Q ~ This is to certify that the

thesis entitled

S QM 15 R AC 5c H72 M C 0/1"? .1: {.fl 7:» out? (‘ Wurst?

CLm/u/H, STTHLJS, ‘A’J I_W'L’£J7,'Ca‘17"“/VI .2? L9
[7. ; (1 (LE 1.5 ”1‘0 ,’fl-’ .1' ( 44/ 2.0 I'M/(1 c‘ru4
presented by
(tang-#55 T, gulf?"

has been accepted towards fulfillment
of the requirements for

'7. r'u,‘ ;‘
b, t? degreeln prvi, .LUL/

'81 6, k {own/v

Major professor

Date MAY 191 I97]

0-7639

     
         

L I B R ER Y
Michigan Sum
Univemty° ‘

//l/

C- O

ABSTRACT

SOME RORSCHACH CORRELATES OF CHANGE IN CLINICAL STATUS:
AN INVESTIGATION OF EGO REGRESSION IN SCHIZOPHRENIA

BY

Charles Thomas Glatt

The capacity of the Rorschach to reflect test-retest changes in
clinical status was examined from the perspective of the ego regression
theory of psychopathology. Both psychoanalysis and comparative-devel-
0pmental theory hold that changes in.the adequacy of overt functioning
are paralleled by shifts in the developmental status of the ego.

Research thus far has demonstrated that Rorschach responses of
adults with serious psychological disorders resemble in many ways the
responses of young children rather than those of normal adults. The
studies purporting to have confirmed the Operation of ego regression in
psychopathology can be criticized, however, for being based solely upon
cross-sectional data. Much more convincing would be a longitudinal
study demonstrating that as a person.becomes increasingly maladjusted,
his ego functions at a more infantile level, and as he improves, his
ego again functions at a more mature level.

In accord with the hypothesis that a retest Rorschach showing
movement toward or away from a more mature mode of ego functioning indi-
cates concomitant improvement or decrement in overball adjustment, it
was predicted that changes in each of ten scores or indices indicative
of the maturity level of the ego are correlated with change in clinical
status.

Four times during a twenty-month period thirty-five‘gs diagnosedzns
schizophrenic were administered both a psychiatric interview to ascerb
tain clinical status (CS) and the Rorsdhach. Measures of CS change and
of Rorschadh score change were obtained for three separate test-retest
intervals--initial testing to six months, six.months to twelve months,
and twelve months to twenty months--thus providing data for three inde-
pendent replications. The delta regressed score transformation was
used to free each measure of change from the effects of regression er-
ror. Product-moment correlation coefficients were then computed for

Charles T. Glatt

each of the three tests of the ten predictions. Significance of the
findings was determined by combining probabilities from the three tests
of each prediction.

Four of the ten measures (Perseveration Score, Scorability of pro-
tocol, _I:+%, and R) correlated with CS change in accord with prediction.
Further analysis of the relationships between Rorschach score changes
and CS change in terms of £3' medication status produced striking re-
sults: six out of nine scores (Perseveration, g, g, g, and the Fried-
man Develomental Level Scores) correlated with CS change in those §_s
not on medication at any time during the test-retest interval , whereas
but one score (3+9 correlated with CS change in gs remaining on medi-
cation throughout the interval . Although medicated and non-medicated
535 did not differ in the amount or direction of change in either CS or
in any of the Rorschach scores, drug-mediated (:8 change was neverthe-
less found to differ in its dynamic qualities from change not accompanied
by the use of medication. The former condition seems to interfere with
the theoretically congruent relationships existing between test-retest
differences in level of adjustment and in certain Rorschach measures of
ego functioning in §_s not getting medication.

It was therefore concluded that the ego regression theory of psy-
chopathology is valid insofar as it accurately predicts the changes
which will occur in certain aspects of ego- functioning when schizo-
phrenics show clinical status change in the absence of drug treatment.
The results corroborate clinical observations that ego regression is
not a monolithic mechanism, i.e., the various ego functions are not af-
fected uniformly during the course of pathogenesis and recovery, and
there are individual differences as to which functions are affected
most by regression or progression.

SOME RORSCHACH CORRELATES OF CHANGE IN CLINICAL STATUS:

AN INVESTIGATION OF EGO REGRESSION IN SCHIZOPHRENIA

BY

Charles Thomas Glatt

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOROF PHILOSOPHY

Department of Psychology

1971

ACKNOWLEDG‘lENTS

There are a number of persons who have been instrumental in the
development and completion of my dissertation. My thanks go to Drs.
Paul Bakan, Albert Rabin, and Robert Zucker for serving so capably on
my committee, and to fellow students Frank Long and Gary Vandenbos for
their assistance in data collection. A special acknowledgment is ex-
tended to Dr. Bert Karon, my committee chairman. I am indebted to him
not only for his supervision of my research but also for the signifi-
cant impact that he has had upon me in his various roles as teacher,
advisor, colleague, and friend during the years in which I have had the
privilege of knowing him. I also owe much to my wife, Karen, for her
valuable contribution as critic, morale-booster, and typist.

C. 6.

ii

TABLE OF CONTENTS

LISTOFTAHJESOOOOOOOOOIOIOOOO
WMWOOOOOOOOOOOOOOOOO
WOFTHELITERATURE...........

Evaluation of previous research

Stataent of the problem in terms of ego regression theory

STATDIENT OF HYPOTHESIS

mu 0 0 O O O O O O O O O O O O O I O O O 0
Selection of Rorschach variables . . . . .

Summary of specific empirical predictions

subj eats O O O O O O O O O O O O 0 O O O

Tests............
Clinical Status evaluations
Analysisofdata......
RESULTS ............
Analysis of drug effects . .
Simaaryofresults.....
DISCUSSION...........
Conclusions in terms of drug

status
Conclusions regarding the effect

Thedynamicsofchange.....

Implications for future studies

SUMMARY

HSTOFW O O O O O O O O O

of medication

iii

Page

iv

14
28
33

34
39
39
4O

42

50
54
S6
S9
62
63
67
69
'72

LIST OF TABLES

Characteristics of Studies of Rorschach Change as a Func-
tionofClinicalStatusChange.. ........... ..

Swary of Research on Rorschach Score Change as a Function
OfelinicalStat‘JSChame0.000.000.0000...

Rorschach Measures of Ego Developmental Level and Relation-
ShipsmlmpmvedC11nj-C313tat'us 0.0.0.0000...

S‘lbjeCt maraCtaj-Stics O O O O O O O O O O O O O O O O O O
Test-retest Coefficients of Correlation for Rorschach Scores

Clinical Status Change upon Shift from Unscorable to Scor-
able Protocols and from Scorable to Unscorable . . . . . . .

Correlations between Rorschach Score Change and Clinical
sums Crime 0 C O O O O O O O O O O O O O O O O O .A O O 0

Correlations of Rorschach Score Change with Clinical Status
ChangeinAccordancewithDrugStatusofgs . . . . . . . .

Comparison of Drug Status Groups on Number of Predictions
AttairdngSignificanceooooooooooooooooooo

iv

Page

15

16

4O

44

49

SO

52

54

INTRODUCTION

This investigation is directed toward determining Rorschach changes
which accompany clinical status change (i.e., improvement or deteriora-
tion in mental health) from the perspective of the ego regression theory
of psychOpathology.

Although a number of studies have dealt with changes in Rorschach
performance as a function of psychotherapy or of some other form of
significant life experience, relatively few have addressed themselves
to the question of the capacity of the Rorschach test to reflect im-
provement and decline in mental healthnirrespective of how that change
came about.

To clarify the nature of the problem upon which the present study
is focused, it is necessary to distinguish among four types of research:

1. Before and after studies of inkblot performance in persons get-
ting therapy (e.g., Carr, 1949; Hamlin, Berger, & Cuxmnings, 1952; Hosak,
1950; Huench, 1947; Oppenheim 8: Brower, 1948; Rioch, 1949; Watkins, 1949).

2. Test to retest changes in Rorschach behavior among S3 in therapy
as contrasted to changes occurring in §_s either not in therapy or getting
a different type of therapy (e.g., Cadman, Misbach, 8: Brown, 1954; Dudelc,
1970; Haimowitz 8: Haimowitz, 1952; Hamlin 8: Albee, 1948; Krout, Krout, &
Win, 1952; Piotrowski & Schreiber, 1952).

3. Comparison of test-retest Rorschach changes occurring in "suc-
cessful” therapy cases to changes occurring in ”unsuccessful' therapy
cases (e.g., Gaylin, 1966; Halpern, 1940; Kisker, 1942; Lipton& Ceres,
1952; Peterson, 1954; Piotrowski, 1939).

4. Comparison of Rorschach changes in persons showing favorable
change in clinical status to persons not showing improvement, without
regard for the nature of the conditions to which each _S_ was exposed dur-
ing the test-retest interval, i.e., some §_s might receive one kind of
therapy, other _S_s get different kinds of treatment, and some might re-

ceive no therapy at all (e.g., Barry, Blyth, & Albrecht, 1952; Goldman,
1955; Rickers-Ovsiankina, 1954; Zamanslcyaz Goldman, 1960).

1

2

Since most of the instances of significant and lasting change which
come to the attention of the clinician are the result of some type of
psychotherapy, it is not surprising that the majority of studies having
to do with Rorschach correlates of change in level of psychological ad-
justment have focused on such occurrences within the contest of a thera-
peutic situation.

The first two research designs are irrelevant to the concerns of
the present study since they utilize Rorschach change to measure the im-
pact of therapy rather than seeking to relate Rorschach change to change
in the clinical status of the _Ss getting therapy. The major difference
between then is that the second, by providing a no-therapy control group
(or a group getting a different kind of therapy), allows one to discount
the effects of practice on the posttherapy Rorschachs.

The difference between the third and fourth designs may seem trivial
inasmuch as both provide independent measures of clinical status change
to which Rorschach change can then be related. A study employing the
fourth design, however, carries different implications than does a study
of the third type. The fourth type of study, albeit infrequent in use,
enjoys the advantage of enabling broader generalizations to be made
about the dynamics of clinical status change.

There is a potentially serious problem involved in drawing conclu-
sions from the third type of study: one cannot automatically assume
that the Rorschach changes which do occur are the result of the therapy.
They may be due, in part or in whole, to extraneous factors whose ef-
fects weren't controlled. For example, significant life events, such as
alterations in occupation, marital status, health, or social envirorment,
can affect psychological functioning independently of or in interaction
with the influence of therapy, making it extremely difficult to separate
therapy-induced effects from extra-therapy effects . The fourth design
obviates the necessity for doing so; the focus is on change in adjust-
ment level and not on the conditions which brought about the change.

With respect to studies employing the third design, the exercise of
prudence in interpreting results can overcome the inclination to attri-
bute (perhaps incorrectly) change in test behavior to therapy. There is a
second potential pitfall, however, which is more difficult to circumvent.

Although equivalent changes in overt functioning may occur in two

3

given individuals, different personality changes might occur in each
case because of differing treatments to which each person was exposed.
The various therapies may have roughly the same goal , the alleviation of
symptoms, but they employ different means to the same end-e.g.,
strengthening a patient's defenses, altering the intensity of impulses
and affects associated with the symptom, restructuring the ego, changing
pathogenic aspects of the enviror-ent.

Dudek (1970), Kraut et a1. (1952), and Piotrowski and Schreiber
(1952) obtained different posttherapy Rorschach changes along _S_s in psy-
choanalytic therapy than among §_s in supportive or drug therapy} Al-
though not substantiated by these studies (perhaps only because none of
the three attempted to relate the observed Rorschach changes to clinical
status change), it seems reasonable to assume that the specific Ror-
schach changes which correlate with clinical status change also depend
toanextentuponthetypeof treatmenttowhichgsareexposed. For
instance, the gross symptoms of a psychotic depression may be amelio-
rated equally effectively by psychotherapy, ECT, or drugs, yet the con-
comitant effects upon underlying processes, as measured by the Rorschach,
are likely to be different in each case.

Since any one study employing the third design typically involves
but one type of therapy, it would be erroneous to conclude that its
findings generalize to all cases of clinical status change. The fourth
design, when properly executed with a heterogeneous array of treatment
conditions, allows one to isolate the inkblot changes which are the
canon derminator to _a_ny and all cases of clinical changeuif such
canonalities do in fact exist. For these reasons, the present study
aployed the fourth design.

 

lAll three reported Rorschach changes of greater magnitude and fre-
quency in reaction to psychoanalytic therapy. There were also differ-
ences in the quality of change. Psychoanalytic therapy was typically
accompanied by a liberation of drives and affects (as reflected in
increased 31, color usage, and _R). Drug, electroshock ,; and supportive
therapies tended to lead to improved contact with reality (34-1.) but at
the ammo of eXpressiveness and spontaneity.

.‘,-,,i.r

 

REVIEWOFTHELITRATURE

Before reviewing previous research on Rorschach changes concurrent
with clinical status change, three alternative methods for defining
Rorschach change should be described:

1. The use of separate raw scores (e.g., g, 1, E) and ratios
(54, E3E+E)e Chem is determined by comparing a given score's
value upon retest to its initial value.

2. The index, or ”signs" approach. An index consists of a coher-
ent set of theoretically meaningful scores; e.g., a "neurotic" index,
such as the Basic Rorschach Score (Buhler, Buhler, s Lefever, 1948) or
the Hunroe (1945) checklist, is a composite of several component scores,
each of which is believed to indicate maladjustment. For each component,
thepresemeofacertainmnberofresponsescmmtsasa"sign"ofneu-
rosis, or as a "sign"of health. Thus, 3229;4’2 is taken to be a sign
of nature aeotionality whereas Eg< 9; +2 is a sign of maladjustment;

52. 3 is an indication of good adjustment where §< 3 is not. The index
value is obtained by simply counting the amber of signs on which g
qualifies because of his having the I"right" quantity of that score.
Change is determined by subtracting the initial index value from the
retest value.

3. The global evaluation. A judgment is made about the overall
quality of a given Rorschach record. The initial and retest protocols
ofeachgarecoeparedtodeter-inewhichofthetwoindicatesagreater
degreeofpsychological health. Indecidingwhetheragivenghasin-
proved, remained the same, or deteriorated, the clinician's intuitive
skill-a-sharpenad by experienceuis called into action, not his nathaa-
tical prowess.

The first and second analyses measure change in quantitative units;

the third analysis provides a qualitative measure of change.2 «

 

2mg global analysis is what Allport (1962) would call a "half-way
norphogenic" approach to research: the comparison of change over time

is not earpressed in abstract units of behavior (such as scores), yet the
4

5

In addition to the availability of three options for the definition
andneasurenentof chargeinRorschach performance, therearetwocoe-
monly used procedures for relating Rorschach charge to clinical status
charge:

1. Analysis of individual data. A count is made of the number of
gs showing each of the possible combinations of charge (i.e. , Rorschach
score increase, clinical status improve-ant; Rorschach score increase,
clinical status decline; Rorschach decrease, clinical status improvement;
Rorschach decrease, clinical status decline). Chi square is then used
to test for association between directional charge in clinical status
and directional change in a given Rorschach score. Thus, concern is
with the frequency and direction of charge rather than with the amount
of charge.

2. Analysis of group data. Two or more groups are formed on the
basis of inprovalent, no charge, or decline in clinical status, and the
group means for Rorschach score changes are compared. The amount of
change is of interest to the investigator as well as is the direction of
change. Obviously, this method of analysis cannot be used with global
evaluations of Rorschach change.

