that
the interpretations made by this investigator are reli
able, valid, and have clinical usefulness. This is an
ambitious task and must be handled separately in specially
designed research studies. Such validation is beyond the
scope of this di~sertation; however , pilot st dies on
these topic are reported in Appendices A and B. The
results will be summari zed here and the implications
discussed.
First, let us consider the meaning of reliability
and validity when dealin with factor names. The inter
pretation of factors is a subjective process. To say
Factor
TABLE XIII
SUMMARY OF FACTOR NAMES
Description
A Bipolar factor: Strength of ego defenses .
➔ Attempts to exercise control and censor
ship of thoughts and behavior. Because
96
of the ego's r ecuperative capacities, prog nosis for remission is good.
- Has poor contact with reality, lack of
judgment, ross eccentricity, and/or care
leasnesa in takin test .
B Bipolar factor: Test- taking attitude.
+ Indicates a
f ak:ing good
arded and evasive attitude,
- Indicates vulnerability , an inability to
protect oneself, or a deliberate atte p t
to fake bad.
C Asocial impul ivity .
D Anxiety and tension
E Bipolar factor:
+ Masterful self-a s ertiven ss .
- Obsequiou self- dis aragement .
F Hypomanla and dominance.
G Introversion , somatic concern , re ressive
H
behavior, tendency to ard psychosis , and
lo s of object rel tionships .
arcissi and dependency .
I Autig ic and pa1anoid tendencies .
Factor
J
K
L
M
N
TABLE XIII (continued)
SUMMARY OF FACTOR NAMES
Description
Depression, feelin s of fatigue, and
inferiority.
Not specifiable.
Emotional reactivity.
Level of aspiration, ambition.
Obsessional rumination.
97
98
that these interpretations are reliable implies that other
investigations will come to the same conclusion, 1. e.,
use the same names, and hence the conclusions appear to
be correct or valid.
A pilot study designed to check the reliability and
validity of the interpretations is reported in Appendix
A. Two Judges, in addition to the original investigator,
were asked to interpret each factor. They were given a
list of the significant factor loadings and a list of
nineteen possible trait names. They were asked to match
the factor with a name from that list or to make up a
name. Their results are summarized in Table XVI. All
three Judges were in complete agreement in the naming of
six of the fourteen factors, and in partial agreement on
five factors; the two Judges disagreed with the names
applied by the original investi ator in only three cases.
The judges worked independently, and it may be that if
they had had the opportunity to discuss their hypotheses
in joint conference, the conclusions would have been more
strikingly unanimous. Even now, the results are much
better than chance, and one mu~t conclude that the factor
names are reliable and can be used with a fair degree or
confidence.
A study designed to supply information on the ap-
99
pl1cab111ty of the factorial traits to clinical data is
reported in Appendix B. The personality of every sub
ject who took part in this study can be described in
terms of the factor loadings. For example, if he has a
high loading on Factors C and D, he would be described
as being anxious, impulsive, and lacking in moral re
straints. Certainly, it is important to find out if this
factorial description is a true and accurate statement
of the patient•s behavior. To check this, the clinical
record folder of each patient was obtained from the hos
pital files. These records contain reports written by
the psychiatrist, psychologist, social worker, and
nurses. The factorial description was compared with the
behavioral description in the clinical record folder by
means of a specially prepared rating scale. Because of
the restricted nature of these records, only the original
investigator was able to make the comparisons, so the
reliability of the ratings cannot be determined. However,
an attempt was made to be as objective a s possible. The
rules for making the ratings were explained and illustra
ted. A sample case history and the ratings made from
it are included in Ap endix B. The results of the study
are summarized in Table XVIII in Appendix B. If, on the
basis of this material, it is agreed that the ratin s
100
were fairly and objectively made, then it follows that
the factor loadings provide an accurate description of
personality. The agreement between the factorial and
clinical data clearly illustrates the usefulness of the
factor analysis approach to personality testing.
CHAPTER VI
SUMMARY AND CONCLUSIONS
A. SUMMARY
The purpose of this study was to investigate and
isolate the personality variables measured by the MMPI.
The MMPI was described and the history of its
development reviewed. Evidence for the reliability and
validity of the test was presented. This was designed to
show that the test was reliable, and that it has been
used successfully to diagnose patients into one of the
three major classifications: psychosis, neurosis, and
psychopathic personality. Precisely because the MMPI
is a worthwhile test, it was chosen as the subject for
the factor analysis. Since the test can be used success
fully for diagnostic purposes, it is important to dis
cover exactly what personality variables it measures.
Eight studies involvin factor analyses of the MMPI
were reviewed. They all used the same technique of inter-
correlating the seal
•
The factors obtained from these
studies were interpreted as being syndrom es of traits
or diagnostic categories. The results of th se statis
tical studies provide additional evidence that the MMPI
can be used to make a differential diagnosis. However,
102
it is a well-known fact that two people with the same
diagnostic label may have different personalities. The
previous factor-analytic studies are limited in that they
do not supply 1nforma.tion about these basic personality
variables.
In order to is olate the personality factors that
are measured by the~ PI, the te twas factor-analyzed
by using the Q-technique. The variables in this study
were thirty-four subjects who wer careful y selected
so as to represent various diagnostic categories. A thir
ty fifth case was added; this was the key or "norm al"
way of classifyin the statements . Tetrachoric correla
tion coefficients were computed for every variable with
every other variable. These correlation coefficients were
modified by equating the means and standard deviation of
the different distributions in accordance with the tech
nique developed by J. w. Holley. The resultant coeffi
cients were arranged in the form of a correlation m atrix
and factor-analyzed. Fourteen principle axes were ex
tracted, and these were r otated orthogonally to achieve
simple structure. The new factor loadings were com
puted and the results interpreted.
lOJ
· B. CONCLUSIONS
The factors obtained are interpreted as follows:
Factor A. Strength of ego defenses, bipolar
Factor B. Test-takin attitude, bipolar
Factor c. Asocial impulsivity
Factor D. Anxiety and tension
Factor E. Self-assertiveness vs. self-
dieparage ent, bipolar
Factor F. Hypomania and dominance
Factor G . Introversion
Factor H. Narcissism and dependency
Factor r. Autistic and paranoid tendencies
Factor J . Depre ssion
Factor K. Not specifiable
Factor L. Emotional reactivity
Factor~- Level of aspiration, ambition
Factor N. 0 sessional rumination
Two other psychologiots independently matched
the factors with pos nible names, and agreed, at least
partially, with the original investigator in all but three
cases . It is ap arent, then, that the s e factor names
are fairly reliable. Factorial description for most of
the patients used in this study were compared with the
descriptions obtained from the clinical records, and the
104
two were found to be in substantial agreement. This ie
further evidence that the factor names are meaningful
and clinically useful.
In the first chapter of this dissertation the fol
lowing four aims were enumerated:
(1) To determine more exactly what personality
variableo are being measured by the items which com
prise the MMPI.
(2) To determ ine whether the MMPI variables as
presently scored (Rs, D, Hy, etc.) are sup orted by t hi
statistical analysis.
(J} To lay a foundation for the future development
of new scales for measurin any additional variables
that are isolated by the study.
(4) To determine whether a classic textbook example
of paranoid schizophrenia or manic depressive psychosis
differs significantl in his MMPI performance from the
usual hospital patient who has been similarly diagnosed.
Each statement will be examined in turn to see how well
the aims have been realized.
The major purpose of the study ha s been to isolate
the variables measured by the 1 I. This has been real
ized. Fourteen factors have been extracted and all but
one interpreted.
105
Regarding the question as to whether the MMPI ecales
are supported by statistical evidence, it is impossible
to give a simple yes or no answer. The purely diagnostic
scales, such as Hypochondriaeis, Hysteria, Schizophrenia,
etc. do not emerge as factors in the present study.
Others more closely related to personality variables do
appear as factors, but in so doing, thy lose their diag
nostic character. For exa ple, Factor F has been called
hypoma.nia, but in he section discua in this factor, it
is clearly stated that Factor Fis not identical with a
manic psychosis. On the contrary, thi f ctor must be
interpreted a.s that personality t rait revealed by a great
deal of activity and by attempts to dominate the social
situation. A mani c patient may have this trait, but so
may many neurotics and normals. This writer, therefore,
concludes that alt ough the I is scored in terms of
diagnostic scales, it is actual y measuring personality
variables. The test interpreter, using profile analysis
techniques, combines these personality traits into syn
dromes and arrives at a diagnosis .
This study has isolated many new variables, and
it should be possible to develop scales consisting of
:MMPI item which measure these traits. Unfortunately,
the Q-analysis technique does not use sufficient subjects
106
to make an item analysis possible. It must be left for
future research to develop new MMPI scales designed to
measure these variables.
The reader will recall the question raised by Dr.
Grayson as to whether the clear-cut case or textbook
example of paranoid schizo hrenia is unlike the usual
paranoid schizophrenics that m ake up the bulk of the hos
pital population. An examination of the factor loadings
reveals that in no instance do the clear-cut cases clus
ter~ together as a unit. It is necessary to conclude that
there is no significant difference between the two groups,
at least insofar as their WI records are concerned.
This writer has strong convictions that the science
of psychology is a unified discipline; in spite of the
fact that, for pedagogical and practical reasons, it is
broken up i nto various sub-branches, such as clinical , ex
perimental, and psychometrics. This is not an original
thought, but it is a true one, and it follows that the
experimentally substantiated conclusions of one branch
should hold with equal validity in another. Psychology
has reached a stage in its development where it becomes
necessary to examine the basic premises of each branch
and eliminate contradictions. In his study, the writer
has examined th differences among some of the variables
107
used by the clinical and experimental approaches to the
study of personality and has demonstrated that they were
in essential agreement. He has also shown that the fac
tor-analytic technique, specifically Q-analyeis, can ad
vantageously be applied to the study of clinical psy
chology. It is hoped that this work will add some small
amount of additional light on the basic personality con
cepts being used in psychology as a whole.
More specifically, and with regard to the MMPI
in particular, this study has supplied a statistical
foundation to the method of profile analysis. In the
previous studies on t he PI, the investigators were
primarily concerned with the diagnostic validity of the
test. The early factor-analytic studies supplied a
statistical foundation for this usa ~e. Lately many
psychologists, notably Hathawa, Meehl, and Gough, have
used the :MMPI as a means for deriving a personality de
scription of the subject. This study, which demonstrates
that the test does in fact measure basic personality
traits, supplies the necessary statistical evidence for
the continued use of this technique.
f
BIBLIOGRAPHY
BIBLIOGRAPHY
1. Aaronson, B. s., and Welsh, G. s. The MMPI as diag
nostic differentiator: a reply to Rubin. J. con
sult. Psychol., 1950, 14, 324-325.
2. Abrams, E. N. Comparative factor-analytic study of
normal and neurotic veterans. A statistical in
vestigation of the interrelationships of intellec
tual and emotional factors as disclosed in the
PMA examination and the MMPI. Unpublished Ph.D.
dissertation, N. Y. u., 1950.
J. Anderson, A. L., and Hanvik, L. J. The psychometric
localization of brain lesions; the differential
effect of frontal and parietal lesions on MMPI
profiles. l· ell~. Psychol., 1950, 16, 177-180.
4. Benton, A. L. The MMPI in clinical practice.
nerv. ment. Dis., 1945, 102, 416-420.
J.
5. Benton, A. L., and Probst, K. A. A comparison of psy
chiatric ratings with MMPI scores. J. abnorm . soc.
Psi~hol., 1946, 41, 75-78.
6. Burros, o. K. The nineteen fqrty mental measurement
yearbook. Highland Park, N. J.: Mental Measure
ments Yearbook Co., 1941.
7. Burton, A. The use of psychometric and projective
tests in clinical ·psychology. ~- Psycp._ol., 1949,
28, 451-456.
8. Cameron, N. The 12.sx_c~ology 9-f behavior disorders.
New York: Houghton-Mifflin Company, 1947.
9. Capwell, D. F. Personality patterns of adolescent
girls: II, delinquents and non-delinquents. J.
aEpl. Psyo~ol., 1945, 29, 289-297.
10. Cattell, R. B. Interpretation of the twelve primary
personality traits. Character and Persona~it~,
1944, 13, 55-91.
11. Cattell, R. B.
personal_i_ty.
The description and measurement of
New-York: World Book Company, 1~6.
12.
110
Cattell, R. B. On the disuse and misuse of P, Q,
Qs, and O techniques in clinical psychology. ~
olin. Peyohol., 151, 7, 203-215.
lJ. Cattell, R. B., and Luborsky, L. P-technique demon
strated as a new clinical method for determining
personality and symptom structure. ~- gen.
Psrchol., 1950, 42, 3-24.
14. Coleman, J.C. Abnormal behav~or and modern life .
15.
New York: Scott, Foresman, and Company, 1950.
Cook, E. B., and Wherry, R. J.
MMPI and aptitude test data.
1950, 34, 260-266.
A factor analysis of
J. appl. Psychol.,
16. Cottle, W. c. Card versus booklet form of the . I.
~- appl. Psychol., 1950 , 34, 255-259-
17. Cottle, W. c. A factorial study of the Multiphasic ,
Kuder, and Bell inventorie s using a population of
adult males. Psyc11_ometr1:ka, 1950, 15 , 25-l}? .
18. Drake, L. E. A social I .E. scale for the MMPI . J.
apP.1:• Psychol., 1946, 30, 51-54.
19. Ellis, A. The validity of personality questionnaires .
Psy:_chol. Bul~, 1946, 43 , J85- '~4o .
20. Ellis, A., and Conrad, H. S. The validity of person ality inventories in military practice . Psychol.
Bull., 1948, 45, 385-426.
21.
22.
Gough, H. G. Diagnostic patterns on the MM I .
clin. Ps~cho~., 1946, 2, 23-37.
Gough, H. G. Simulated patterns on the M MPI .
abnorm. soc. Psychol., 1947, 42, 215-225 .
J .
-
J .
2J. Gough, H. G. A new dimension of status: I . Develop
ment of a personality scale . Amer. social . Rev.,
1948, 13, 401-409.