One of the shortcomings to the use of group data is that the occur-
rencewithinafewésofchangesofextre-enagnitudenaydistortthe
general trend of retest changes for the grwp as a whole. By analyzing
individual data, the influence of atypical _S_s is prevented from becoming
disproportionate.

Studies involm gantitative m 9.53:3. The earliest of the
published accounts is Piotrowski's (1939) investigation of hospitalized
schizophrenics given insulin coma therapy. Unfortunately, there was no
presentation of data amenable to statistical analysis . Only his general
observations were recorded. Clinical status improvenent was fourg to be
associated with an increase, from pre-therapy to post-therapy, in (a)
the ease and speed of responses, (b) the logical quality of percepts,
(c) the amber and quality of 5 responses, (d) the amber and percentage
of}. responses, (a) percentage of sharply perceived forms, (1’) the

unique charges in each _5_ are used only to establish group trends which
are in turn used to discover general laws regarding the relationship of
Rorschach charge to clinical status charge. It is possible to quantify
global judgments (see Karon I: O'Grady, 1970). No study to date, how-
ever, has quantified global judgments of test-retest charge in Rorschach
performance.

6

umber of good quality 3 responses, and (g) the amount of difference
between the second protocol and the first, i.e., the appearance of new
responses and the inproved quality of responses carried over fron the
initial test.

Helpern (1940) examined the perfornance of 17 male schizOphrenics
before and after one to three nonths of insulin treatnent . Follow-up
data were utilized to separate those who recovered and naintained ade-
gm. social functioning for at least a year followirg treatment fron
those who either failed to respond favorably or who suffered relapses
within the year. For all but one of'a umber of Rorschach scores, in-
cluding g, 151%, 2+1, 5, g, _Ag, 5, and gain, there were no significant
charges on retest in either group. There was a significant decline in
the average umber of Q: among inproved patients whereas no 9: charge
occurred among unimproved _S_s.

A third study involving schizophrenics and insulin coma treatment
(Kisker, 1942) noted that patients who recovered gave fewer _D_ and 95
reaponses upon retest, had a drop in 5- 1., and showed a tendency for ini-
tial Confused or Irregular Sequences to shift toward Hethodical Sequence.
g+%increased, but its charge was less consistent than was the §-% de-
crease, largely because soIIe initial _F_'+ responses became mltiple-
deterreinant responses upon subsequent testing. The relative paucity of
Rorschach changes in improved _S_s led Kisker to question whether shock
therapy , though having an effect upon overt behavior , had actually
brought about any charges in fundamental personality structure. Kisker,
like Piotrowslci (1939), didn't provide data that could be statistically
analyzed.

Rioch (1949) studied the pre- and posttherapy Rorschach of a group
of hospitalized schizophrenics. Host of her attention was directed
toward detectirg changes associated with psychoanalytic therapy, so her
discussion of Rorschach changes ampanying clinical status changes was
unsystenatic and generalized. There was one score, however, for which
data amenable to statistical testirg was available: Incidence of in-
crease in g was associated with improved clinical 'statns.

' Several researchers have used patterns of scores or composite indi-
ces rather than merely tabulatirg quantitative changes in separate re.
sponse categories.

nuanch (1947) touched off a rash of such studies in the wake of his

7

creation of a set of signs of satisfactory adjustment. Acting on the
basis of clinical experience with the test, he established minimally
acceptable levels for each of 22 scores. A protocol containing, for
exuple, at least 3 5 responses, or 2 E responses, or 5% below 50,
was considered indicative of satisfactory adjustment with respect to
these particular scores. It was held that the greater the tuber of
these signs to reach satisfactory values, the better adjusted was the
individual.

Although his was essentially a shady of the effects of client-
centered therapy on a group of 12 persons (the first research design),
acrudeneasureoftheoutconeoftherapywasavailableforeachcase,
thus allowing the writer to analyze Huench's data for Rorschach changes
as a function of clinical status change. Although the mean number of
satisfactory signs increased significantly for the entire group after
they received therapy, the pretherapy to posttherapy gain vanished when
gs were separated into a group showirg considerable improvement and a
group that didn't, i.e., the ilpmved §_s failed to differ significantly
from the unimproved §_s in the average gain in number of satisfactory
signs on retest.

Since there was no control for practice effects, it is conceivable
that gainforthegroup as awholewas afunctionofincreasedfamili-
arity with the Rorschach and not necessarily “cause of the beneficial
effectsoftherapy. Ifthisweretrue, thenboth inprovedandunim-
proved _S_s would have been equally affected by this source of Rorschach
change.

Hamlin and Albee (1948) were aware of the unsatisfactory state of
Huench‘s findings, so they provided the missing no-therapy control group.
‘meirgs didn't showanydifferenceinmeannumberof satisfactory signs
from initial exuination to retest, so they concluded that practice
didn't account for the posttherapy imase observed by niench. Left
unanswered, however, was the question of why fluench's successful therapy
cases didn't show a greater gain in satisfactory signs than did his un-
successful cases.

There is another factor which could account for the failure of
Huench's signs to differentiate on the basis of clinical status charges.

8

With but 12 Se, any statistical test lacks power. The improved group
did show a greater increase in the amber of satisfactory signs than did
the unimproved group, but the difference was insignificant. Perhaps the
use of a larger ample would have disclosed significant differential
gains for the two grmps.

Carr (1949), using all but 2 of the 22 signs, tried to duplicate
fluench's results on a new ample of therapy clients but failed. Several
reasons for his inability to find a significant posttherapy gain in the
amber of satisfactory signs have been advanced (Peterson, 1954; hard,
1966), including the fact that nuench's signs involved a greater degree
of qualitative specification (see below) and that a greater proportion
of Huench's patients improved with therapy than was the case for Carr's
gs. One cannot expect to obtain Rorschach changes when there is little
charge in clinical status.

The writer examined Carr's data for changes in Rorschach scores
that could be related to the mount of gain made during therapy (as
judged by _S_‘s therapists). There were no significant relationships.
Since Carr had but nine _S_s, it is possible that his data analysis suf-
fered from the same problem as did unench's. Small sample size might be
responsible for failure to reject the mall hypothesis in Halpern's (1940)
study, also.

Tucker (1951) used Carr's modification of March's signs and related
sign change to an independent measure of clinical status change (not
based on therapist's judgments). As was the case for his preucessors,
Huench and Carr, Tucker's research efforts were not blessed with signi-
ficant findings.

In reviewing the Blanch-inspired studies of Rorschach charge at-
tendant upon exposure to therapy, Peterson (1954) noted that the inves-
tigators (Haimowitz & Haimowitz, 1952; Buench, 1947) who found signifi-
cant posttherapy index changes had considered the quality of the respon-
ses used as signs, whereas the unsuccessful studies (Carr, 1949; Hosak,
1950) hadn't. huench, for example, counted only 5 or 3 responses of
goodformqualitytowardtheninimalmmberrequiredforanindication
ofgoodadjustment, whereasCarr-andnosak countedallgandgresponses.

In the hope that the successes accruing to greater qualitative
specification of responses would generalize from a situation involving
therapy-induced index charges to a situation in which clinical status

9

change is related to index dianges, Peterson (1954) scored the pre- and
posttherapy protocols of 42 file using Huench's scoring method first, then
Hainowitz and HaiJIowitz's, and finally Carr's. After using water's
criteria to obtain independent neasures of the success of therapy for
eachg, hecorrelatedthe‘l‘uckerscoreswith theanountofgainshownby
gain tens of each of the three sets of adjustment signs. The product-
nonent correlations were insignificant in all three instances.

Hamlin et a1. (1952) used two different indices to compare VA
patients who improved after six months of therapy to those who didn't.
”either the Basic Rorschach score nor Muench's signs differentiated
between the groups.

Thus nuench's sign approach, though souetines sensitive to before
and after therapy group differences, consistently failed to distinguish
batman persons who responded favorably to therapy and those who didn't.

The hunroe Inspection Technique (Munroe, 1945), a checklist de-
scribed as the "nost useful set of signs available" by a renowned critic
of the uisuses of statistics with Rorschach data (Cronbach, 1949), was
deployed by Lipton and Ceres (1952). When the records of successful
therapy patients were compared to the records of unsuccessful therapy
patients, a significant decline in flunroe's index of maladjustment was
noted in the posttherapy protocols of the former group, but no change
occurred in the latter group.

Gaylin (1966) used both the single score and the sign procedure to
analyze his data. He derived several hypotheses regarding Rorschach
change with clinical status change fro. the traditional "nor-ative-
adjust-eat" theories of natal health and then from the self-actualizing
theories propounded by Rogers (1961) and Haslow (1962). His gs were
well-educated, predcninantly ymmg persons who had applied for help at a
universitycounselingcenterandhad received twentysessionsofclient-
centered therapy sandwiched between two. adainistrations eachof- the Bor-
schach and the Butler-Haigh Q sort (Rogers 8: Dylond, 1954). Each thera-
pist used a nine-point scale to rate the amount of gain nade during
therapy.

Bach g was then assigned to one of the four possible therapy outcoee
groups on the basis of the direction of therapist-rated change and the
directionofthechangeindiscrepancybetweenperceived selfandIdeal
Self, obtained free the g sorts. gs which both measures indicated were

 

 

 

 

,7; ~-

10

improved made up the "pure-success" group; _S_s which both measures showed
were not improved constituted the "pure-failure" group; those for whom
there was disagreement between the two measures formed the "mixed suc-
cess-failure" groups.

A composite "structure" score, consisting of a smumation of the
standard score values for _P_‘+%, 3, £9- - (EC: 4- [Cid-SJ), and (yang/3,
all of which were assumed (on the basis of nonnative-adjustment cri-
teria) to increase with health, didn't differentiate among any of the
outcome groups . However, a composite ”function" score, derived from a
self-actualizing concept of health and including _lg, 9+, iii-simple 1',
Egg (number of content categories), non-2%, 5, and a modified version
of Beck's Experience Actual (_Efi), did separate the “pure-success" group
from each of the other groups.

Separate score analyses were made for each of the components of the
structure and function composites. Gaylin discovered that 9+, gf-sim-
ple fl, 2933, and 5, each taken by itself, didn't discriminate among any
of the four groups. 91, non-3%, and EA; did distinguish between the "pure-
success" group and the "therapist-rated failure, 3-sort-rated success"
group. .35 also discriminated between the "pure-success" and "pure-
failure" groups, and non-5% was significantly lower in "therapy-rated
failure, g-aort success" gs than in gs with the reverse status.

There have been four studies which used the fourth research design.
The first was done by Barry et a1. (1952). Written reports of pre- and
posttherapy clinical status interviews of cooperating former VA patients
were evaluated in terms of a rating scale for over-all personal, social,
and vocational adjustment. The posttherapy interview and Rorschach were
given six months or longer after the discontinuation of treatment at the
VA outpatient clinic. The authors tested a number of hypotheses gleaned
from the literature in regard to Rorschach changes associated with im-
provement. Pre- to posttreatment changes in the raw scores derived from
the clinical status rating scale were correlated with the changes in
seventy Rorschach scores before and after treatment. Only three of the
seventy rank-order coefficients survived the test for statistical sig-
nificance-152, 32%, and (g+gg)%. Clinical status improvement was asso-
ciated with increases in each of them.

Rickers-Ovsiankina (1954) used the Rorschach to determine whether

11

changes in psychiatric status reflect primarily alterations in basic
personality stmctures or alterations in the more overt aspects of
behavior. Adult schizophrenics were retested several years after they
had been initially examined, and, on the basis of their case histories,
were classified as having shown clinical status improve-ant during that
tineintervalorashavingr-ainedroughlythe sanesincetheinitial
evaluation. 5- (92+g), 5+1», and (_li-o- :I.ncl)%3 differentiated between the
"improved” and “static” groups, with ”improved" gs showing a retest in-
crease in each score. Other neasures, 5, fl, (g+g)%,§§_, g+g, fi%,§_%,
2%, and 2%,didn't show retest changes which corresponded to clinical
status change.

In an exceptionally well-designed study, R. Goldman (1955) investi-
gated the validity of the theory that schizophrenic behavior is funda-
mentally a withdrawal syndrome. According to the theory, the schizo-
phrenic individual has retreated from his own sectional experiences .
Therefore, recovery should involve a reversal of the withdrawal process.
Rorschach variables which clinicians usually interpret as being indica-
tive of withdrawal were used to measure each of the three aspects of
withdrawal. Like fluench, Goldnan didn't simply count every response
within a given scoring category but differentiated between good and poor
quality responses.

The ratio of "healthy" 5 and 5 responses to ”unhealthy" g and 11 re-
sponses and the color balance (2939;4-5) were chosen as indices of the
nature of the individual's interpersonal relations. 5%, 8-9-10%, Beck's
regressive shift,4 and the ratio of fan-dodnant to for- subordinate °
and fornless color, shading, vista, and texture-detenined responses
(Q'; &§§:EIEE +1) measured the expression and control of affect; and
§_+ ‘1», g, pathological content (destructive and anxiety-laden percepts) ,
and a conposite of rejection, denial, and self-reference responses,
served as neamres of attitudes toward the environment.

In a fashion similar to Gaylin's, the separate but presmably

 

3(_1:+ incl)!» is calculated on the basis of the quality of Pan‘-
doninarrt responses (32, _1-3, 31, etc.) as well as on the basis of pure
For. responses.

gamut away from _P_g toward 2 becomes progressively pathological
in the following sequence: 22+, g, g:+, g}, g.

12

related scores were combimd to font coeposite indices of attitudes
toward (a) other persons, (b) the environment, and (c) one's own emo-
tions. .

Golchan's §_s were acute schizophrenics who were examined shortly
after admission and were re-examined several days before discharge or
transfer to another institution. They were rated for degree of im-
prove-entatthetimeofdischarge, roughlyone-thirdbeingjudgedto
be "markedly improved," "slightly or moderately improved," and "not im-
proved," respectively.

Both groups rated as improved differed significantly frm the unim-
provedgroups inthefrequencyoffavorable changesineachofthethree
attitudinal areas. Furthermore, they differed with respect to most of
the separate scores. Improved _S_s showed (a) an increase in the ratio of
"good" to "poor" 5 and 31 responses, (1)) an increase in _rg responses
relative to other kinds of color responses, (c) mt toward the op-
ti-n level of g “55-80), (d) movement toward the optima level in
8-9-10%(30-45), (e) reversal of the regressive shift,5 (15) a shift
toward the optin- level of {+1. (70-90), and (g) a reduction in the fre-
quency of responses containing pathological content.

uMarkedly improved" _S_s stowed, in addition to the increases in the
sevenscoresjustmentioned, (h) anincreaseintheproportionofform-
dominant shading, texture, color, and vista responses to fan-subordin-
nate and fonless responses of the same kind. Although there was no
differential change associated with change in clinical status within the
rejection, denial, and self-reference category, further analysis indica-
ted that, when considered separately, both the frequency of rejected
cardsandofperceptsofferedbyafibut subsequentlydeniedornegated
declined more frequently among improved §_s than among unimproved gs.

 

580th quantitative and qualitative changes were counted when deter-
mining whether a reversal had occurred. Quantitative changes can be
determined for the color balance. In the absence of shifts in the color
balance, any alterations in the form quality of the color reaponses were
noted. 8.9., if a reSponse was scored g— on the initial test and was
repeated on the retest, but was modified enough so that it could be
scored 93, then a favorable change was said to have taken place. Thus,
reversal of the regressive shift meant either an increase in 32 relative
to Q: and g or a transformation of a poor form (3:) color response into
a good f (5%) color response.