24. Gough, H. G. A new dimension of status: II. Rela
tionship of the St scale to other variables .
Amer. sociol. Rev., 1948, 13, 534-537 .
111
25. Gough, H. G. A new dimension of status: III. Dis
crepancies between St scale and "objective status."
Amer. social. Rev., 1949, 14, 275-281.
-
26. Gough, H. G. Factors relating to the academic a
chievement of high school students. J. educ.
Psy9hol., 1949, 40, 65-78.
27.
28.
29.
30.
31.
32.
33.
Gough, H. G.
the MMPI .
413.
The F minus K dissimulation index on
J. consult. Psychol., J.950, 14 , 408-
Gough, H. G. A scale for measuring dominance.
abnorm. soc. sycpol., in press .
Gough, H. G. A scale for measuring prejudice.
abnorm. soc. . chol., in press .
Gough, H. G., cClosky, ., and Meehl, • E.
scale for measuring social responsibility.
abnor~. soc. sychol., 1952, 47, 73-81.
J.
J.
A
J.
Guilford, J. F. Ps~chometric ethods . New ork:
McGraw-Hill, 193; second edition in preparation.
Guilford, J.P. Fundam ental statistics in ps~chol ogy and education. New York: McGraw-Hill, 1950.
Guilford, J. •
ment program .
Factor analysis in a test develop
Ps~chol. Rev., 1948, 55 , 79-94.
J4. Guthrie, G. M . Six MPI dia nostic rofile patterns.
J. Psyc~ol., 1950, 30, 317-323.
35. Hanvik, L. J.
back pain .
354.
MM I profiles in patients with low
J. consult. ychol., 1951, 15, 350-
36.
37.
38.
Hathaway , S. R. A codin system for I nrofile
classificati on . _. £.Q!l~Ult. sychol., 1947, 11,
334-337.
Hathaway, S. R., and Meehl, _ . E. An tlas for the
clinical use of the . ..v1PI . iinneapolis : Univ. Minn.
Hathaway, S. R., and cKinely, • C. M_anual for the
fl'I . rew rork: P yoholo ical Corporation, 1951,
(Re.vis ion) •
I
112
39. Holley, J. w. A psychometric method of item scoring
for inverted factor technique. Unpublished paper,
1950.
40. Holzberg, J. De, and Alessi, S. Reliability of the
shortened~ ~I. J. consult. fsiohol., 1949, 13,
288-292.
41. Factor analysis:
Chicago, Ill.:
Holzinger, K. J., and Harmon, H. H.
A synthesis of factorial methods.
Univ. ·of Chica.go Press, 1941.
42. Hotelling, H. Analysis of complex statistical vari
ables into principal components. J. educ. Psychol.
1933, 24, 417-441, 498-520.
43. Hovey, H. G. Somatizat ·on and other neurotic reac
tions and the MMPI profiles . J. clin. Psychol.,
1949, 5, 153-159- -
44. Hunt, H.F., and others. A study of the differential
diagnostic Gfficiency of the M MPI . J. consult .
Ps~chol., 1948, 12, 331- JJ6 .
45. Landis, c., Zubin , L., and Katz, S. The validity of
certain questionnaires which proport to measure
neurotic tendencies. l· aERl· Psychol., 1934,
18, 343-356.
46. Leverenz, Maj., c. W. MMPI: An evaluation of its
usefulness in the psychiatric service of a station
hospital. War Medicine , 1943, 4, 618-629.
47. Little, J . W. An analysi of the PI . Unpublished
Master's thesis, niv . of North Carolina, 1949.
48. Louttit, C •• , and Browne, C. G. The use of psy
chometric instruments in psychological clinics.
~- consult. Es~chol., 1947, 11, 49-54.
49. Meehl, P . E. The dynamics of structural personality
tests. ~- clin. Psichol., 1945, 1, 296-JOJ.
50. iieehl, P . E. Profile analysis of he _ I in differ
ential diagnosis. ~- appl. Psycqol., 1946, 30,
517-524.
113
51. Meehl, P. E. Configu.ral scoring. J. consult.
Peyohol., 1950, 14, 165-171.
52. Michael, J. C., and Buhler, C. Experiences with per
sonality testing in a neuropsychiatric department
of a public general hospital. Dis. Nerv. Syet.,
1945, 6, 205-211.
53. Moore, T. V., Stafford, J. W., and Hsu, En Hsi. Ob
verse analysis of personalit r. J. Personality,
1947, 16, 11-48. -
54. Morris, W.W. A preliminary evaluation of the "diag
nostic" scales of the vMPI. J. clin. Psychol.,
1947, 3, 370-374.
55. Portenier, L. G. Personality tests in a un versity
guidance program. J. educ. Psyc~ol., 1948, 39,
479-487.
56. Rubin, H. The ll'I as a diagnostic aid in a veterans
hospital .. J . consult. _sychol., 1948, 12, 251
2.54.
57. Rubin, H. A note on "Rep ly to Rubin." J. c onsult.
58.
59.
Ps~chol., 1950, 14, 327-328.
Schmidt, Lt. Col., H. o. Test profiles as a diag
nostic aid: The M_ PI. l· appl. Pslchol., 1945,
29, 115-131.
Stafford, J. W~, and su, En Hsi. The super-factor
of persons. l· Psi~hol., 1947, 24, 63-70.
60. Staner, R. Ps~cholog,y of Eers~~a~it~. New York:
McGraw-Hill, 1937•
61. Stephenson, W. Correlating persons instead of tests.
Cba:racter a nd Personal ity, 1935, 4, 17.
62. Stephenson, i . Th e inverted factor technique. Brit.
l· Psychol., 1936, 26, J34-J61.
63. Stout, M. An analysis of the structure of the M¥ J.PI.
Unpublished Ma ter•s thesis, Penn. State Colle e,
1949.
114
64. Thurstone, L. L. Primar~ mental abilities. Ps~cho
metrics Monograph #1, 1938, Chicago, Ill: Univ.
of Chicago Press.
65. Thurstone, L. L. MultiE~e factor analysis. Chicago,
Ill.: Univ. of Chio go Press, 1947.
66. Tucker, L. R. Simplified punched card methods in
factor analysis. Proceedin s of the Research
Forum Endicott, N •• , 1946, International Busi
ness Machines Corporation.
67. Tyler, F. T. A factorial analysis of fifteen MMPI
scales. ~. consult. Psychol., 1951, 15, 451-457.
68. Van Vorst, R. B. An evaluation of test performance
of a group of psychopathic delinquents . Ps~cho~.
Bull., 1943, 50 , 583 . (abstract.)
69. Welsh, G. An anxiety index and an internationaliza
tion ratio for the MMPI. J. consu~. Psychol.,
1952, 16, 65-75.
70. Wheeler, W. M., Little, K. B., and Lehner, F. J. The
internal structure of the n>I. J. consult.
Psychol., 1951, 15, 134-142.
71. Wiener, D •• Differences between the individual
and group forms of the MMPI . ~- consult. Psychol.,
1947, 11, 104-106.
72. Wiener, D. N. Subtle and obvious keys for the MMPI.
J. consult. Ps~chol., 1948, 12, 164-170.
73. Winne, J. F. The factorial composition of normal
and neur tic responses to an a daptation of the
MMPI. Unpublished Ph.D. dissertation, Univ. Penn.,
1950.
74. Winne, J. F. A scale of neuroticism: An adaptation
of the MMPI. • clin. Psychol., 1951, 7, 117-122.
75. Zimmerman, W. S. A simple gra hie method for orthogo
nal rotation of axes. Psichometrika, 1946, 11,
51-56.
APPENDIX A
THE RELIABILITY AND VALIDITY OF THE FACTOR NAMES
APPENDIX A
In all factor-analytic studies, it le the respon
sibility of the principle investigator to name, and to
give psychological meaning to, the factors he has ex
tracted. Should another psychologist disagree with the
names, it is his privilege to develop his own hypotheses
and to rename the factors accordingly. In order for him
.to be able to do this, he must have access to the orig
inal data, in this case the factor loadings and profiles.
The faotor loadings are found in Table XII, page 73.
The raw scores for each of the thirty-five cases are
found in Table XIV, and the T-scoree are in Table xv.
The profiles can be reconstructed from these scores.
Although it is customary for an investigator to
name the factors he has extracted, it is certainly de
sirable to assess the reliability and validity of the
factor-naming hypotheses by having a number of investi
gators do this work independently, and then seeing how
well the results agree. There are a number of ways in
which this problem could have been formulated. Perhaps
the simplest and most obvious way was for the other in
vestigators to start from the beginning, with the pro
files and factor loadings, and independently name the
factors. There are also obvious disadvantages to this
117
TABLE XIV
MMPI RAW SCORES CORRECTED FORK
S c a 1 e s
Subject ? L F K Hs D Hy Pd Mf Pa Pt Sc Ma
1 0
5
11
9 17 30 20
23 27 11
30 29
24
2 0
7
21
17 30 38 36
33
21 20
J6 49 24
3 5
1 8 16
13
24
19 23 23
12
27 32 16
4
3
8
55
13 27 29
24 28
33
20 46 60 24
5
6 2 21
5
23 32 30 31
41 24 46 61 28
6 0 2 10
17 JO 36
4o 28
36 17
44
50 21
7
0
5
20 14
17 32
28
36 43 17 43
40
28
✓
8 6 2 8 11
9 13 12
25
)9 10
29 25 21
9
21
3
2 11
17
14 22 28 22 10
27 30 18
10 1 8 4 28
15 21
2.5
26
23
10 28
30 17
11 1 6 22
13 2.5 36 34
35 39
18 41 46 26
12 0 2
7
22 18
15 27 Jl J2 10
27 34 22
13
0
5 3
17 15 17 22
27 JO 7
21 22 21
14
30
4
16 11 11 18 11 22 26 22
32 4J 23
15 1
3 9
12 11 28
19
33 32 17
39 39 25
16 1
5
1 18
9 17 20 24 21 8 18 22 20
17 0
7 9 19 17 19
24
29 29 10
27
33 29
18 0 2 6 10 20 21 26
29 31
12
27 19 23
19
0 6
9
20 14 22
23 36 17
9
28 28 20
20 0 4
12 13 19 JO
35 38 47 14
36
41
30
21 0
3 15 12 8
29 22 40
36 12 JO
35 27
22 0
9
4
21
13 25 17
33
17
9 29 26
25
23 0 4
5
22 14 28 21
33
22 10
27 26
17
24 0 2 6 16 20
27 28
31
26
17 34 31 21
25
0
3
4
15 18
2.5 26 26 14 6
29 29 21
26 0
5
11 12 28
. 39 35
31 34 18
51
48
23
27 1 2 8 14
15
26
29 20 24 10
J6 36 27
28 2 1 8
7
22 29
33
24
31
14 J4
33
17
29 1 1 11
7
10
31
18
33
24
15 36 36
26
JO 0 1 14 12 16
39 25 29
35
15
41
34 17
31 1 0 10
5 5
29 12 24
36
10
33 3.5
22
32
0 2 6
17 32
~a
34
24
27
9 33
31
20
33
0
5
10 10 20 28
35
28 12
35
32 31
34 1
5 J
16 22
19 26
25 22 4 22 22 22
35
0 0 0 21 11 16 21 16 18 8 24 24 12
118
TABLE XV
MMPI STANDARD SCORES CORRECTED FORK
-
Sub- S c a 1 e s
Ject ? L F K Ha D Hy Pd Mf Pa Pt Sc Ma
1 50
.53
68 44
65 82 56
60
63
59
64
63
68
2 .50
60 80
59
98 101 86
83 51 8.5 77 101 68
3
.50
40 62
57
54 68
55
60
55
62
58 69
48
LJ,
50 63
80
51 90
80 64
71 74 85 97 120 68
5
50 43
80
36
80
87
75 79 90 97 97 120 78
6 50 43
66
59
98 96
93 71
80
76 93 103
60
7 50
53
80
.53
65 87 71 90 94 76 91 101 78
8
50 4J 62 48
L}4
41 42 64
47 56
62
55
60
9
.50
46 48 48
65
44 60 71
53
56 58 65
53
10
50 63
53 79 59
60
65 67
55
.56
60
65 50
11 50 56
80
51 85 96 82 88 86
79
87 96 73
12 50 43 60 68
67
46
69 79 73 56 58
73 63
13 50
53
50 59 59
51
60 69 69 47
46
50
60
14
50 50
80 48 49
53
40
57
61
91 69 90
65
15 50
46 64
49 49
77 55
83
73
76 83
82 70
16 50
53
46 61 44
51 56
62
51 .50
40
.50 58
17 50
60 64 62
65 .56
64 74 67 56 58 71
81
18 .50 43 58
46
72
60
67 74 71
62
58
44
65
19 50 56
64· 64
57
63
62 90 43
53
60 61
58
20
50 50 70 51 70 82 84
95
102 67
77
86 83
21 50
46
78 49 41 80 60 100 80 62 64
74
75
22 50
66
53
66
54 70 51 83 43
53
62
57
70
23 50 50
55
68
57 77 58 83
53
56 58 57
50
24
50 43 58 57 72 75
71 79
61 76
73
67 60
25 50
46
53 55
67 70 67 67
37
44
62 63
60
26
50
53
68 49 93 104 84
79
76 79 107 99 6.5
27 50 43 62
53 59
72 73 53 57
56 77 76
75
28 50
40 62 40
77
80 80 62 71 67
73 71 50
29 50
40 68 40 47
84
53
83 57
70
77
76
73
30 50
40
76 49 62 104
65
74 78 70
7 73
50
31 50 36
66
36 34 80 42 62 80
56 71 74 63
32 50 43 58 .59 103 70 82 62 63
53
71 67 58
33
50
53
66 46
72
68 71
88
65 62
75
69
86
34 50
53 50 57 77 56 7
64
.53
38
48
.50 63
35 .50 36
44 66 49
48
.58 43 45 .50 52
.53
38
119 l
method, for it is highly unlikely that different people
with different orientations would use the same terms in
describing a personality trait. Should different terms
be used, how could one judge whether the traits being de
scribed are really identical?