13

The same direction of change occurred for marks involving a reference
to the examinee, but the relationship wasn't significant.

Nmber of 3 responses was the only variable in the study which
failed to show a differential change among the three clinical status
change groups.

The most recent research of the fourth type (Zamansky aeoldman,
1960), involved 96 chronic mental patients, some of whom had received
somatic therapy: others having gotten occupational and recreational
therapy. Some had participated in group therapy as well. The only con-
dition for inclusion in the study was that of giving at least nine scor-
able responses to the Rorschach. Clinical status change was ascertained
by the use of scales for rating social adjustment on the psychiatric
ward.

Eleven signs were chosen on the basis of their traditional inter-
pretive significance or because of empirical support gained in previous
research. They consisted of eight separate scores-45+, _EE, 3-, g, _S_e_x_,
£4» %(Klopfer's good form level), (§+§g)%, and (5+ég+£bl)%-and
three composite values indicative of mature cognitive developentuan
index of integration (2' ample: aids and Eva/Z: all 31 and 2 responses),
an index of primitive thought (2 of confabulations , containations , and
fabulized colbinations/Sl of all _W_ and 2 responses), and the Ex index
(form-dominant color and shading responses minus form-subordinate and
fonless color and shading responses).

The only score associated with clinical status change was (_A_+§_d_+
5221) %. Zamansky and Goldman noted that satisfactory ward adjustment
for chronic patients is facilitated by the tendency to engage in stereo-
typed thinking and behavior. This propensity was clearly indicated by
the increase in (5+52+_A_gg1) %.

Studies involm Qualitative 2.11% 931:3. In addition to the
quantitative studies, several atteepts have been made to relate a holis-
tic or global evaluation of Rorschach change to an independent measure
of clinical status improvesent.

Carr (1949) utilized a global assessment of Rorschach change as
well as the sign approach. A judge's decision as to whether the ink
blot test indicated either improve-est or no change was wepared to the
evaluation of improvaent made by each §_'s therapist. Agreuent was
reached for seven of the nine cases (2- .12).

14

Lipton and Ceres (1952), in addition to using the Munroe checklist,
instructed a psychiatrist unfamiliar with the gs or their protocols to
judge whether each _S_'s second Rorschach showed improvwent over the
first. In seven of nine cases his evaluation of the direction of charge
concurred with the independent clinical assessment.6

Global judgments of Rorschach charge were associated with psychia-
tric evaluation of charge among male schizophrenics (Rickers-Ovsiankina,
1954). The outstanding impression about Rorschach change, however, was
that almost every _5_ showed greater impoverisl'nIent on the retest. The
only instances of movuaent toward "richer, better-organized” records
occurred amorg four of the nineteen psychiatrically improved _S_s.

In Zamansky and Golchan (1960) a clinician judged which of the two
records for any _S_ showed the greater amount of social adjustment. Al-
though he lcnew which two protocols belonged to any one §_, he didn't know
which was the original performance and which was the retest. Whenever
he was unable to make a confident decision as to which of the two rec-
ords was the better, it was concluded that there was no change. Judg-
ments of the direction of charge concurred to a highly significant det- ‘
gree with behavior scale ratirgs of improvement .

Sample characteristics and the criteria for improvaent in each of
the precedirg studies are listed in Table 1. Sunarized in Table 2 are
the findings of the various studies with respect to each Rorschach vari-
able.

Bvaluation 93 previous research

An attempt to draw conclusions from the data presented above rust
take into consideration the methodological adequacies of each study as
well as the extent of agreement among the several studies. Following

 

68ince no information was given as to the amber of file judged to be

improved or unimproved by either the Rorschach or clinical status, the
statistical significance of the results is unknown. The authors did
disclose, however, that one g called improved by the Rorschach was
judged not improved by the clinical status evaluation, and there was one
S for which the reverse was true. By placing these cases into diagonal
cells of a 2x2 table and successively assuming every possible combina-
tion of values for the other two cells, the one-sided 2's for the cor-
rect matching of seven of nine cases, with the marginal values uricnown,
range from .17 to .89 (Fisher's Exact Test). The 11111 hypothesis is not
rejected in any event.

15

Table 1

Characteristics of Studies of Rorschach Charge
as a mnction of Clinical

Status Change

 

 

Study

Inscription of gs

Measure of Clinical Status

 

 

Barry et a1. (1952) 31 forter VA patients

Carr (1949)

Gaylin (1966)

Goldlan (1955)

Halpern (1940)
Hamlin et a1.

(1952)
Kisker (1942)

Lipton & Ceres
(1952)

fluench (1947)

Peterson (1954)

Piotrowski (1939)

Ridcers-Ovsiawcina
(1954)

Rioch (1949)

Zamansky 8: Goldman
(1960)

9 neurotics in
client-centered
therapy

57 recipients of
client-centered ther-
apyo-both sexes

42 acute schizophren-
ics

13 young male schizo-
phrenics getting ICT

20 male neurotic and
psychotic VA patients

28 psychotics in ICT

11 neurotic VA pa-

tients in psychoana-
lysis

12 recipients of
client-centered ther-
apy-~both sexes

42 college students

60 schizophrenics in
ICT

38 adult male schizo-
phrenics , half of
whom got ICT or ECT
22 hospitalized
schizophrenicsuboth
sexes

96 chronic psychotics
«both sexes

Personal-social adjustment
determined by standardized
interview

Therapist evaluation of im-
provement

Therapist rating and Butler-
Haigh 3 sort

Socio-sexual-emotional ad.
justment--rated by therapist
and another psychiatrist

Post-treatment resumption of
normal social role

Therapist ratirg of improve-

ment

Personality charge "likely
to help in ultimate adjust-
ment"

Analyst '3 judgment of symp-
tomatic charge

Therapist evaluation of im-
provement

Tucker's multiple-criterion
measure of therapeutic suc-
cess

Judged by medical staf --
criteria unspecified

S-judge consensus on the ba-
sis of written case histo-
ries

Therapist judgment of im-
prove-ant

Social behavior as rated by
ward personnel

 

 

16

Table 2

Sr—ary of Research on Rorschach Score Change
as a Function of Clinical Status Change

 

 

Direction of score change with

 

 

5”“ clinical status improvunent Study
32 Increase Barry et a1. (1952)
Increase Piotrowski (1939)
No change Carr (1949)
No change Halpern (1940)
No change nucnch (1947)
No charge Ridcers-Ovsiankine (1954)
No charge Zamanslcy 8: Golchan (1960)
fig : (:3: +9 Relative increase in §_'C_ Goldman (1955)
Relative increase in _PE Rickers-Ovsiankina (1954)
No charge Barry et a1. (1952)
No change9 Carr (1949)
No changeb Gaylin (1966)
No change8 Halpern (1940)
No change“ Huench (1947)
Q: Increase Rioch (1949)
Decrease Halpern (1940)
No charge Barry et a1. (1952)
g Increase Piotrowski (1939)
Increasec Gaylin (1966)
No charge Barry et a1. (1952)
No charge Carr (1949)
No change Halpern (1940)
No change Rickers-Ovsiarkina (1954)
g-o- Increase Piotrowski (19 39)
No change Muench (1947)
No charge Zamanski 8: Goldman (1960)
{4- 1 Increase Kisker (1942)
Increase Piotrowski (1939)
Increase Rickers-Ovsiankina (1954)
No charge Barry et a1. (1952)
No change Carr (1949)
No charge Gaylin (1966)
No charge Halpern (1940)
Move toward optimu- level Goldman (1955)
No shift toward optinn Carr (1949)
No shift toward Optima Halpern (1940)
5 No change Barry et a1. (1952)
No change Carr (1949)
No charge Gaylin (1966)
No charge Halpern (1940)
No charge huench (1947)

No change

Ridcers-Ovsiarkina (1954)

 

 

17

Table 2 (cont'd)

 

 

 

 

 

b:
3 Direction of score charge with
clinical status improvasent Study
a z. Decreased Gaylin (1966)
No charge Barry et a1. (1952)
No charge Carr (1949)
No change Halpern (1940)
Move toward Optimum level Goldman (1955)
No shift toward optisum Carr (1949)
No shift toward optimm Halpern (1940)
g No change Barry et a1. (1952)
No charge Carr (1949)
No charge Gaylin (1966)
No change Goldman (1955)
No change Halpern (1940)
No change Huench (1947)
No change Zamansky 8: Goldman (1960)
Rejection of cards Decrease Golchan (1960)
No change Halpern (1940)
No change wrench (1947)
Basic Rorschach No charge Barry et a1. (1952)
Score No charge Hamlin et a1. (1952)
Muench's 22 signs No charge Hamlin et a1. (1952)
No charge Huench (1947)
No charge9 Barry et a1. (1952)
No char-gee Carr (1949)
No chargef Peterson (1954)
llunroe checklist Decrease Lipton a Ceres (1952)
No charge Barry et a1. (1952)
Global charge Successful Rickers-Ovsiankina (1954)
Successful Zamansky at Goldman (1960)
Trend toward success? Carr (1949)
Trend toward successg Lipton & Ceres (1952)
“so : 2g; + 3_c_:_.

cApplies only to the comparison of ”pure-success" _S_s to "therapy-
failure, 3-sort success" _S_s.

dApplies to "pure-success” group vs. "t11erapy-fai1ure,Q-sort-success"
and "therapybsuccess, g-aort-failure" vs. ”therapy-failure, Q-sort-success."

eCarr's modification of Huench's signs.

f

Used both Huench's and Carr's signs.
978 %(7 out of 9) agreement between Rorschach change and CS chame.

 

18

the presentation of the conclusions to be gained from previous research
on Rorschach charges as indicators of clinical status charge is a dis-
alssion of the procedural pitfalls to which the studies were suscepti-
ble. Each Rorschach variable which has been investigated by two or more
independent studies is considered below. Other variables, each havirg
been found to be related to clinical status by a single study, are in
need of replication before they can be seriously considered.

Separate 39952 change. InSpection of Table 2 reveals that the
findings are inconsistent with regard to most of the scores which have
been tested for association with clinical status change. There are four
exceptions which merit discussion: _T:_‘C_ : g+g, 2%, g, and 5.

Favorable shift in ‘Fg : g +2 means that _52 increases relative to g:
and 9. Although the overall impression is that there are contradictory
findings with this variable as is the case with most other scores, it is
noteworthy that change in color balance coinciding with change in clini-
cal status was reported in two of the three studies using schizophrenics
as _S_s. No _Fg : g; +2 shifts occurred in any of the four studies using
less severely impaired individuals as _S_s. In the one study of schizo-
phrenics (Halpern, 1940) not findirg favorable color balance charge with
improvement, there was a trend in that direction. The small sample size
may have precluded the possibility of its attaining statistical signifi-
cance.

A phenomenon similar to that observed above with color balance oc-.-
curred with regard to change in §+%: four out of the six studies which

used psychotic gs found a direct relationship to change in psychologi-
cal health; none of the three studies which used neurotic gs found a
relationship. Halpern (1940) was one of the two studies of psychotics
not finding a relationship, the other being Zamansky and Goldman (1960).
Goldman (1955) pointed out that a hypothesis more consistent with
Rorschach theory is that §+% moves toward an optimum value (70 to 90)
with improved mental health, for scores’which are either unusually high
or low are considered pathological. This hypothesis is thus capable of
rendering intelligible both a modest decrease in {+7. in an overly-con-
trolled and constricted individual with an initial score of nearly 100%
and an increase in 2+1» in a person with initial score below 60%. In
Goldman's study, schizophrenics with either type of change showed

19

improvement in mental health.

It is quite possible that the link between _1;+% charge and clinical
status change found exclusively among the studies involvirg psychotics
is due to the fact that such individuals typically have low {3% values
(reflecting their impaired reality testing). Clinical status improve-
lent would logically involve an increase in Ed. . The situation is dif-
ferent anong neurotics; if they are abnormal in 3+1. , they are more like-
ly to have extremely high values rather than very low scores. Improve-
ment for the individual with a constricted protocol would involve a
loosening of defenses with a consequent drop in 3+1. toward the level
usually found in the psychologically normal individual . It then follows
that Rorschach studies of clinical status improvement in neurotic or
other highly defensive individuals should properly look for £+% to move
toward the Optimum level rather than toward the raaxinun score. One can-
not help but wonder whether the conclusions of the studies using neuro-
tics as gs would have been charged had they used this means of measuring
5+ %charge.

There are but two scores for which the various studies have ren-
dered a unanimous verdict: there is no relationship between change in
adjustment level and change in g or in g.

"812" chge. With the exception of one (Lipton 8: Ceres, 1952) of
the two studies usirg the Munroe checklist, composite ”sign" charges
Quench's 22 signs, the Basic Rorschach Score) have consistently failed
to correlate with clinical status inprovement.

Global M. The two studies (Rickers-Ovsiankina, 1954; Zanansky
& Goldman, 1960) obtaining significant findirgs provided considerably '
larger samples than did the two studies which, although showing a trend
toward agreement between global Rorschach and clinical status Judgents
of charge, did not reach the conventionally accepted probability level
for rejection of the null hypothesis. Once again, there appears the
tendency for research involvirg psychotic gs to be more likely to get
positive results than does research using neurotic gs.

M gf findigs. Although changes in 5+7. , and, to a less cer-
tain extent, in color balance, have been demonstrated to nirror charges
in the Dental health of schimphrenics , no single Rorschach score or
index of "signs" has been consistently demonstrated to reflect clinical

20

status change across all types of §3--neurotic, psychotic, or otherwise.
To those who have insisted that the Rorschach cannot capably serve as a
research instrument unless it is used idographically, it comes as no
surprise that the global” judgments of charge appear to have done the
best of the three types of Rorschach analysis in reflecting alterations
in one's ability to deal effectively with his enviroment.

Comparison 3f the merits 9: global FE sepaiate £9952 studies.
Global evaluations enjoy a significant advantage over coutparisons based
on separate scores or checklists in that the former technique can uti-
lize all the relevant information present in a protocol. Two aspects
to this more caplete usage of the data are worthy of mention. First,
methods using sirgle scores and signs cannot readily take into account
the context within which the discrete score charges take place. An in-
crease in §+ %from 55 to 90 may indicate one kind of personality change
when accompanied by a decrease in 5 and a drop in Experience Balance
from 2 : 5 to 0 : 0 but signify something quite different when accompanied
by an increase in g and a shift toward ambiequal g : gsum. In the purely
quantitative evaluation of score charges there is no simple way to spec-
ify which of the alternative meaningsfor an 3+ %change may be correct.
The some difficulty applies to most other scores as well.

Second, to the extent that the qualitative attributes of Rorschach
performance are more sensitive to charges in §_'s clinical status than
are the quantitative characteristics of the protocol (see Oppenheim &
Brower, 1948; Piotrowski, 1939; Piotrowski 8: Schreiber, 1952), a global
analysis of charge is superior to an approach which considers only charges
in the frequency of scores. Oppenheim and Brower (1948), in fact, re-
ported that many _S_s undergoirg therapy did not show an increase in abut
did manifest changes in the quality of g: reSponses took on a more ac-
tive character, larger areas of a blot were involved in the responses,
and §_s were more confident and specific in describirg their percepts.

Now, the eager score or sign investigator could attanpt to over-
come this handicap by quantifyirg charges in the quality of responses,
e.g., counting changes in "warm" color responses only, or charge in _rg g
reSponses but not in 5 fig responses, or the amber of _S_s who shift from
giving 2+ responses to giving _FE- responses. There is a decided risk
involved in such an action, for what good would it do to look for a

’L

21

relationship between clinical status improvement and shifts from "sta-
ticgtoagrayblot" to "extensorgtoacoloredblot" if buttwo or
three §_s out of the entire sample happened to give such responses?
Adoption of the quantitative approach to the neasuring of Rorschach
change requires the selection of variables which appear with sufficient
frequency so that §_s will show differential change upon retest.

Instead of continuing the heretofore fruitless search for score
charges that are foolproof indicators of clinical status change, the
student of the Rorschach might be sorely ternpted to concede victory to
the front-runner. Granting the more successful history of global eval-
uations in reflecting changes in overt adjustment, it can nevertheless
be demonstrated that exclusive reliance upon the global method does not
really represent a satisfactory solution.

A major drawback to its use is that test-retest change in inkblot
performance is usually described in the most general of terms: "im-
proved” or"not improved." Just what is the nature of the alteration
that has taken place in the person whose charge in Rorschach behavior
has been thus characterized? To be more specific, were the criteria
for ”improved Rorschach performance” similar or different in the four
previously cited global studies? Questions such as these remain unan-
swered despite the modest success of global Rorschach studies in de—
tecting clinical status charge, for (given this particular research
context) the global method of measuring Rorschach charge tells as such
about the carpetence of the judge as diagnostician as it does about the
specific dynamics of personality change.

Answers to the questions posed in the preceding paragraph can be
gained only by focussing attention upon the specific psychological pro-
cesses which are believed to be intimately related to overt manifesta-
tions of mental health. The challerge to the Rorschach investigator is
to identify and subsequently utilize variables which will provide valid
treasures of the inferential processes. A purely global evaluation can-
not do this .

Problems 2: design, Although Rorschach textbooks abound with ex-
amples of separate scores and patterns of scores which are credited
with being indicative of the level of adequacy of personal functioning,
not one score, upon submission to controlled mpirical scrunity, has

Lu,

P
4 'JC

22

lived up to its promise as a measure of clinical status change. Why
hasn't the utility of Rorschach scores in assessing personality charge
been demonstrated? Although the previously discussed shortcomings in-
herent to quantitative methods of ascertaining Rorschach change may be
to blame , the confounding influence of avoidable methodological errors
in the research to date cannot be overlooked. Perhaps the fault lies
notsonrchinthemethodofinquiryasintheexecutionof thatmethod.
Below is a list of the salient factors the writer believes are respon-
sible, in part or in whole, for the inconclusive state of existing re-
search on the relationship between Rorschach score change and clinical
status charge. The first four items deal with the fact that the re-
sults of the various studies are not always comparable. The next six
points deal with problems of faulty design:

1. The use of different pepulations. The types of individuals
used as §_s has run the gaunt from well-educated, college-aged narrotics
in therapy voluntarily to middle-aged mental patients with psychoses of
long duration. As Schafer (1955) has pointed out, there is no reason
to expect Rorschach charges in one individual showing clinical status
change to be identical to those of another individual who shows a simi-
lar change in clinical status. Improvement for a schizophrenic, for
instance, might involve a strergthenirg of defenses with resultant di-
mimtion of primary process intrusions , while improveaerrt for an overly-
controlled neurotic might entail a relaxirg of defenses with consequent
increase in modulated expressions of primary process material. Obvi-
ously, the kinds of Rorschach charge expected of the two individuals
would be quite different. The evidence suggests, in fact, that popula-
tion differences do have a decided influence upon the nature of the
relationship between charge in clinical status and change in at least
two scores-§+%and _Fgrg-rg. Therefore, comparisons of different
studies must take heed of the kinds of peeple to which the findirgs
apply.

Piotrowski (1960) is of the opinion that the relationship between
single Rorschach scores and overt behavior is more direct amorg poorly-
integrated individuals than it is among betterbintegrated persons, i.e.,
that the context of given score charges becomes less important as the
degree of psychopathology increases. His contention gains credibility

23

upon recognition of the fact that the two scores found to vary with
clinical status charge did so in schizophrenics but not in neurotics.

It is also possible for significant findirgs to be hidden in stud-
ies having decidely heterogeneous samples. Schafer (1949) claimed that
serrples containirg persons of similar personality structure and social
class are more likely to obtain consistent changes in separate scores
that reflect changes in psychological well-being. It is for similar
reasons that Goldman (1955) argued against including both chronic and
acute cases in a study of schizophrenics. There is some empirical sup-
port for this view. Hamlin et a1. (1952) included both neurotics and
psychotics in their sample and found no therapy-induced Rorschach
charges for the group as a whole. When considered separately, however,
the neurotic _S_s did show significant change.

2. The use of differing criteria in the evaluation of clinical
status charge. Barry et a1. (1952) and Zamansky and Goldman (1960)
illustrate the point quite nicely; improvement in the former study was
based upon the adequacy of overall functioning within a normal environ-
ment, whereasimprovementinthelatter studywasdefinedintermsofa
patient's observing the amenities of the psychiatric ward-«cooperation,
tidiness, and participation in planned social activities. The ability
to be a "good" patient does not necessarily imply the capacity to func-
tion within a normal social milieu. Thus, it would not be surprising
to get contradictory findings from the two studies.

3. Exposure to different kinds of conditions during the test-
retest interval . Since the type of therapy a person gets influences
the course of subsequent personality charge, it is advisable that the
investigator qualify his conclusions whenever results come from re-
search based on reSponse to only one kind of treatment. Adoption of
the fourth research design (see pp. 2-3) minimizes the danger of over-
generalization because clinical status change stealing from any and all
causes is studied.

4. The use of different scoring systems. The Beck and Klopfer
systems differ in their standards for scorirg certain types of respon-
ses, especially those with regard to form quality, the shading deter-
minants, and popular content. It is conceivable that two or more stud-
ies might obtain conflictirg results regardirg a variable, such as 3+ ‘1,

24

because in each study different classes of responses were considered
scorable for 5+ or g- .

5. Sample size too small to permit tests of significance to uti-
lize their full power. Nearly half the studies had less than 25 gs,
thus providirg fertile ground for the proliferation of type II errors.

6. The use of theoretically indefensible hypotheses. A prime
exasple is the assumption of a linear relationship between the value of
every score and the level of adjustment, i.e., that maximum equals ep-
timm. The most col-Ion victim of this fallacy has been the interpre-
tation of 3+1: charge (see pp. 13-14). Goldman (1955) is to be consen-
ded for making allowances for curvilinear relationships in her deduc-
tion of hypotheses regardirg change in 34-7,, £1», and 8-9-10%with im-
provement in clinical status.

7. The confusing of a charge in productivity (3) with a change in
the values of other scores. Piotrowski and Schreiber (1952) offers an
excellent example of this problem. Patients gettirg psychoanalytic
therapy demonstrated considerable Rorschach change upon retest , espe-
cially in human movement and color. No significant changes occurred
amorg individuals getting a more superficial , supportive therapy.
Moreover, the psychoanalytic therapy group averaged a nearly two-fold
increase in R on the posttherapy Rorschach, while the other group's
average productivity didn't change. Since _R correlates with most other
scores (Cronbach, 1949; Fiske & Baughman, 1953), could the increases
made by the psychoanalytic therapy _S_s be merely the result of becomirg
more productive? Although the possibility wasn't discussed by the
wthors, it does merit serious consideration. Unless proven to the
contrary, application of the law of parsimony would require such a con-
clusion despite the striking increases observed in scores other than 5.

Piske and Baughman (1953) published a table of the average values
observed for each cos-on Rorschach score at varying levels of _R_. When
each mean score value in Piotrowski and Schreiber's psychoanalytic
therapy group is compared to its "expected" value at the correspondirg
level of _R-n-an admittedly crude procedure necessitated by the availa-
bility of group, and not of individual, data-nit can be seen that the
statistically significant increase in germ upon retest (from 4.8 to 7.5)
wasofthemagnitudetobeexpectedwhengincreasestotheextentthat

25

it did (free 41.6 to 76.8). Not all the score increases, however, can
beexplained away asnereartifactsof increasesg; gains ing‘gand!
were disproportionately high when cocpared to their predicted values at
the higher 5 level.

Barry et a1. (1952) was one of the few studies to show concern
over the effect that 5 change night have on other score changes.

8. The inflating of significance levels by running a large umber
of tests. The validity of the findings reported by Barry et a1. (1952)
is highly questionable. Chance alone dictates that three or four of
the seventy statistical tests they made would reach the .05 level-oand
just three of the scores tested in their study happened to mach this
level of significance. Gaylin's (1966) research, which involved a
total of 66 co-parisons, was also especially prone to this problal.

9. Bias and/or inconsistency in the gathering and interpreting of
data. The failure to insure that the clinical status criteria are ap-
plied consistently to every 3:; in a study can lead to erroneous measure-
ments. The adoption of a standardized neans for obtaining the relevant
infornation is essential. Equally important is that the judglental
process be uniforn for every g. The best way to do this is by having
the sale rater or raters make the evaluations of change for all gs,
thus eliminating the effect that interrater differences would have on
the validity of judgments.

Half the studies (Barry et al., 1952; Goldman, 1955; Hamlin et al.,
1952; Lipton &Ceres, 1952; Peterson, 1954; Rickers-Ovsiankina, 1954;
Zmansky &Goldnan, 1960) utilized both standardized rating scales and
the seas rater(s) for each §_. Although a co-on scale for clinical
status inprovelent was used in Gaylin (1966), there were different
judges for different gs, with each g's therapist rating his patient for
ilprovuent. No attenpt whatsoever was nade to establish unifor- cri-
teria of change or to use the same rater(s) in Carr (1949), march
(1947), or Rioch (1949). Information was insufficient to deter-ine the
adequacy of the procedures followed by Halpern (1940), Kisker (1942),
and Piotrowski (1939).

10. Failure to control for examiner-subject interaction effects
in Rorschach testing. The effect of the examiner's personality upon
Rorschach perfomance can be considerable (Sanders 82 Cleveland, 1953;

26

Zax, Stricker, 8: Weiss, 1960). The ideal study involving repeated ad-
ministrations of the Rorschach would have the same examiner give the
test to every :5 both times. Test-retest changes may be spurious due to
the differential effects that two different examiners have had upon gs.
On the other hand, it is equally possible for genuine Rorschach changes
to be "wiped out" because of the new examiner's influence.

It is clear that Barry et al. (1952) and Muench (1947) didn't con-
trol for the examiner effect. The information available isn' t suffi-
cient to ascertain whether Carr (1949), Gaylin (1966), Goldman (1955),
Halpern (1940), Hamlin et a1. (1952), Kisker (1942), Lipton and Ceres
(1952), Peterson (1954), or Piotrowski (1939) attenpted to deal with
the problem There are but three studies in which it is known that an
effort was made to use the same examiners for all Rorschach testing--
yet even these weren't entirely successful. Rioch (1949) gave 62 of
the 72 Rorschachs herself and instructed the other examiner to achinis-
ter the inkblots in the same manner as she had. Zamansky and Goldnan
(1960) controlled for the influenCe for §-_S_ interactions to the extent
that the same individual gave the initial and second Rorschachs "to any
one patient in almost all cases." Kickers-cvsiankina (1954) adminis-
tered the initial test to half her 533 and the retest to all _S_s.

The reality-oriented researcher is forced to admowledge that in
every study there will be a gap between the ideal and the attainable in
methodological rigor. One procedure, that followed by Zamansky and
Goldman (1960), although less desirable than that of having one emin-
er for all gas at both times, is preferable to a situation in which sev-
eral persons administer the test with no effort made to insure that
«mg is tested both times by the same person. Another procedure, in
whichoneexanimrgivestheRorschachtoeverygthefirsttimeand
another examiner administers it to every g the second time, keeps the
examiner influence relatively constant in comparing g's, even though
the _s_-§ relationship differs free initial test to retest for all _S_s.

It is not surprising, then, that the lack of consensus and the
multiplicity of causes thereof may cause the interested observer to
seriously doubt that quantifiable shifts in separate scores can reveal
the subtle changes in personality postulated by dynamically oriented
psychologists to underlie or aces-pany significant changes in clinical

27

status.

Yet, it can be deuonstrated that such analyses, when done properly,
bear fruit. Research such as Goldman's (1955), in which explicitly
stated hypotheses about score change were formulated and subsequently
confirmed, suggest that this is so. One Goldman, however, does not a
consensus lake.

In all fairness, it should be acknowledged that several of the
studies which have been presented and criticized (Carr, 1949; Halpern,
1940; Hamlin et al., 1952; Huemh, 1947; Rioch, 1949), were designed
not for the investigation of Rorschach change related to clinical sta-
tus charge but for the study of the effects of therapy on inkblot per-
formance. Their inclusion in the present review was possible only
because the writer was able to glean the necessary data from the psy-
chograns which each author so thoughtfully included in his publication.

A peculiar and most troublesome attribute of the Rorschach test is
that a particular kind of charge in a given score can mean different
things for different persons. The context of the score charge and the
value of that score on the individual's initial protocol determine the
meaning of observed charge in the score upon retest. Thus, the inves-
tigator who has utilized separate score analysis (and who has been suc-
cessful in circumventing the methodological problems which have trapped
a umber of his predecessors and caused their findings to be suspect)
is < confronted with the formidable task of attespting to discover ex-
actly what his findings mean.

Therefore, it behooves the investigator to establish a pgiori a
clearly articulated rationale for his selection of variables to be
examined empirically, for the analysis of charges in separate scores is
meanirgful only to the extent to which these changes can be related to
theories of personality and psychopathology. It is for this reason
that Schafer (1949, 1958) has advocated the use of interpretations,
rather than scores, as the basic unit of research with the Rorschach.
The investigator formulates hypotheses about shifts in covert processes
that he expects will vary vis-a-vis change in overt measures of adjust-
ment. Rorschach scores are then chosen on the basis of their capacity
to measure the hypothesized inner changes. The present study applies
this procedure in that a particular theory of personality change has

28

determined the selection of Rorschach variables for investigation.

In addition to the intelligibility which it lends to the data, the
prediction of Rorschach changes enables the researcher to use one-tailed
tests of significance, thereby increasing the power of the tests he
makes.

Statement 25 the problem 3.3 39s; gf Egg aggression theory

The present study explores the relationship between Rorschach
change and clinical status charge within the framework of the ego re-
gression theory of psychOpathology . The conceptual approach involved
represents a combination of psychoanalytic ego psychology and of
Warner's (1948, 1957) couparative-develOpmental theory.

Psychoanalytic theory considers regression to be a dynamic process
which is amorg the ego's repertoire of defenses (Afraid, 1937). Re-
gression may occur totally or segmentally, and it may apply to either
id or ego functions. Not surprisingly, psychoanalytic theorists have
often applied the concept to the schizophrenic syndrome (Arieti, 1955;
Cameron, 1963; Fenichel, 1945; Jacobson, 1967). The schizOphrenic pro-
vides the most striking example of regression as a defensive maneuver
against intolerable anxiety.

werner's (1957) orthogenetic principle, that "Whenever developuent
occurs it proceeds from a state of relative globality and lack of dif-
ferentiation to a state of increasing differentiation, articulation,
and hierarchic intergration,” describes not only the maturation of the
organim as a biological whole but the develOpment of psychological
pmcesses as well. He listed five pairs of "polar Opposites" as beirg
descriptive of the direction of develOpmental changes , the i-ature
organism being characterized by the first term of each pair, the mature
organism being characterized by the second term: (a) diffuse-
articulated, (b) labile-stable, (c) indefinite-definite, (d) syncretic-
discrete, (e) rigid-flexible.