In order to resolve the above-mentioned difficulty
and to shorten the amount of time required to name the
factors, it was decided to formulate the work in terms of
a matching problem . The sychologists who participated
in this study were ,1ven the following material:
(1) A set of fourteen 3
11
x 5
11
cards. Each card
contained the identification let er for factors A through
N and the significant factor loadings in each case.
(2) A set of nineteen 3
11
x 5" cards. Each card
contained the name and description of a ersonality trait.
These included tbP. thirteen traits which the original in
vestigator used in namin t he factorR, and six other
trait name s obtained from Cattell's (11) l i ste.
1
1
The additional six tra.it names are:
(a) ypochondriacal , taciturn retroversion vs.
interest in future, eloquence; bi olar
(b) Severity of the di order
(c) Alcoholism
(d) Self-sufficiency, independence an sociability
(e) Rhathymia, adventurous vs . withdrawn schizo-
thymia, obstructiveness
(f) General ne1r tic m aladjustment
120
(3) A set of 35 MMPI profiles.
(4) The Atlas of the MMPI to which the investi
gator could refer.
The Judges doing the matching were familiar with
this particula,r study, with :MMPI profile-analysis tech
niques, and with factor analysis in general. They were in
structed to st8rt with the card for a particular factor,
and then to select those cases from the group of profiles
that have sin fic~nt loadin son this factor. The next
step was to determine, from the profile , what personality
trait these case s had in common. If the investigator
wished, he could look up the profiles in he Atlas and in
that way increased the number of cases and get new ideas.
After getting some eneral notions as to what the factor
could be, he was to look through the cards with he fac
tor names, and i~ one of these names descried the trait
he had in m·nd, he was to use t. If not, he could name
the factor as he saw fit. It was su ested hat when in
doubt, he was to make a first and second choice. In th.J..8
way, it was made po ible for him to use the ame name
more than once and thus minimize the tendency for the
disagreement to occur in pairs -- the first time by usin
the trait name for the wrong factor, and the second time
for not attach n that name to he correct factor. If
121
the same name could be used more than once, this spurious
element would be eliminated from the matching problem.
The investigator and two psychologists
2
partici
pated, as indicated in Table XVI. It can readily be seen
that there is complete agreement in six of the fourteen
factors, a percentage far greater than chance. One of
the two other judges agreed with the ratings of the origi
nal invectigator in two cases, and in three additional
oases there was agreement on the second choice. This was
summed up as partial agreement in five cases, and dis
agreement in the remaining three. There was one other
point of interest; this concerne factor N, wherein the
two Judges agreed among themselves but disagreed with the
writer.
3
The following is a summary of the results of this
study on the matching of the f ctor names with the pro
files:
(1) For the three judges there is complete agree
ment in naming of six of the fourteen factors. These
2
nr. Edwin Shneidman , who is on he staff of the
Veterans' Administration Hospital, and Dr. s. M. Wesley,
formerly an Associate Professor at the University of South
ern California and now in private practice .were the judges.
3
The writer's hypotheses are defended in the ma.in
body of the dissertation, and even after reviewing the
data, he continues to support his original views.
TABLE XVI
COMPARISON OF THE FACTOR NAMES USED BY THE DIFFERENT JUDGES
Factor Borko
A Ego Defensiveness
B Test-taking
Attitude
C Asocial Impulsiv-
ity
D Anxiety
E
F
G
H
Self-assertive
ness
Hypomania
Introversion
Narcissism
*A= Agreement
I n v e s t i g a t o r s
Shneidman
Ego Defensiveness
Test-taking
Attitude
Asocial Impule1v-
1ty
Depression
Self-assertive
ness
Narcissism
Obsessional Rum
ination or In
troversion
Introversion or
Narcissism
P = Partial Agreement
Wesley Results*
Ego Defensiveness A
:,
Test-taking
Attitude A
Asocial Impuls1v-
1ty A
Asocial Impulsiv-
ity or Anxiety P
Self-assertive-
I
ness
Hypomania
Severity of Dis
order or Para
noid
Anxiety or Emo
tionality
D = Disagreement
A
p
p
p
....
N
N
Factor
I
J
K
L
M
N
- -- -
TABLE XVI (continued)
COMPARISON OF THE FACTOR NAMES USED BY THE DIFFERENT JUDGES
I n v e e t i g a t o r s
Barko Shneidman Wesley
Paranoid Paranoid Paranoid
r:I
Depreseion Alcoholism or Neurotic MaladJust-
Anxiety mentor Nar-
oissiem
Not Specifiable Not Specifiable Not Specifiable
Emotionality Anxiety or De- Obsessional Rumin-
pression ation
-
Level of Aspira- Level of Aspira- Obsessional Rumin-
tion tion ation or Anxiety
Obsessional Rum- Emotionality or Emotionality
1nat1on Anxiety
Results
A
D
A
D
p
D
....
N
y
124
factors are:
(2)
agreement:
(3)
factors:
A. Strength of ego defenses
B. Test-taking attitude
~
C. Asocial 1mpuls1v1ty
E. Masterful self-assertiveness
I. Autistic and paranoid tendencies
K. Not specifiable
For the following five factors there was partial
F. Hypomania
M. Level of aspiration
D. Anxiety and tension
G. Introversion
H. Narcissism
There was disagreement in naming the following
J. Depr~ession
L. Emotional reactivity
N. Obsessional rumination
APPENDIX B
✓
A QUALITATIVE STUDY OF THE APPLICABILITY OF THE
FACTORIAL TRAITS TO CLINICAL DATA
APPENDIX B
Having decided upon the interpretation of the fac
tors, and having named them on the basis of hypotheses
derived from the interpretation of the MMPI profiles, it
was decided to check the validity of these hypotheses by
reference to the clinical records of each of the patients
who participated in this study. The point is that four
teen variables were isolated by factor analysis. These
variables were interpreted in terms of the MMPI profiles ,
and now it was deemed desirable to find out if the psy
chiatrists and psychologists found the same traits in
the personalities of the patients they described. In this
way a check could be made as to whether the factor loadings
agreed with the psychiatric reports. For example, if a
particular patient has a high loading on Factor C, Impul
sivity, it should be possible to check the clinical records
and see if his behavior was in fact described as impulsive.
The question is raised as to whether traits found by fac
tor analysis actually show up clinically.
The first step was for the invest! ator to get the
patients• clinical folders from the hosuital files. These
folders contain the reports of the neuropsychiatric board
on the admisoion and discharge of the patient, the reports
of the social worker and the psychologist, the daily prog-
12?
ress notes, the description of the patient's behavior on
the ward, and various laboratory and medical reports.
The second step wa to prepare rating sheets for
1
each subject. On these sheets the twelve named factors
were listed in descending order, the one with the highest
loading appearing first. To the right of each factor was
a four-point rating scale: Stated Agreement, Inferred
Agreement, Inferred Disagreement, Stated Disagreement.