Comparative-developIental theory shares with psychoanalysis the
view that psychopathology involves regression. Ruiniscent of Hughlirgs
Jackson' 5 observations about the effects of cortical injury, that the
"higher" or phylogenetically more recent functions become impaired first
while the "lower" functions main intact, is Werner' s conceptualization
of regression as the relative accentuation of genetically early levels

29

as the higher and more recently developed functions--the products of
differentiation and integrationu-become less dominant.

Goldman (1962) demonstrated that the ontogenesis of perceptual ,
learning, thinking, anotional, social, and motor processes readily fits
into the framework of Werner's theory. Moreover, he noted the similar-
ities between schizOphrenics and children with regard to these six ego
functions and concluded that Werner's theory provides a valid descrip-
tion of schizOphrenic behavior. He thus argued for a comprehensive
approach to schizOphrenia that would include consideration of the for-
mal structure of psychological processes as a necessary complaent to
the traditional psychoanalytic emphasis on the content and function of
these processes.

The psychoanalytic theory of regression is compatible with, but not
identical to, Warner's formulations. The former does not require that
development proceed along the lines Specified by Werner, and it implies
a dynamic process, whereas Werner's treatment of regression is purely
descriptive. Thus, Goldman, in working within the narrower comparative-
developmental framework, used regression as a descriptive concept with-
out implying that it causes schizophrenia. For Goldman, regression is
coezdstent with, but not necessarily a determinant of, the psychosis.

The concept of regression employed in the present study-ma return
to any less mature form of behaviorm-is necessarily descriptive. Re-
gression as an explanatory concept cannot be empirically tested with
the research design at hand (although it may very well be such a dynam-
ic process).

'_1‘_h5_ Rorschach _a_; 3 measure 2; _a_gg deve10pment. The Rorschach me-
thod has been used for the purpose of investigating the maturational
changes specified by Werner. Styles of response, particularly with
regard to percept location, were postulated by Friewan (1952) to re-
flect the mrmal developmental sequence of perception. He devised a
scoring system to classify Rorschach responses on the basis of the
amount of differentiation and integration expressed in each percept .
Helmendinger (1953) used the Friedman scores to study groups of chil-
dren ranging in age from three to ten. He found a significant decline
in the preportion of developmentally "imuature" responses and a corre-
sponding increase in the proportion of "mature" reSponses as the age of

30

the groups increased. His data on adult Rorschach performance dis-
closed a continuation of the trend established by the samples of chil-
dren.

Although it is true that some of the traditional scoring catego-
ries originated by Rorschach (1951) have also been reported to vary as
a function of age, Friedman's (1952) scores represent the foremost
effort to date at devising Rorschach measures of genetic level on the
basis of a specific theory of development.

Evidence _fgg the m regression 2119.951 9; psychopathology. Per-
ception is the one ego function discussed by Goldman (1962) which has
been investigated by mans of the Rorschach. Friedman (1952) and
Siegel (1953) reported that the proportion of develOpmentally mature
responses produced by schizophrenics was significantly lower than it
was for normal adults but that it was the same as the proportion given
by yams children-

Corroborative evidence of ego regression in schizophrenics has
come from recent laboratory experiments (Flynn, 1965; Klorman & Chapman,
1969) which disclosed that schizophrenic adults, when compared to nor-
mal adults and to children in performance on several cognitive tasks,
made errors similar to those couitted by the children.

In a critical review of the experimental and clinical evidence
pertaining to Goldman's (1962) application of regression theory to
schizophrenia, Buss (1966) concluded that there was modest support for
the belief that schizophrenics manifest perceptual regression but that
the findings with respect to the areas of learning, thinking, emotion,
and social behavior were negative-wr equivocal at best. The primary
objection raised was that there are many behavioral aspects in which
schizOphrenics do not resemble children.

The existence of similarities, rather than identity, between the
behavior of the schizOphrenic adult and of the normal child is, however,
congruent with both psychoanalytic and comparative-develomental theo-
ries of regression, for

. . . just as any stage preserves vestiges of the earlier stages
from which it emerged, so will degeneration bear signs of the
higher level from which it retrogressed (Werner, 1948).

For instance, Friedman (1952) discovered that schizophrenics, al-
though resembling children in the proportion of deve10pmentally mature

31

responses, were similar to normal adults in that both adult groups used
a greater umber of content categories than did the children. Friedman
eXplained this difference between schizophrenics and children on the
basis of differential degrees of experience. The young child hasn't
yet had time to become familiarized with the myriad of objects which
couprise the perceptual world of the adult.

A corollary to regression theory is that the more severe the
psychopathology, the more extreme is the regression. Hebephrenic and
catatonic schizophrenics are generally assumed to be more seriously
regressed than are paranoid schizophrenics, and they do in fact give
less mature Rorschach responses than do the latter. With the use of
the Friechan developmental level scores, Siegel (1953) found hebephren-
ics and catatonics to resemble the performance of normal three to five
year-old children while the responses of paranoids resembled those of
six to ten year olds.

Additional support for the corollary regarding degree of regres-
sion and severity of maladjustment came from studies by Becker (1956)
and Van Pelt (1961). In the former, the degree of perceptual regres-
sion (in terms of Friedman's scores) was related directly to the sever-
ity of symptoms and inversely to the level of premorbid adjustment.

The latter study disclosed that in successive groups of _S_s having in-
creasirgly severe levels of psychopathology, responses characterized as
beirg deve10pmentally imature tended to increase while responses con-
sidered to be mature decreased.

The evidence thus appears to support the theory that mental dis-
orders in general, and schizOphrenia in particular, are characterized
by regression in ego functioning. A rather embarrassing question can
be raised, though. Every one of the above-mentioned studies on regres-
sion was cross-sectional, i.e. , schizOphrenic and nonschizOphrenic _S_s
were examined at one point in time. How does one know that the empiri-
cally duonstrated ego immaturity of schizOphrenics was actually concom-
itant to the develOpment of the psychosis? Perhaps the persons who
served as the schizophrenic gs had always looked "imamre"--even in
the premorbid state. If such were the case, it would follow that the
afore-mentioned studies failed to provide a legitimate denonstration of
regression as a phenonenon associated with the onset of psychosis .

tive

32

Buss and Lang (1965), in reviewing studies of regression in cogni-
functioning, concluded that

...it is not sufficient to demonstrate vague similarities be-
tween schizophrenics and children. It must be shown that as
an individual becomes schizophrenic, he retraces the develop-
mental sequence (if there is one) of learning and thinking;

as he recovers,‘he then moves farward toward demonstrably more
mature modes of learning and thinking. These corollaries of
regression theory have yet to be established empirically, and
therefore we conclude that the regression theory of sdhizo-
phrenic deficit is unproved.

The present research has addressed itself to the task by using

longitudinal data to study ego processes within sChizophrenics.

STATETENT OF HYPOI‘HESIS

It is hypothesized that change in the clinical status of individ-
uals who are initially schizophrenic-~whether it be in the direction of
recovery or of further decline--is accompanied by an increase or de-
crease, respectively, in the functioning developmental level of the ego.
Considered from another perspective, it is hypothesized that whenever a
retest Rorschach indicates movement toward or away from a more mature
mode of ego functioning, there is a concomitant improvement or decre-
ment in overall adjustment level.

Confirmation of the hypothesis would indicate that ego regression
is a salient mechanism in the formation of the schizophrenic syndrome.
Investigation of the ego regression theory of psychOpathology simlta-
eously provides an Opportunity to augment the present meager state of
knowledge with regard to whether the Rorschach is capable of reflecting
changes in mental health status.

33

 

 

 

 

 

METHOD

Selection 31: Rorschach variables

The Rorschach scores used in the present study are listed in Table
3 along with the predicted direction of change indicative of clinical
status improvenent. The justification for their use is given below and
is followed by a listing of the specific predictions. The ego functions
to which these Rorschach variables are related are taken from Bellak's
(1969) comprehensive listing.

Table 3

Rorschach Measures of Ego Develomental Level
and Relationships to Inproved Clinical Status

 

 

Predicted direction of change

 

5““ with clinical status improvatent

9&3 Nature 11/ Total 3 Increase

nature 2 / Total 2 Increase
£32 + _C_c_>_l_t_ Decrease
_P§_ Decrease
F+ % Increase
3 Increase
5 Increase
£2 mg Increase
g Increase
5 Increase
Unscorable Protocol Scorable Protocol

 

 

 

Friedman DevelopIental Level scores (£9) . These have been used in
a variety of research settings with impressive results (Hemnendinger,
1960). The present study, however, offers the first opportunity for

34

35

the Friedman scores to be applied to test-retest data.

The ratio of genetically “mature" Whole and Detail responses to
all Whole and Detail responses measures the degree of differentiation
and integration eXpressed in perceptual activity.7 Both of the devel-
opmental level scores, Mature 3/ Total 3 and Mature 2/ Total 2, in-
crease progressively from early childhood to adulthood (Hemendinger,
1953). Among the principal ego functions affecting these scores are
Reality Testing, Thought Processes, and Autonomous Functioning.

Pabulized Combinations and Contaminated Responses (gag +933). In
addition to the developmental level scores, Friedman (1952) used a
value consisting of the sum of Fabulized Combinations and Contaminated
Reaponses as an indicator of genetic level. Each type of reSponse
involves a fusing of two percepts solely on the basis of either spatial
contiguity (Egg) or spatial identity (995). In both instances cogni-
tive processes are stimulus-bound rather than based on logic. Fusion
is illustrative of the syncretic, labile qualities which characterize
primary process thinking. Secondary process thinking, on the other
hand, is typified by the attributes of discreteness and stability. £39
+ £95 declines with age and is usually absent in the records of normal
adults (aner, 1953).8

Perseveration £9253 (11$). Friedman (1952) used this measure of
perseveration, devised originally by Phillips and Elmadjian (1947), as
an index of perceptual rigidity. gs is a value, expressed as a per-
centage, which is obtained by crediting g with one point for every time
there is repetition of the entire content of an earlier response and
giving one-half point for every time there is repetition of part of the
content of a previous response. The sum of perseveration points is
then divided by the total number of responses.9 The more differentiated

 

7ReSponses scored 113+, 31+, or Mediocre )1 are considered to be ma-
ture, whereas responses scored Mg E, m E, 3-, or Q! are im-
mature. A similar classification is made of all Detail responses. For
definitions and examples of the scoring of the various kinds of respon-
ses, see Friedman (1952) or Phillips & Smith (1953).

8Unfortunately, FaC +CoR couldn't be used in the present study be-
cause so few _S_s gave these types of response.

9The precise rules for calculating the Perseveration Score are

found in Friedman (1952).

36

and organized an organism becomes, the more flexible is its behavior,
thus allowing a variety of situations to be dealt with adequately
(Werner, 1948). The amount of response perseveration decreases with
age until adulthood is attained (Fox, 1956; Heile-Dworetzki, 1956;
Siegel , 1953), reflecting the maturation of capacities for Autonomous
Functioning , mastery-Capetence , and Reality-Oriented Thought .

The investigator proposed, as additional measures of ego deveIOp-
ment , the use of several determinant- and content-based Rorschach
scores. Some of the new scores reflect primarily the Operation Of cog-
nitive processes, much as do Friedman's scores: several others, though,
are more intimately related to other important ego functions , such as
Drive Control and Object Relations. The investigator's selection of
scores which were not used by Friedman or his associates at Clark Uni-
versity was made on the basis Of fulfilling two requirements: (a) the
frequency of the score has been demonstrated to vary as a linear func-
tion of age from early childhood to early adulthood, and (b) the score
reflects the Operation of one or more ego functions. Every Rorschach
variable clearly ascertained to meet these two criteria was included in
the study.

5+1». This score measures the effectiveness of conscious discrimi-
nation and judgment in cognition and is considered by Beck (1945) to be
the foremost indicator Of ego strength. Discreteness of mental func-
tioning is reflected in 23% in that the mature individual is capable of
perception which is not distorted by the intrusion of needs or effects.
§+ %increases with age to a range of 70 to 901. in the healthy adult
(Amos, 1966; Rabin & Beck, 1950; Thetfort, Holish, 3: Beck, 1951), there-
by disclosing the improved Reality Testing capacities of the mature
individual .10

a. Human content reSponses indicate interest in and sensitivity

 

10Ehrtended 3+ % (Schafer, 1954) is a more inclusive measure of Real-
ity Testing than is §+ %since the form quality of multiple-determinant,
as well as of pure form-determinant, responses is considered. Although
published data on the relationship between extended §+ ‘1. and age wasn't
available, at the suggestion Of A. Rabin (Personal commmication. July,
1970), in whose experience the score is age-related, extended 2+% was
analyzed as was {+1. to see whether the former would provide a more sen-
sitive measure of clinical status change than did the latter.

37

to other persons, thus frequency of _11 provides an index Of social ma-
turity (Draguns, Haley, 8: Phillips, 1967). 5 increases with age (Ames,
1966; Ledwith, 1959; McFate & Orr, 1949; Thetford et al., 1951), re-
fleeting a greater maturity in Object Relations.

5. The production of Human Movement reSponses indicates a capac-
ity to delay impulse discharge in deference to eXpression through fan-
tasy (Beck, 1945; Phillips & Smith, 1953). g also implies an empathic
identification with one's fellow man, a prerequisite for mature Object
Relations (Phillips 8: Smith, 1953). Further, the ability to separate
mental image from overt act implies, in Warner's terminology, a discrete-
ness Of mental functions (Meile-Dworetzki, 1956) . Presence Of several
_a_ in a protocol also indicates a capacity to Regress in the Service Of
the Ego. Once it appears in children, 5 increases in frequency until
adulthood is attained (Ames, 1966; Ledwith, 1959; Meile-Dworetzki, 1956;
Thetford et al., 1951).

_P_C:_g‘_+<_2_. Another measure Of impulse control, pertaining to the
regulation Of affective and motoric discharge rather than to ideational
expression, utilizes the quality of color responses. 5 and g; are con-
sidered to be immature color responses and are characteristic of impul-
sive and egocentric individuals. _F_'C_, on the other hand, is the most
mature type of color response, representing a kind of emotional reactiv-
ity which is tempered by a regard for one's social environment (Beck,
1945). _C_ and 39. are located at Opposite ends of a continuum ranging
from a state Of passive perception of color, in which the ensuing per-
cepts are diffuse and global , to a state of active integration Of color
with form, resulting in a well-articulated response; 9; occupies an
intermediate position (Shapiro, 1960).

Studies have generally shown that neither §'_C_, g", nor 9 is linear-
1y related to age. However, the relative proportions of each kind Of
color response are so related. DevelOpmental prOgression is thus re-
flected by comparative increase in {g and decreases in g and g (Ames,
1966; Ames, Learned, Hetraux, 8; Walker, 1952; Hamuendinger, 1953;
Ledwith, 1959; Meile-Dworetzki, 1956; Phillips 8: Smith, 1953; Shapiro,
1960).

Color Balance (£2 : 995), then, is the apprOpriate measure of
changes in the quality of color reSponsivity and has to do with

38

Regulation and Control of Drives and Affects. _FE/Z'S is a more con-
venient form in which to tabulate Color Balance since the wnputed
values are expressed as percentages.

13. The production of POpular responses implies that _S_ is capable
of perception which is in conformity with consensually-based reality.
It can therefore be considered to indicate the degree of socialization
(Klopfer, Ainsworth, KlOpfer, & Holt, 1954; Phillips 3. Smith, 1953). g
increases with age (Ames et al., 1952; Ledwith, 1959; McFate & Orr,
1949; Rabin & Beck, 1950; Thetford et al., 1951). w

_R. Productivity level indicates ability to reSpond to a task
(taking the Rorschach) which requires a measure Of ingemity in a novel
situation, a flexibility Of attitudes, tolerance of ambiguity, and an
expenditure Of effort (Neiger, Sleman, 8: Quirk, 1965). Hastery-
Competence and Adaptive Regression in the service Of the Ego are the
principal ego functions expressed in _R_. Studies have consistently
shown an increase in _R from early childhood to early adulthood (Ames,
1966; Beck, Beck, Levitt, & Holish, 1961; Hemendinger, 1953; Ledwith,
1959; Heile-Dworetzki, 1956; Rabin & Beck, 1950; Thetford et al., 1951).

Scorable Protocol . Whether or not §_ gives a scorable protocol
provides a relatively gross measure of the level of his ego development.
There are several qualities that need to be present in order to Obtain a
record which can be accurately scored. First, is the recognition by g
Of the nature of the task being presented to him by the examiner. An
illustration is provided by Fox's (19 56) characterization of Rorschach
behavior in the very young child as "nagic wand perseveration." In
giving a response to an inkblot, the child is actually reacting to the
testing situation as a Gestalt. Thus, whatever response is elicited to
the first card is likely to be repeated to all ten cards since that re-
sponse obviously "pleased" the examiner. Because it cannot be truly -
said that the child is responding to the blots themselves, his associa-
tions cannot be considered to be legitimate responses. Hallucinations
and other confabulatory responses represent a somewhat similar failure
to discriminate blot-related stimuli from all the other stimuliuboth
inteer and external--which simultaneously comprise _S_'s perceptual
field. A second prerequisite to the scorable record is the possession
of sufficient verbal skills to make possible mutually meaningful

39

communication.with the examiner; This is especially important during
the Inquiry phase of the test. Obviously, the very small child isn't
yet able to use language effectively. Also necessary for the Inquiry
is §fs ability to focus attention long enough so that he can locate his
percepts and recall their determinants. Among the ego functions in-
volved in giving a scorable Rorschach are Reality Testing, Mastery-
Competence, and Thought Processes.

Whenever 25% or more Of a S_'s reSponses could not be scored because
of inability to conduct an adequate inquiry, the protocol was considered
Unscorable.

Summary g§_§pggi§ig empirical predictions
In accord with the previously stated hypothesis that:

...whenever a retest Rorschach indicates movement toward or
away from a more mature mode of ego functioning, there is a
concomitant improvement or decrement in overall adjustment
level...

the following predictions were made:

1. Increase in each of the following scores is associated with
improved clinical status; decrease is associated with declining clini-
cal status:

(a) Mature 3/ Total 2

(b) Mature 2/ Total 2

(c) Ԥf%

(d)
(e)
(f)
(g)
(h)

2. Decrease in each of the fOllowing scores is associated with
improved clinical status; increase with declining clinical status:

(a) Perseveration.Score (g§)
(b) Fabulized Combinations and Contaminated Responses (EEQBQEBQ

3. Change from Unscorable protocol to Scorable is accompanied by
improved clinical status. Change from Scorable to Unscorable is asso-
ciated with a decline in clinical status.

Sub ects
Thirty-three participants in the Michigan State Psychotherapy

IR! ['0 [a It! In:
B
In

40

Project (Karon & O'Grady, 1969) served as‘gs.11 They had been selected
from among first admission patients who not only had been diagnosed as
schiZOphrenic but also were, according to medical examination, appar-
ently free from mental retardation, organicallybbased psychOpathology,
and prior exposure to ECT. Since emphasis in selection was on Obtain-
ing persons who were clearly schizophrenic, there was a tendency for
the more seriously disturbed to be chosen for the project (See Table 4).

 

Table 4

Subject Characteristics

 

 

 

Characteristic E Characteristic EjT
Sex Education
Male 16 None
Female 17 1-5 1
Race 5-9 7
High School
Neg“ 25 I 1 t 12
white 8 m? e 9
High School
Age Cbmplete 10
16-17 3 University
18-22 9 Incomplete 2
23-28 9 University
29-35 6 Complete 1
36-44 5
45-49 1

 

 

 

Each‘g was given a battery of diagnostic tests as soon as possible
after being admitted. Examination was in.most cases done from one to
three weeks after admission. After being tested,Ԥs were randomly
assigned to one Of three groups receiving different kinds of treatment:
modified psyChoanalytic therapy without use of medication, psychoanalyt-
ic therapy with use of drugs, and supportive therapy with use of drugs.
22.8.22

The Rorschach and a psychiatric interview to ascertain clinical
status (CS) were given along with the TAT and several tests of intelli-
gence. All tests were given to any oneԤ.on the same day.

 

11It wasn't possible to Obtain complete test data for 2 Of the 35

project'Ss, so they were deleted from the present study.

41

Each §_ was re-examined six months after his initial testing. Each
was tested again after another six-month interval, and final examina-
tion was administered after an additional eight months. Thus, each S
was tested four times: right after admission (T1), at six months (T2),
twelve months (T3), and at twenty months after admission (T 4).

A doctoral student in clinical psychology administered the Ror-
schach under standardized conditions to every §_ for all save the last
testing. He was no longer associated with the project at that time, so
another clinical graduate student gave the twenty-month Rorschach to al-
most all gs .12

The protocols were scored blindly by the investigator and another
doctoral candidate in accordance with the Beck system (Beck et al.,
1961). The form quality tables in Hertz (1961) and Phillips and Smith
(1952) were consulted whenever the Beck £4» and 5- table provided insuf-
ficient guidance. Special care was taken to insure that the scoring
was consistent from T1 to T4 as well as from one §_'s record to another.
The additional scoring necessary for assessing Developmental Level
scores was made in accordance with the rules established by Friedman
(1952) .

An eXperienced psychiatrist conducted all the interviews for each
_S_. He was instructed to delve into each patient' s adjustment status in
terms of 11 pre-determined criteria of mental health .13 Each interview
was recorded on tape. TO prevent the judges from gaining knowledge
about the nature of the treatment given to any S, all remarks made con-
cerning therapy were edited from the tapes. Random deletions were made
in the tapes of 33 not making references to their therapy.

Clinical _S______tatus (CS) evaluations

The CS Of each _S was determined by two judges, both doctoral

 

12Several Ss had to be tested at T by a third graduate student and

were therefore- dropped from further consideration in order to maintain
a homogeneous examinerbsubject relationship for all 83.

13Ability to take care Of self, ability to work, sexual adjustment,

social adjustment, absence of hallucinations and delusions, relative
freedom from anxiety and depression, amount Of affect, variety and
spontaneity of affect, satisfaction with life and self, achievement of
capabilities, and benign rather than malevolent effect on others.

42

candidates in clinical psychology, who listened to the tapes at a later
date. They evaluated each _S in terms of the 11 indicators of mental
health by integrating the latter criteria into a single, global judg-
ment.

_S_s were rated in accordance with the scaling technique for unidi-
mensional judgments devised by Estavan and modified by Karon andO'Grady
(1970).

The CS rating technique resulted in each _S_‘s being assigned a
quantitative score. A hierarchy of values for all _S_s was obtained for
each of the four times they were interviewed. Thus, within each hier-
archy, a given §_'s mental health status relative to that of any other _5_
was easily ascertainable.

Intrarater consistency and interrater reliability were both quite
high. Split-half E's for the former measure were .83 and .79 at T1 and
remained as high on subsequent ratings . The values for interjudge re-
liability were .87 (TI) and .82 (T2). Moreover, the predictive valid-
ity of the interview-based ratings was demonstrated in that they cor-
related significantly with the number of days of hospitalization during
the six months following each CS evaluation (-.71 for T1, -.64 for T2) .
The latter criterion traditionally has been felt to be the most obvious
and socially useful indicator of recovery in psychotic patients.

Neither the CS raters nor the Rorschach scorers had personal con-
tact with §s or had knowledge of the results of each others' ratings or
of the kind of therapy given any :3, thereby eliminating information
from sources outside the imediate material necessary for evaluation.
The raters and scorers did have access , however, to basic background
data on each S which included age, sex, marital status, occupation,
educational level , and brief etiological sketches.

Analysis of data

Preparation of the data preliminary to statistical analysis in-
volved two steps: (a) obtaining measures of CS change, and (b) obtain-
ing measures of Rorschach score change. The actual hypothesis testing
also involved two stages: (a) testing for association between Rorschach
change and CS change, and (b) analysis of the influence of certain un-
controllable factors upon the relationships of interest.

Measures _a_; _C§ charge. Regression toward the mean is a notorious

43

by-product in the counpilation of test-retest data on any variable which
cannot be measured with perfect reliability. Thus, difference scores,
which involve the simple expedient of subtracting each individual's Tl
score from his T2 score, are inappropriate measures of change. Among
§_s who initially score below the mean, the amount of actual gain upon
retest would be overestimated by such a procedure, whereas a decline in
score upon retest would be underestimated. For _S_s who initially score
above the mean, the deviations of observed change from true change
would be in the opposite directions.

A scheme devised to prevent error due to regression toward the
mean involves the use of regressed scores. Meltzoff and Kornreich
(1970) recomend the use of the delta score transformation, a variant
of the regressed score}4 Not only is the resulting measure of change
free from regression error; it is also independent of the level of the
initial score.

A necessary condition for the calculation of regressed scores is
that the regression of retest scares on initial scores be linear. A
scatter-plot revealed that the regression of CS2 raw scores on C81 raw
scores was crudely V-shaped, (i.e., the variance of CS2 scores increased
as a function of C81) making it imperative that a scale transformation
be applied to obtain a linear relationship. Inspection of the respec-
tive distributions of scores revealed the identity of the culprits--a
positive skew in C51 and a strong negative skew in C82. There was no
skeamess to either the CS3 or the CS raw score distributions. Since

4

the regression of CS4 on CS3 was linear, the test-retest reliability

coefficient needed to derive the regressed scores was readily computed.
The ass1mtption of a normal papulation distribution for CS is a

reasonable one, therefore the C81 and CS2 scores were normalized by

 

14

A = z - I Z
--cs2 12‘--<:sl

5's are individual CS values expressed
in standard score form.

512 is the correlation between CS

and CS2 scores .

The resulting CS change score (A) denotes the amount of movement
relative to that of the other SS and implies nothing about overall
change in the mean of the total group.

1

44

means of the T-scale (Edwards, 1954; Guilford, 1956; HcNemar, 1962).
The T1 to T2 correlation and the regressed scores were then derived
from the normalized scores. The test-retest coefficient for T2 to T3
was based on the C52 T scores and the C33 raw scores.

CS stability was lower between T1 and T2 than it was between
either of the other time intervals (512- .366; 523: .563; £34: .759).
Due to the lack of complete information on several of the original 35
_S_s, _R_- 33 for CS change calculations from T1 to T2, _R_:- 32 for T2 to T3,
and E- 30 for T3 to T4. .

Measures 2: Rorschach score change. Since Rorschach scores are
also affected by regression toward the mean, the delta score transfor-
mation was applied to them as it was to CS measures.

Listed in.Table S are the test-retest Pearson Efs applied to T1,
T2, and T3 scores in order to obtain the regressed scores used in the
respective delta score transformations from T1 to T2, T2 to T3, and T3
to T4. In general, Rorschach score stabilityb-like CS stabilityb-was
greater between 6 and 12 months and between 12 and 20 months than.be-
tween initial and 6 months. This occurred in spite of the fact that a
new examiner administered the 20-month Rorsdhachs and would suggest

that greater change occurred during the early months of the study.

 

Table 5

Test-retest Coefficients of Correlation for Rorschach Scores

 

 

 

F: Tl‘Tz T2"1'3 Ta'T4
3”“ 1“. r .L‘ .r. .1! 5
Mature w/A11 g 23 .46 25 .59 22 ‘ .66
Mature D/All _a_ 23 .18 25 .69 22 .74
gg 23 .68 25 .80 22 .80
9% 26 .53 3o .55 25 .69
g 27 .62 3o .75 26 .78
£4. 27 .62 3o .76 26 .77
Egg/25; 19 .26 22 .42 19 .29
g 27 .79 30 .80 27 .75
3 27 .54 30 .79 25 .76

 

 

 

45

Regressed scores for g and 5 were based on the square roots of
their raw scores so as to maximize test-retest change at the lower end
of the raw score continuum. In this way, allowance was provided for
the generally accepted view that a shift from O to 1 has greater clini-
cal significnace than does a shift from, say, 4 to 5.

The measurement of 373/22 change presented several problems.

Since the denominator (total number of color reSponses) is almost al-
ways quite small, any change in the numerator (number of _F_'_C_ reSponses)
causes a big change in the color balance, a phenomenon which typically
leads to highly unreliable measures of score change over time (Cronbach,
1949). In fact, none of the test-retest correlations for color balance
were significant, in distinct contrast to the other Rorschach score
correlations, all of which were beyond the .005 level of significance
(except for Mature 2/ Total 2 from T1 to T2).

Thus, it might be difficulttn obtain significant 3's between 22/22
change and CS change-«even if there were a strong relationship-oif the
low test-retest correlations were due to unreliability rather than to
genuine changes in the attributes represented by color balance.

5 for _Fg/gg change was lower than for the other Rorschach scores.
Some _S_s gave no color reSponses at all. Others gave color reSponses at
one testing but none on the other occasion, thus making it impossible to
ascertain color balance change.

Although Rorschach protocols were obtained from 33 S3 at both T1
and T2, the records of several of the more severely disturbed patients
consisted primarily of unintelligible or purely confabulatory reSponses
and were thus unscorable. It was necessary to exclude these _S_s from
consideration for score changes. Unfortunately, as in so much of the
research on psychotics, this led to a curtailment in the range of psy-
chopathology covered by the data of the study. Nevertheless, it is
believed that the present study uses a greater proportion of persons at
the lower end of the clinical status continuum than is typical of most
research using psychotic gs. In addition, the study of CS changes
among _S_s who change from a scorable to an unscorable record, or vice-
versa, does include precisely those individuals who had to be excluded
from the analyses of separate score changes.

46

On some protocols the ambiguity of one or several associations
prevented accurate scoring of _B_I_._, _I_>_S_, _R, gig/2g, and §+% , yet g, g, and
g could be scored with confidence. An additional qualification for
inclusion in the testing of predictions was applied to ES and QE change.
Each protocol had to contain at least ten reSponses, since such has
been the practice in previous studies involving these scores in the
measurement of ego regression.

A further reduction in g for the testing of Rorschach and CS change
from T3 to T 4 was made necessary because a third examiner had been used
for six of the _S_s at the time of the final testing.

The prediction concerning §_a_C_+_C_g_R couldn't be tested, for the
incidence of the two responses was too infrequent to make statistical
tests plausible.

Testing 9}: predictions. The strength of association between CS
change and each Rorschach score change was determined by Pearson _r_.
Thus, both the direction and the magnitude of change for each S_ were
considered. The only assumption that need be met whenever g is used to
test the significance of relationship is that of linearity (McNamar,
1962), so scatterplots were made to check for possible nonlinear rela-
tionships.15 None were found.

The prediction regarding shifts from Unscorable to Scorable proto-
cols, and vice-versa, was tested by means of Student's E.