If one were to design a good statistical study to
validate the hypotheses by which the factors were given
psycholo ical meaning, a fairly large number of clinicians
would ha e to go through th records of the patien sin
dependently and rate the amount of agreement between the
psychiatric descriptions and the factorially determined
traits. Thie project is a study in itself and is not part
of the present investigation which aim aim ly at determin-
ing those traits measured by the PI . However, the ap-
plicability and validity of the factor traits do deserve
some consi eration, and for that reason it was decided to
review the records qualitatively .
In order to assure the objective use of the rating
scales, the following more or less arbitrary rules were
1
K and M not included becau e they could not be
named with any degree of confidence.
128
adopted:
{l) All loadings above .20 are to be considered sig
nificant. In other words, if a particular individual has
a loading greater than .20 on a certain factor, he is pre sumed to possess that trait in sufficient degree that men
tion be made of i t in his clinical records.
(2) All loadings of .20 and below are considered to
be not significant. This means that if a particular in
dividual has a loading of .20 or less on a certain fac
tor, he is pre urned to lack that rait, and consequently
no mention should be made of it in his clinical records.
(3) Stated a r emen is to be scored:
(4)
(a) "When the patient has a significant loading
on that t rai and the reports in the clinical
record file specifically mention that he poss
esses the trait in question.
(b) When the loading is not si nif cant and the
recor s specifically sate that the patient does
not posses the trait or posoesses it only to
an insignif · cant de ree.
Inferred agreement
(a) When thee 1
hould be scored :
a si nificant loading and the
general enor of the report is such that the
pa ient can reasonably be assumed to pos
0
ess the
129
trait, although it is not mentioned specifically.
(b) 'When the loading is not significant, and the
general impression from the clinical records sup
ports the conclusion that the patient does not
possess the trait in question or possesses it
only to a slight extent.
(5) Inferred disagreement should be scored:
(a) When the factor loadings are significant and
the general t enor of the records implies that the
patient lacks that trait.
{b) When the loadin is not significant, but it
can reasonably be inferred from the record that
the patient does possess that personality trait.
{6) Stated disagree ent should be scored:
{a) When the factor loading is significant and
it is specifically stated in the records that
the patient lacks that trait.
(b) When the factor loading is not significant
and it is specif oally stated in the clinical
records that the patient does possess that trait .
It is readily apparent that these rules, although
they attempt to make the ratin objective, leave ample
room for subjective jud ment by th rater. This diffi-
lJO
culty can usually oe overcome, as was stated previously,
by having several judges make the ratings, and thereby
establish their reliability. It is regrettable that this
could not be done, but the hospital records are restricted,
and the task is time-consuming. As a result, only the
original investigator did the ratings.
Before presenting the composite rating scale which
ie the result of this sub-study, the writer deemed it de
sirable to present the eignifican material from the clini
cal record folder of a particular patient, and hie rating
scale, so that the reader can decide whether he agrees
with that ratin and be better able to evaluate the writer's
concl11s1ons on this topic. Case o . 10 was selected at
random. The complete psychiatric and sychological reports
for this case are reproduced here with the identifying data
removed.
REPORT OF l EUROPSYCHIATRIC Ex.M-1INATION
ON ADMISSION
PATIENT'S NAME: Case No. 10
FAMILY I TORY : The patient 's father died at 36 years of
age after suetainin injuries in an automobile accident.
The patient was nine at this time . He claimed that he
did not now his father very rel becau
0
e he was away most
lJl
of the time. He described the events surrounding the
death 1n this manner: " fuen he died everybody was upset.
I began to cry and mother told me to shut up. I heard a
psalm then, and always remembered the psalm word for
word.
11
His mother was a :tiard worker, inclined to worry
over household affairs, and very serious. The patient is
said to be very much like his mother. He felt that he
was loved by her and that there was no partiality shown
any of the children. His mother remarried a man seven
years her Junior. He was described as being very likeable ,
easy in his manner, and very accepting of the children.
The patient had frequently gone on drinkin sprees with
the stepfather. There were four siblings, three brothers
and one half-sister. He was unable to describe his sib
lings adequately, but felt that they were all very fine
people. His stepfather occasionally drank to excess, but
although irresponsible when drunk, he usually remained
out of trouble.
PERSONAL HISTORY: The patient wa s he oldest of five
children. There was nothing unusual about his develop-
ment or health during infancy . He was born in ___ ,
on a farm and lived fiv e miles from s chool ; he waled this
distance every da to and from chool. He interrupted his
schooling in the first year of high school for lack of
132
funds. After school he worked on the farm and delivered
papers; although he had no very close friendships, he
stated his adjustments were easy, and he seemed to be
getting along satisfactorily. During this period he moved
quite frequently with his family. Although he fought fre
quently with his brothers, he claimed that this was not
unusual and most of the time he was friendly with them.
He was described as a nervous child who always bit his
nails. The most unusual aspect of his ch. ildhood and pu
berty dealt with his psychosexual development.
While in his early teens, he was introduced to hom04-
sexual practices, and he remained an active homosexual for
three or four years.
At the end of this peri d he joined the CCC. Fol
lowing his work with this group, he returned home and
began school; at the age of 19 he enlisted in the Navy
for "educational. opportunities." His adjustment in the
Navy, at best, was only fair. He had few friends, drank
moderately, and frequently visited prostitutes. He began
to go with one girl steadily, whom he "used experimentally
for knowledge purposes." She married someone else. In
1943 he met another girl and became engaged to her. W hen
she expressed her love f or him, he became very suspicious
of her and began to torment her in order to test her;
133
"I wanted to see how much she could take." He began to
aggravate her by interrupting her. She often became very
angry, but contained herself. (The patient smiled inap
propriately at this point.) He continued in this manner
until he had developed the habit of constantly arousing
her anger. "She lifted my ego because she was willing
to take so much from me." On one occasion she tried to
seduce him but he refused to have a sexual experience with
her; finally she married someone else .
MILITARY HISTORY: This patient had three enlistments in
the Navy . All told, he served nine years in the military
service; his most recent enlistment was in January, 1946,
and he was CD D'd in June, 1946. He served as an electri
cian 2/c and saw duty only in the Southwest Pacific Theatre.
There was little combat.
PRESENT ILLNESS: While he was in the Pacific Theatre,
his girl friend became pregnant by another man; when he
returned home he learned of this . She wished to marry
him but he in sisted that she marry the other m an, and she
did. On his third reenlistment he brooded con s i derably
over this. He e an to experiment with himself in order
to determine whether he could live without sleep . He
went without sleep for three days; then sleep overcame him.
134
Soon after this he thought he could do better and went
without sleep for six days. He didn't explain to anyone
what he was doing, but felt that he could gain a great
deal of knowledge and endurance in this way. He had the
idea that,
11
Life was Just a series of experiences and ex
periments.
11
At this time he had a strange dream. "I had
a dream I was with her; I saw a black cloud coming; I
tried to get her to run with me; she didn't. The black
cloud came over us and I woke up . It was at this time
that I suddenly busted loose." He developed the omnipo
tent idea that he knew everythin hat was going on . He
began to feel flashes passin through him and had to ex
pend a tremendous amount of energy. It was at this time
that he picked up his bicycle and smashed it a ainst the
wall. He began to drink considerably after this. He was
hospitalized in ___ Hospital for excessive drinking.