ReEication 2: M. An oft-forgotten aSpect of research is that
of replication (Smith, 1970). This neglect is unfortunate because re-
plication is an especially appmpriate procedure for occasions on which
a large number of significance tests are made. Spurious findings (i.e.,
those due purely to chance) can be readily identified, for they don't
replicate. The present study provided what Sidman (1960) has called
"intrasubject" replication, i.e., repeated observations on the same _S_s.
After the initial (T1) and six-month (T2) data were tested, the same

 

lslt is usually assumed that normality and homoscedasticity are re-
quired for proper use of 5;. However, Pitman (1937) proved, byuse of the
randomization criterion, that 5 can be used as the most efficient non-
parametric test for association, and that the resulting coefficient
from any shape distribution can be tested for significance by referral
to the standard tables for 5 (the distribution of _r_'s generated by ran-
domization closely approximates the distribution of 5's generated by
normal curve theory).

47

procedures were applied to the sixpmonth and twelve-month (T3) data,
and these procedures were then reapplied to the twelve-month and twenty-
month (T4) data.

Since there were three independent tests of each prediction, it was
permissible to combine probabilities.16’17 Obviously, combining the
three'p's increased the power of the test.

 

16For any single replication, the correlation between.Rorschach

score change and CS change was not influenced by the correSponding
relationship between the two variables at any previous replication,
particularly since regressed scores were used as the measure of Change.

17Significance levels are uniformly distributed by chance. There-
fore, the cumulative distribution of the three significance levels
emanating from the independent tests of each prediction can.be compared
to the theoretical cumulative distribution of probabilities by means of
the KolmogorOVbSmirnov test to determine whether the observed distribu-
tion departs from the distribution which results from the operation of
chance factors alone.

RESULTS

The findings are summarized in Tables 6, 7, and 8. Four of the
ten predictions-~those pertaining to Perseveration Score (_P_S_) , 51%, S,
and the scorability of a protocol--were confirmed.

None of the correlation coefficients for the nine Rorschach scores
during the Tl-T2 interval reached the generally accepted level of sig-
nificance. The situation was quite different, however, for the second
(Tz-TB) replication: four coefficients (those for E, S+%,S, and S)
attained significance beyond the .05 level, and one (Mature 2/ Total S)
fell just short of it. During the T3-T4 replication, none of the nine
coefficients reached significance.

Upon combining probability levels, change in SS, in S+%,18 and in
S were found to correlate with charge in C3 (the respective 2'5 were
(.02, (.01, and <.02). For each of these scores, the coefficients for
all three replications were consistent in direction and magnitude with
each other and with the prediction. Such was not the case for the
other predictions (save for S, wherein the coefficients were consistent
across replications but were too low to enable rejection of the null
hypothesis).

Mean CS gain for Ss changing from Unscorable protocol to Scorable
protocol upon retest was, as predicted, significantly greater than for
Ss who shifted from Scorable to Unscorable (see Table 6).

Influence gi: extraneous variables. A question could be raised as
to whether the observed correlation between charge in a given Rorschach
score and change in CS might actually be attributable to their sharing
a common relationship with a third factor rather than to a causal rela-
tionship between the two variables of research interest.

 

18The correlations between extended F+% and CS charge were general-

1y similar to but lower than were the coefficients between gm change
and CS change.

48

49
Table 6

Clinical Status Change upon Shift from Unscorable to
Scorable Protocols and from Scorable to Unscorable

 

 

2

 

 

Direction of charge 3 Z S - j_: .
Unscorable to Scorable 6 +28.l7 3710.57 2.22.
Scorable to Unscorable 2 -204.50 18818.00

 

 

Note.--Test of significance is one-sided.
I"Significant at .025 level.

 

Information was available on several demographic variables-“includ-
ing sex, race, marital status, social class, and age-so a more detailed
analysis of the relationship between Rorschach change and CS change was
carried out.

Socioeconomic status (SES) was a significant correlate of CS change
from T to T2 (5;: .56, p<. 005) but not of CS charge from T to T

l 2 3
(r=-. 05) or 2from T to T4 (1:... 02).19 The presence of a direct rela-

tionship between soot3:ial4 status and amount of CS improvement was due to
correlation of SES with CS at T2 (5: .52); the correlation between the
two variables was absent at T1. (gs-.01) and quite low at T3 (5- .23)

and at T4 (5- .24). None of the other four variables were related to

CS charge at any of the replications.

Since SE3 did correlate with CS charge from T1 to T2, the correla-
tions between Rorschach score change and CS charge were recomputed,
holding constant the effects of $35. A similar adjustment of coeffi-
cients for the potentially confounding effects of an outside variable
was necessary in the case of the relationships between changes in S, S,
and g and change in cs for the Ti, to T3 test of predictions. Although
changes in these three scores were correlated with change in _R_ during

the other replications as well, T2 to T3 was the only occasion on which

 

19SES was determined by use of a modified version of the Hollings-

head Index of Social Position (Hollingshead 82 Redlich, 1958). The
scale was based on a weighted combination of S's level of occupation
and of education. Although in the original version the higher the
Index rating, the lower the social standing, the mnnberirg was reversed

in the present study in order to make correlations of CS with high SES
be positive in sign. Since the distribution of SES scores within the
sample was quite skewed (negatively), they were normalized before being
related to CS values.

50

change in productivity was correlated with CS change and therefore the
only occasion on which the three coefficients would be appreciably af-
fected by charge in 5.

The combined significance of the relationship between ES change
and CS change didn't change as a result of correction for $85. Sig-
nificance for §+% change, however, fell from <.01 to (.03 while sig-
nificance of _R charge fell from <.02 to <.04. Correction for 5 change
caused a decrease in the coefficients for S, E, and S but did not alter
the combined 3 for any one of the three scores.

The SES- and _R-corrected coefficients are included in Table 7.

Table 7

Correlations between Rorschach Score Change and Clinical Status Charge

 

 

 

k
Score T1 - Tza T2 - T3 T3 - T4 Comb. 2
E .5. Pl 5 l“. E.
Mature 1 / A11 5 23 -.03 25 .21 22 .22 n.s.
Mature 2] All S 23 -.26 25 .31 22 .24 n.s.
3;; 23 -.26 25 -.16 22 -.22 .02
3+7; 26 .11 30‘ .40“ 25 .23 .03
g 27 .02 30 .06b 26 .05 n.s.
g 27 .14 30 .48b" 26 -.19 n.s.
_99 mg 19 -.14 22 .06 19 .22 n.s.
g 27 -.21 30 .33”: 27 .04 n.s.
5 27 .10 30 .34‘ 2s .14 .04

 

 

 

Note.--A11 2's are one-tailed tests of significance.
apartial correlation coefficients (SES controlled).

bPartial correlation coefficients (5 controlled).

.2 < .05-
on
p < .025.

Analysis 9}: drug effects
The original intent of the study was to look for Rorschach score

changes which parallel CS change without qualification as to the cause
of the latter. In view of the afore-mentioned evidence suggesting that

51

various kinds of therapy produce differential personality change, how-
ever, it was considered advisable to check also for the existence of
specific treatment interactions with Rorschach change and CS change.
The most pronounced treatment difference had to do with the use of
medication. Drug usage was avoided whenever possible in the therapeu-
tic program of some Ss; for other Ss, drugs served as an adjunct to
psychotherapy or even as a substitute for it.

Every S was on phenothiazine at the time of initial testing, but
13 of the 33 were no longer on drugs at T2. Rorschach change and CS

charge from T to T2 were correlated separately for the 13 persons with-

drawn from medication sometime during the test-retest interval ("Drug
to non-drug"[D-ND] S3) and for the 20 remaining on drugs throughout the
interval ("Drug to drug"[D-D] Ss). From T2 to T3 nine more Ss had been
taken off drugs, so Ss were now separated into three groups: D-D, D-ND,
and ND-ND (not on drugs at either testing). By T 4, only one additional
S had been removed from medication and one had been reinstated on drugs.
Thus, for T3 to T4 there were two classes of S3: D-D and ND--ND.20

The findings according to drug status are presented in Table 8.
Some striking differences in the nature of the relationship between
Rorschach change and CS change appeared among the three groups.

Four of the Rorschach variables followed a similar pattern: signi-
ficant relationship of score charge to CS change for ND-ND Ss but not
for D-D nor for D-ND Ss.

1. ES. The negative relationship between PS change and CS change
observed for the group as a whole was even stronger when ND-ND Ss were
analyzed separatelyudespite the reduction in E ensuing from elimina-
tion of all the other Ss. Not only was the combined 2 highly signifi-
cant (.01) , but the correlation for the T replication was signifi-
cant by itself (2- .02).21

2. 2. Although i: change for the entire sample wasn't associated

3'T4

 

20The two Ss charging drug status between T and T4 were excluded
from the D—0 and ND—ND groups but were included in the analysis of all
Ss as a single group.

21The formula which was used whenever two, rather than three, inde-

pendent experiments were combined is 2
(21 + 22)
2 O

 

52

Table 8

Correlations of Rorschach Score Charge with Clinical Status Change
in Accordance with Drug Status of S5

 

 

 

f
'1' - T T - T T - T
Score 82:29.15 1 2a 2 3 3 4 Comb. p
l". E I! E. .19. E

Mature 31/211 3 0.0 12 -.40 7 .27 6 .38 n.s.
D-ND 11 .23 8 -043 nos.
ND-ND 10 .42 14 .15 .09

Mature S/All 2 D-D 12 -.O7 7 .32 6 .17 n.s.
D'ND 11 -027 8 .21 n.s.
ND-ND 10 .48 14 .23 .05

£3- D‘D 12 .09 7 -015 6 .13 “.8.
D‘ND 11 -030 8 -005 “.8.
ND-ND 10 -.39 14 -.S3” .01

§_+ % 0-0 14 .10 10 .63” 7 .27 .05
D’ND 12 .31 9 .08 11.8.
ND-ND 11 .21 16 .12 n.s.

fl D'D 15 -036 10 -025 7 .03 nose
D-ND 12 .45 9 .64b‘ .01
ND-ND ll .01b 17 .08 n.s.
0.110 12 .53- 9 .26b .05
ND-ND 11 .6013. 17 .16 .05
mm 8 -009 8 .26 n.s.
ND-ND 7 -312 14 .08 n.s.

E D-D 15 -016 10 .49 8 .04 11.3.
D-ND 12 -.15 9 .12b n.s.
ND-ND 11 .34b 17 .39 .03

_R- D'D 15 .36 10 .06 7 -036 nos.
D‘ND 12 -001 9 044 n.8-
ND-ND 11 .42 16 .19 .06

 

 

Note.--A11 2's are one-tailed tests of significance.
aPartial correlation coefficients (SES controlled).

b

cDue to the small 5, 5 was not calculated.

'2 (.05.
"p < .025 .

Partial correlation coefficients (5 controlled).

 

53

with CS change, the'predicted positive relationship was confirmed for
ND-ND gs (comb. 2:- .025). In addition, the correlation between E charge
and CS change during the T3-'1‘4 replication approached significance (2:-
.06).

3. 92. The relationship in ND-ND _S_s between change in Nature 111/
Total )1 and change in CS was in the predicted direction (2- .09), al-
though not as strong as was the case for change in Nature 2/ Total _Q
(comb. p- .05) .

4. 3. Combined p for the correlation of productivity change with
CS change approached significance for ND-ND gs (£2 .06).

Not only was the prediction regarding 5 change and CS change con-
firmed for ND-ND ES (2:: .05); it was confirmed for D-ND _S_s as well (3-
.05). Moreover, each of the non-drug (ND-ND and D-ND) groups had one
replication which attained significance: Tlm'l‘2 for D-ND (33- .05) and
Tz-T3 for ND-ND (2- .03). As was true for 5 change, correlations for
D-D gs were high but inconsistent across replications.

A situation somewhat different from that found in the preceding
five scores characterized the influence of drug status on 5 change.
Confirmation of the prediction concerning CS change appeared in D-ND
group instead of the ND-ND group. In fact, combined 2 for D-ND S5 was
highly significant (p: .01), and the correlation at TZ-T3 was signifi-
cant by itself (2' .05), whereas the coefficients for the three D-D
replications and for the two ND-ND replications were inconsistent.

§+% provided a different type of interaction with drugs. Correla-
tion of 3+1» change with CS change for the sample as a whole appeared to
be attributable primarily to the Dan group rather than to the non-drug
groups: the former was correlated with CS change at T2-T3 (2: .025)
and just reached the conventionally accepted level of significance for
the combined 2's (.05). All four coefficients for D-ND and ND-ND §_s
were in the predicted direction but were insufficiently strong to at-
tain significance even after probabilities were combined.

_F_C_/2£ change was the only variable for which there was neither
a relationship to CS charge for the entire sainple nor for a particular
drug or non-drug group.

It is interestirg to note that a strikingly greater number of the
original predictions from ego regression theory were confirmed in S5

54

under ND-ND conditions than in _S_s under D—ND or D-D conditions (see
Table 9).

Table 9

Comparison of Drug Status Groups on Number
of Predictions Attaining Significance

 

 

No. of predictions

”mg “with“ with combined 2

 

< .10 > .10
Nondrug - Nondrug 6
Drug - Nondrug
Drug - Drug

 

 

 

The drug and non-drug groups for each replication were similar to
each other in age, sex, 883, marital status, race, and initial level of
CS (i.e., at T1 for the first replication and again at T2 for the sec-
ond and at T3 for the third). Group differences cannot, therefore, be
accounted for by demographic differences but would appear to be due to
whether or not as were on medication.

9.1122: 2!.- _a_—stats

Changes in ES, §+%, and 3 correlated, in accord with prediction,
with changes in CS, i.e., uprovement was acconmanied by increased {+75
and g and by decreased _13_S_. There was no relationship between change in
CS and charges in 31, g, a, 59/89, or in either DL score. gs showing
change from an Unscorable protocol to a Scorable record showed signifi-
cant CS gain relative to §_s showing the reverse trend in scorability.

Examination of S3 in terms of drug usage revealed that, with the
exception of 3+% , the original predictions were confirmed only among
individuals who either were not on drugs at any time during the test-
retest interval or who had been taken off medication before the retest.
Changes in six scores, g_S_, Mature 31/ Total 1,22 Mature 2/ Total 2, g,

 

22A'thhough the predictions regarding Mature W/ Total W and R did

not quite reach the conventionally accepted .05 level of significance,
they have been included with the other, clearly significant, scores
since they were mucher closer in p to these than to the three scores
having negative findings in ND-ND §_s.

55

g, and 3, were correlated with charges in CS among ND—ND gs; changes in
two scores, 51 and g, were correlated with changes in CS among D—ND _S_s.
The correlations between §+% change and CS change were consistent across
all replications for all drug and non-drug groups, but the D-D group
was the only one for which the combined B's reached significance. With
regard to the scorability of a record, 5 was insufficient for analyzing
the three drug status groups separately.

DISCUSSION

The results provided qualified support for the hypothesis that im-
proved clinical status (CS) is accompanied by movement toward a more
mature level of ego functioning. The justification for qualifyirg the
conclusions regarding the validity of the hypothesis will be apparent
once the results are considered in detail.

Charges in four of the ten testable23 measures of ego regression
and progression24--Perseveration Score, Scorability, §‘_+% , and gucovar-
ied with charge in CS for the entire sample as predicted. The inter-
pretation of each result is discussed in descendirg order of the
strergth of correlation.