Later he was transferred, dischar ed for nervousness , and
received 100% disability . From 1944 to 1946 he drank ex
cessively. During th a period of time he had very few
friends, frequently visited prostitutes, and spent a good
deal of money . When he was confined to neuropsychiatric
hospitals in the avy, he was always i n l ocked wards . He
was not sure why he was confined to the hospital, but be
lieved that it was due to somethin he said to a woman.
135
He couldn't enlarge upon this. Once while on pass, he
met a pregnant girl who had been deserted. He immediately
talked her into marrying him. The marriage, however, was
annulled when his medical officer found out about this,
because he was considered to be incompetent. "I felt good
about her accepting my proposal, but they put me in a
looked ward. I suppose I was in a locked ward because I
said something that frightened the doctors. I can't re
member what I said."
PHYSICAL EXAMINATION, NEUROLOGICAL EXAMINATION, AND LABOR
ATORY REPORTS: All essentially ne ative.
MENTAL EXAMINATION: Attitude and eneral behavior: He
was very pleasant and affable but evasive in dealing with
personal questions regarding sex and his previous hospit
alizations. He objected to bein interviewed at first be
cause he said al l the information had reviously been
given. However, after a few minutes, he cooperated with
examiners much be ter . He bit his nails frequently ,
smiled inappropriately, spoke slowly, and often became
loot in circumstantiality . In his emotional reactions he
showed a discrepancy between ideation and emotional re
sponse. He rarely smiled, and when he did, it was in ap-
propriate .
136
~ental trends and content of thought; There was
a p:reat deal of vague, bizarre ideation. He felt that
life was a series of experiences and experiments. He
nointed this out in his ideas relating to cex and sleep .
He tried to regulate his life in such a manner in 1943
so that he did not have to sleep at all. When his girl
friend expressed her love for him , he began to test her
to see how dee ly she as in love with him. He did this
by constantly in err ptin her or accusin he of goin --:- - out with other men . He final ly drove her to despair.
His comment concerning this was, "My e o wa lifted when
I aw that she could ake so much from m e." In the latter
part of 1943, while talking with someone, "I suddenly
busted loose.
11
At this time he had a maniacal episode ,
unable to be controlled, an dashed his bicycle against
the wall. He had the sense of knowing what was going on
about him. From this period until his release from the
Navy in 1946 he drank excessively but evidenced no form
of alcoholic p sychosis . I wa felt th t t he more the
exam er pre ed h m for information, the more bizarre
hi ideation became . The patient showed no insi ht . He
sta ed thR. he Vi as no a ho asexual , although he had had
homosexual exuerienc o er eri od of four years . In
all his re ponse he ried to e cooperative , but was
137
usually evasive and circumstantial. At times he had
omnipotent ideas.
Seneorium and mental grasp: He appeared fairly
oriented, was moderately confused with regard to events
in the immediate past, and often mixed up dates. He was
unable to state why he was hospitalized. His insight is
lacking; Judgment le poor.
DIAGNOSIS: Schizophrenic reaction, mixed type, m anifested
by mildly paranoid non-systematized delusions and bizarre
ideation.
PSYCHOLOGICAL REPORT
PATIENT'S NAME: Case No. 10
TESTS ADMINISTERED:
1
Wechsler-Bellevue I
Rorschach
TAT
BEHAVIOR AD MANNER: Althou h the patient , when asked
the few personal uestions prior to the Wecholer admin
istration, displayed some annoyance because he had al
r eady iven information oevera ti es before , after a
1
The PI was administered at a later date and
was not available when th s re ort was writ en .
1~
brief explanation, he became very cooperative and remained
so throughout the testing. Prior to the administration of
the Rorschach, he said that he had taken this test pre
viously. Before responding to the cards, he looked at
them from all angles, even lookin · at the reverse side.
Hie conoepto were usually carefully formed in mind before
making his response. He was very critical of both Ror
schach and TAT cards. \vhile on the surface the p tient
appeared to be cooperative, hi responses in general were
evasive. Contrary to the instructions on the TAT of giv
ing stories, the patient limited himself to rather detailed
descriptions of the cards, particularly of the minute de
tails in the background.
SUMMARY AND CONCLUSIONS: Intellectually, the patient is
within the bright-normal range (I = 114). His poor cul
tural and educational environment and also his vocation
may account in part for the ide difference between hie
verbal (104) and his performance (121) r .. •s, but the
relatively large scatter would lend support to the hy
pothesis that some intellectual impairment 1c present,
probably as a result of his anxiety.
This man has a basically ohizoid personality, but
the teats do not show any evidence of a frank psychotic
139
break. He has avoided psychonis by developing rigidity.
This defense is relatively weak. Contact with reality
has been maintained by avoiding emotional involvement
and by substituting a detached, intellectual relationship
with the environment. This is evidenced by his general
evasiveness, hie critical att tude of the test materials ,
and his cone rn with impertinent, minute detail . It is as
if he were desperately hanging on to reality by concen
trati g on the objective, factual element
0
of th envir
onment, avoidi any i nterpre ation of the . His adjust
ment appears to be of a chronic, sub- psychotic nature.
Under pres ur e, however, when his ri idity proves o be
inadequate, his behavior may ap roach psychotic propor
tions.
On a more superficial level, the patient likes to
impres other as being a happy, el -adjusted person .
When this is questioned , he responds in characteristic,
non-commi -tal manner, being reluctan to reveal his in~·
adequacy, insecurity, and lack of self-c onf dence.
There is evidence that this man has never progressed
beyond an infantile sexual level of development . An in
terpre ation of he projective material indicates that
he is anall y or entedo Psycholo ical y, he is not capable
of mature sexual expression . From all indications, there
140
is no tendency to regress, but rather to remain fixated
at a pre-genital level. -
In view of the chronic nature of the disorder and
the absence of any signs of a serious attemp t to achieve
a mature adjustment, the prognosis is poor.
These are the reports from the clinical record
folder. In order to illustrate the specific bases for
making the ratings , a brief statement about each trait
or a quotation from one of the reports will be iven where
possible.
Ego defensiveness . Si nificant loadin : Stated
agreement.
"He has avoided sychosis by developing rigia_i ty
• • •
It i as if he were desperately hanging on to reality
by concentrating on the objective , factual elements of the
environment , avoidin a ny interpretat on of them .
11
Test-takin a.ttitud.e - fil!_arded and evasive .
nificant loadin~: Stated a r ement .
"Before respondin to th cards, he looke
from all angles, ~v n lookin at the reverse side .
Sig-
at them
His
concept were usually carefully formed in mind before
making his resp onse. He was very critical of both the
Rorschach and T AT cards. While on the urface the patient
appeared to be cooperative , his responses in gene~a1 were
14!
evasive. Contrary to instructions on the mAT of giving
stories, the patient limited himself to rather detailed
descriptions of the cards, particularly of the minute
details in the background."
Aooc1al impuleivity and moral deficit. Signifi
cant loading: Inferred agreement.
"Once while on pass, he met a pregnant girl who had
been deserted. He immediately talked her into marrying
him.
11
He drank alcoholic beverages excessively. He ad
mits a few yea ·s of active homosexual life . He frequently
visited prostitutes.
Hypomania. Significant loading: Stated a.greement.
"At this time he had a maniacal episode, was unable
II
to be controlled, and dashed hie bicycle against the wall.
Narcissism an dependenci. Si nificant loading:
Inferred agreement.
"When his girl friend £Xpr essed her love for him ,
he began to test her to see how dee ly she was in love
with him. He did this by cons ant l y interrupting her or
accusing her of going out with other en. He finally drove
her to despair. His com ment concerning his wao, 'My ego
was lifted when I saw that sh could take so much from
ma. ' "
Self-a sertiveness.
ferred agreement.
142
on-si nificant loadin : In-
There was no significant loading on this trait and
no information in the report to support a belief that the
patient was self-assertiv·e.
Autistic and naranoid tendencies. Non-significant
loading: Stated d~sagreement.
The diagno is iven to the patient is, "Schizo
phrenic reaction, m xed pe manifested by mildly paranoid,
non-systematized delusions an bizarre ideation.
11
On the
basis of this diagnosis one would expect the loading on
this factor to be hi her.
Demorali zed general emotionalit.
loading: In erred a reement .
on-significant
owhere in the report is there any evidence that
the pat i en has t hi trait .
Anxiety. on- si~nifican loadin: Inferred agree-
ment.
The only place where the word anxiety is mentioned
is in the psycholo ist
1
s report, where it is stated
". . . but the relatively l : 1r e catter would lend support
to the hypo hesi that some intellectual impairment is
pre ent, prob bly a r esult of his anxiety .
11
The psy-
cholo it is a p rently no sure of this nference, and
145
it may be that the anxiety was ma.nifested only in the
taking of this test. At any rate, lt le not supported by
the rest of the behavioral data, and ~o it is felt that
the patient does not have much anxiety.
Obsessional rumination. Non-significant loading:
Inferred agreement.
No information that the patient pos~esses this
trait was available.
Introversion. on-significant loading: Inferred
disagreement .
The patient states that he had no close friends
while at school, and none while in the Navy. The implica
tion are that he was by himself most of the time.
Depressio;11.
1
on-significant loadin : Inferred
agreement.
There is no evidence that the patient ever felt
very depressed.
The rating scale for case No . 10 is found in
Table XVII so that he reader may obtain an overall view.
Twenty- seven of th thirty-four cases were rated
in a manner similar to that descried for case No . 10. It
was impo sibl to obtain the clinica folders for the
other seven cases, ecause ei hr the patients and their
records were tran ferred to another hospital , or they were
...
TABLE XVII
RATING SCALE SHOWING THE AGREEMENT BETWEEN THE FACTOR LOADINGS
AND CLINICAL REPORTS FOR CASE NO. 10
FACTOR RATING SCALE
Inferred
Stated Inferred Dis- Stated Dis-
Label Description Loading Agre_ement. Agreement agreem.ent agreement
A Ego defensiveness .63
x·
B Test-taking attitude .46 X
C
Asocial impulsivity .32 X
F Hypomania .31 X
H Narcissism .28 X
E Self-assertiveness .13
X
I Paranoid .13
L
Emotionality .13 X X
D Anxiety .09 X
N Obsessional rumina-
tion -.01
X
G Introversion -.07
X
J Denression -.08
X
.....
..
...
145
still being treated, in which case their ol1n1cal folders
were unavailable.
A summary sheet of the rating scales is included,
Table XVIII, and it can readily be seen that the trend
1a in the favorable direction. Apparently, the peyohia
triste and psychologists, in describing the patients,
agree fairly well with the description of these patients
as obtained from their factor loadings. It is to be em
phasized that these results are to be considered as a
trend which, as yet, ie unsupported by good statistical
evidence. However, the available information does support
the validity of the hypotheses by which the factors were
named, the names themselves, as well as indicating that
the factors have clinical significance.
TABLE XVIII
SUMlvf.ARY OF THE RATED AGREEMENT BETWEEN THE FACTOR LOADINGS
AND THE CLINICAL REPORTS
Factor Number of Cases
Label Description
A Ego defensiveness
B Test-taking attitude
C
Asocial impulsivity
D
Anxiety
E Self-assertiveness
F
Hypomania
G Introversion
H Narcissism
I Paranoid
J Depression
L
Emotionality
N Obsessional rumination
Stated
Agreement
8
10
7
11
5
7
2
5
5
10
9
4
Inferred Inferred
Agreement Disagreement
16
3
16 1
14 4
7
4
19
3
17 1
18
5
11
9
14 2
5 7
12 4
19
4
Stated
Disagreement
0
0
2
5
0
2
2
2
6
5
2
0
...,
~
Ql
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The relationship of the pretreatment strength of social responses of schizophrenic patients to improvement with electric shock treatment
PDF
The effects of anxiety on the alpha rhythm of the electroencephalogram
PDF
An analysis of the research literature used by American writers in the field of speech
PDF
The isolation of personality traits in the domain of military leadership
PDF
A study of the organic basis for behavorial deviations in school children
PDF
Social functions of the churches in a changing community with special reference to social processes
PDF
The effects of partial reinforcement on a whole human response
PDF
A survey of theatre organization as practiced by a selected group of community theatres in the Los Angeles area
PDF
An experimental study in fluctuations among successive play performances
PDF
Psychological factors associated with aspirations for socio-economic mobility
PDF
Studies on cactus alkaloids and related synthesic [i.e. Synthetic] compounds
PDF
Clothing as a factor in the social status rating of men
PDF
Sociological aspects of training programs in selected industrial organizations in the Los Angeles metropolitan area
PDF
The fisherman and his wife : theme and variations for soprano, narrator, and chamber orchestra. Text from Grimm's Fairy tales
PDF
Some personality variables in pvert female homosexuality
PDF
W-B data as indices of aptitude for law and engineering a contribution to vocational counselling
PDF
Claus Spreckels of California
PDF
The selection of textbooks and other instructional materials in the high schools of California
PDF
The social ecology of delinquency in Los Angeles county: a structural analysis
PDF
An investigation of oxygen-pulse time course as a measure of cardiorespiratory adaptation to exercise
Asset Metadata
Creator
Borko, Harold
(author)
Core Title
A factor-analytic study of the Minnesota multiphasic personality inventory using a transpose matrix (Q-technique)
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Psychology
Degree Conferral Date
1952-06
Publication Date
06/01/1952
Defense Date
06/01/1952
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
factor analysis,Minnesota multiphasic personality inventory,OAI-PMH Harvest,Personality tests
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC112724836
Unique identifier
UC112724836
Identifier
A no.32 (Film) (call number),Ph.D. Ps '52 B734 (call number),etd-BorkoHarold-1952.pdf (filename)
Legacy Identifier
etd-BorkoHarold-1952
Document Type
Dissertation
Format
theses (aat)
Rights
Borko, Harold
Internet Media Type
application/pdf
Type
texts
Source
20230207-usctheses-microfilm-box7
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
factor analysis
Minnesota multiphasic personality inventory
Personality tests