Perseveration _Sgggg (3:3) . As CS improved, gs decreased; as CS
declined, 5% increased. Hence, §_S_ not only differentiates varying de-
grees of mental disorder at a given point in time (as previous research-
ers have shown) but has now been demonstrated to indicate changes in
the level of pathology over time, i.e. , it accurately measures within
individuals as well as across individuals. §§_ is interpreted as an
index of the overall adaptability of the ego. The lower the score, the
greater the ability to modify behavior when the situation calls for it.
The higher the score, the more apparent _S_‘s "dulling of reaction to
stimuli." (Phillips & Elmadjian, 1947). Small children are limited in
their capacity to respond in an appropriately differentiated manner to
an ever-charging envirorment; schizophrenics are likewise handicapped
by behavioral rigidity. With regard to this particular measure of ego
functioning, then, schizophrenics became more child-like as the level
of pathology increased. And, conversely, recovery for them was akin to
the normal maturation of the ego in the growing child.

 

230m of the original predictions, involvirg E29. +CoR, couldn't be

tested because of the rarity of the reSponses.

24"Progress'ion" herein refers to a process which is the reverse of

regression.

56

S7

Scorability. Test-retest charges in Scorability of a protocol
have never before been examined as signs of change in clinical status.
Although but eight _S_s shifted from Unscorable to Scorable or vice-verse,

this variable differentiated improvement from decline quite well. §_s
changing from Unscorable to Scorable gained in CS while _S_s changing
from Scorable to Unscorable dropped considerably in CS. The inability
to give a scorable protocol is characteristic of extreme psychopathol-
ogy in adults and is also typical of very young children. It implies
the impairment of several key processes, such as the focussing of at-
tention on the task of takirg the Rorschach, discrimination of blot-
related stimuli from other stimuli, real or imagined, and coumunication
of information between examiner and _S_. Like ES, Scorable-Unscorable
was shown to represent a dimension of ego functioning, albeit quite
gross, which reflects CS differences across individuals as well as CS
change within individuals over time. Obviously, such a measure is use-
ful only when patients coverirg a wide range of pathology are tested,
for just as unscorable records are rarely encountered in healthy per-
sons of school age or older, they usually occur only among the most
disturbed of schizophrenics.

The problem posed by unscorable responses and protocols in Ror-
schach research is rarely mentioned in the literature, suggestirg that
most studies of schizophrenics are based on samples consisting of the
relatively more healthy and coOperative patients. It is clear from the
findings of the present investigation that an unselected group of schiz-
ophrenic patients will contain an appreciable number who are most dif-
ficult to test. By restricting the range of the population sampled,
the researcher may spare himself the problw of how to deal with proto-
cols which cannot be scored or interpreted with confidence. He may
also, however, spare himself the Opportunity of obtaining significant
findings for certain Rorschach variables. Most importantly, his find-
ings may hold only for an unrepresentative subgoup cf schiwphrenics.
This unfortunately is probably true not only for studies of the Ror-
schach but also for a good deal of the experimental literature on
schizophrenia.

2+ %. The positive findings with regard to §+% corroborate previ-
ous claims of its validity as an indicator of clinical status change in
psychotics. CS decline was marked by a drop in _E_‘_+%, recovery by an

58

increase. Disturbance in the reality testing function of the ego is
one of the hallmarks of psychosis, so it is not surprisirg that the
degree of accuracy in perception has been found to be intimately re-
lated to charges in adjustment hvel. The capacity for a critical and
objective appraisal of stimuli is vital to the attairment of high £31.
The mental processes of the small child are still governed to a consid-
erable extent by the pleasure principle or (to put it into Wernerian
rather than Freudian terminology) by syncretism. Severe psyChOpatho-
logy is similarly characterized by inability to accurately match con-
ceptual images with perceptual reality. Thus the decline in 2+1. atten-
dant upon worsenirg CS represents a regression to an infantile mode of
dealing with the objective world. While there is an optimal §+%above
which increase represents movement toward pathology, such findirgs are
more a problem with a less severely regressed population. For §_s in
the present study 33% was generally low, and increase represented ap-
proaching the optimal level .

_R_. The correlation of 3 charge with CS change (improvement was
accompanied by increased productivity, decline by a decrease in the
production of responses) was especially noteworthy inasmuch as the re-
lationship, although clearly consonant with regression theory, had not
been found in any of the studies of schizOphrenics cited previously
(Halpern, 1940; Rickers-Ovsiankina, 1954). Regression error and small
_bl may very well account for the negative results in 3 in the former.
The latter is typical of the many studies which excluded §_s gettirg
fewer than ten reSponses. Although it may be feasible to eliminate the
uninformative or unreliable records of highly defensive _S_s when examin-
irg charges in many Rorschach scores (Phillips & Smith, 1953), there
are some variables in which test-retest change can be meanirgful even
in severely constricted records. The present study included §_s who
failed to give the minimally "acceptable" number of responses, and it
showed that change in _R does indicate CS movement.

Beck (1952) spoke of R as giving an indication of the amount of

"liberated drive” at a S's disposal. Thus, changes in the general ade-
quacy of functioning were correlated with changes in the quantity of
controlled energy available for constructive purposes (i.e., giving
responses). A certain flexibility of response and tolerance of ambi-
guity are necessary to enable g to utilize his perceptual-cognitive

a”!

59

resources with efficiency. The normal adult has develOped this capac-
ity which is lackirg in the child. A drop in productivity therefore
represents a reversal of the normal develOpmental sequence for this ego
trait.

Thus improved mental health in the schizophrenic patients of the
present study appeared to be associated with greater flexibility in
behavior, more effective utilization of personal resources, and increas-
ingly more realistic and objective perception and thinkirg.

Although the above results are suggestive, judgment as to the valid-
ity of the ego regression hypothesis becomes more conclusive only upon
investigation of a most striking phenomenon-«he drug effect.
Conclusions _ig 1:351:13 9_f_ drug status

The data offered impressive evidence that improvement or decline
in mental health is accompanied by quite different Rorschach score
changes dependirg upon whether or not a person was on medication during
the time in which the changes occurred. _S_s not gettirg drugs durirg
the test-retest interval (ND-ND) had a greater umber of predictions
which were confirmed than did _S_s who were on medication throughout the

interval (13-0) or who were withdrawn from medication at some time
during the interval (D-ND).

The scores related to CS change under the three drug conditions
are discussed below, by group and in the order of significance of the
findings.

§.§ 395 gr; medication (912-112). Despite the considerable reduction
in E necessitated by analysis of each drug status group separately (and
the consequent reduction in the power of statistical tests), six of the
nine predictions were confirmed for the ND-ND group.

1. 3S. As was true for the entire sample, the results for ES

 

were more highly significant under ND-ND conditions than were the re-
sults for any other score. The negative correlation between ES charge
and CS charge was, in fact, more striking in ND-ND Ss than it was when
_Ss under every drug condition were considered together.

2. 5. Although charge in g didn't correlate with change in CS
for the sample as a whole, it did among Ss who experienced charge with-
out the use of medication. SchizOphrenic patients who improved gave
more Popular reSponses on retest whereas patients who got worse gave

l‘

) .

60

fewer. Increased 3 indicates a tendency toward greater conformity to
the group standards of perception and conceptualization, an important
aspect of becoming socialized. Also, increase in E signifies better
orientation toward reality--a significant sign of improvement in some-
one who formerly was disassociated from reality. The child, for whom
the socialization process is not yet complete, typically scores low in
g. In schizOphrenics, g is also low, reflecting the failure of the
psychotic patient to meet the minimal requirements of cognitive and
behavioral conformity which are necessary to function effectively in a
normal social enviroment. The test-retest data thus demonstrated that
worsening clinical status involves regression in the type of behavior
measured by 3.

Again, as with 5 change, the present study's findirgs with reSpect
to 2 seem to contradict the experiences of previous investigators who
were unanimous in concluding that g doesn't change in relation to alter-
ations in clinical status. For instance, Phillips and Smith (1953) had
stated that "g is remarkably impervious to the impact of psychOpathology."
It appears likely, however, that most Ss were on medication in the other
studies of 2 charge in schizophrenics. Since the present study demon-
strated that the relationship of 3 change to CS charge was manifest
only when ND—ND _Ss were examined separately from the other Ss, there is
really no contradiction with previous findings.

3. 5. SS not receiving medication during the test-retest interval
showed increased g with CS improvement and decreased 5 with CS decline.
Recovery therefore appears to be associated with increased maturity in
important ego functions such as the ability to delay discharge of ten-
sion and the capacity for empathic identification. Another consequence
of improvement is a greater proclivity to engage in fantasy. Thus, CS
decline in schizOphrenic patients not on drugs represented a reinstate-
ment, in the adult schizOphrenic, of the relatively imature status
that these ego functions possess in the child.

4. Mature 12/2932}; 2. This and Nature E/ Total 1 have been the
most frequently used Rorschach variables in the genetic studies of
mental disorder. Mature 2/ Total 2 increased with improved health and
decreased with deterioration in the S3 not on medication. This 9;.
score, although similar to 33% in that both provide a measure of the

61

adequacy of reality testing, also involves qualification of the respon-
ses in terms of the differentiation and integration that they show.
Psychopathology in the adult not on medication has thus been demon-
strated to involve a retreat from the relative complexity of percepts
attained by the healthy adult to the less differentiated and less
organized responses typical of the normal child.

5. 3. As was the case with ES, 5 charges with CS charge in ND-ND
_Ss alone as well as for the entire sample. Persons improving in psy-
chological health become more productive, those with increasirg path-
ology become less productive. The fact that the significance of cor-
relation for ND-ND S5 was not as high as it was for all S3 is attribut-
able to the decrease in E, for the coefficients in the former group
tended to be higher than they were in the entire group.

6. Mature 31/2233}; 1!. Improved CS was accompanied by an increase,
decline in CS by a decrease. The meanirg of charge in this score is
identical to that for the other _DE score, except that one score is
based on the quality of Whole reSponses, the other on the quality of
Detail responses.

The relationship between movement in {+7. and in CS among ND—ND Ss
was in the predicted direction but not strong enough to be significant.
The coefficients for the remaining two scores, a and _Fg/ng, were not
in the least suggestive of relationship to CS charge.

Thus, improvement in _S_s not on medication seems to involve greater
adaptability, more effective use of personal resources , increasing con-
formity to culturally established norms of perception and cognition,
greater capacity to delay impulse discharge in deference to expression
of inner tensions through fantasy, and increased couplexity and inte-
gration'of cognitive processes.

_S_s SE medication (9;?) . One of nine predictions was confirmed.

2+1: . This was the only score related to CS change when _S_s remained
on drugs. Since §+% increased with CS gain and decreased with CS loss,
change in reality testing coincided with change in adjustment. Dynam-
ically speaking, change as a function of improvement did not occur in
the other ego functions measured by the Rorschach.

S3 withdrawn m medication (9:19). Two of nine predictions were
confirmed.

62

l. 5. Although charge in g didn't coincide with CS charge in the
entire sample or under the separate ND-ND or D-D conditions , it was
very highly correlated for the D-ND condition, an intriguirg develOp-
ment. Improvement was reflected by increased 5, increasirg maladjust-
ment by decreased 3;. Recovery from schizOphrenia thus entailed greater
interest in and sensitivity toward other persons. Declinirg CS involved
retreat to a less mature mode of dealing with people.

2. E. This was correlated with CS charge in both ND-ND and D-ND
Ss. Despite this, a charge didn't relate to CS charge in the entire
sample because the coefficients for D-D _Ss were high but inconsistent
in sign across replications. With respect to E, the nature of CS
change in the D-ND group seemed to be similar to charge in the ND-ND
group.

_FE/zg was the only score examined which did not relate to CS
change under any drug or nondrug condition. Its failure may be due to
its being unreliable as a charge score (see previous marks about £2 /
22 under Method) . Interestingly enough, two of the studies of schizo-
phrenics reported in the Review found _P_g : 2+9 to be a consistent in-
dicator of charge in mental health.

Conclusions Legardim £139 effect 9}: medication

The findings demonstrated that regression and progression in cer-
tain ego functions is correlated with gains and losses in clinical
status. The data further disclosed that the ego charges accompanying
alteration in adjustment are treatment-Specific: the ego functions
represented by ES, _R_, g, and _D_I_. regressed and progressed in direct re-
lation to CS improvement and decline only under condition that _S_s were
not on medication during the time when the charges occurred; the ego
functions represented by g changed in relation to CS change only when
Ss were being taken off drugs; the functions represented by g changed
in relation to CS charge in S8 either not on drugs or beirg withdrawn
from them; and the functions represented by §+ % varied in relation to
CS change only when S3 were on medication.

The predictions made on the basis of the ego regression hypothesis
held up much better under ND-ND conditions than under D-D or D-ND con-
ditions . The use of medication seas to interfere with the theoretical-
ly congruent relationship between changes in ego structure and in overt

63

adjustment that exists under non-drug conditions. It is not clear
exactly how the use of medication affected the relationship between
behavioral charge and ego change, for the data indicated that the drug
and non-drug groups did not differ from each other in the amount or
direction of test-retest charge in either CS or in the nine Rorschach
scores. Therefore, the loss of correlation between Rorschach and CS
change cannot be explained on the basis that drugs either enhanced or
hindered charges in CS or in Rorschach performance.

All that can be said with certainty about the drug effect is that
beirg on medication during the test-retest interval led to the disap-
pearance of all the relations between CS movement and Rorschach score
charge which were predicted on the basis of ego regression theory and
which were confirmed in S5 not on drugs. The lone exception was real-
ity testing, which appears to be impervious to the effects that drug
usage has on the other ego functions. The correlations of §+%change
with CS change were consistent in direction for every replication under
every drug and non-drug condition. Another way of describing the drug
effect is that ego-structural change coincided, in a theoretically mean-
irgful way, with behavioral change to a far greater extent in ND-ND Ss
than it did in D-D or D-ND Ss. It can be concluded that drug-mediated
clinical status change is characterized by greater heterogeneity in
charges in ego functionirg than is charge in clinical status in the
absence of medication. The need for more research into the effects of
medication on both overt behavior and inner processes is obvious.

It is difficult to determine how medication influenced the findings
for the D-ND group since it was not always known when these S3 had been
taken off drugs. Some were withdrawn from medication shortly after the
initial test, others may have remained on drugs throughout much or most
of the test-retest interval.
gh_e_ dmmics 2g chagge

A question can be raised as to why certain aspects of ego function-
ing were related to CS charge whereas other aspects weren't. Even
under ND-ND conditions, three of the nine scores for which predictions
were made weren't confirmed.

On the grounds of the evidence presented above , acceptance of the
generality of the ego regression hypothesis would therefore be highly
questionable. That is, it is unreasonable, on the baSis of both

64

emmdrical and theoretical considerations, to believe that all aspects
of ego functioning regress and recover with every change in CS. The
statement that “psychopathology involves ego regression” must be made
Inore explicit to be truly meaningful. Inasmuch as "ego" is a collec-
tive name that refers to the various adaptive processes which mediate
ibetween the needs of the organism and the demands of its environment,
the concept of ego regression isn't very helpful unless there is speci-
fication of the specific functions that are affected.

Discussion of the nature of ego regression in schizOphrenia pro-
vides insight into the failure of some Rorschach measures of change in
ego developmental level to correlate with changed mental health status.

Two ideas in relation to change are especially relevant. The first
is embodied in.Schafer's (1955) comments to the effect that there are
individual differences in the dynamic characteristics of improved or
decreased psychological adjustment, i.e., a personality change which
indicates movement toward health for one person.may indicate increasing
pathology in another. The second is Bellak's (1969) contention that
ego regression is not a uniform process among individuals diagnosed as
schizophrenic.

Since Schafer's view was discussed earlier in the paper, only
Bellak's position need be elaborated. He pointed out that clinical
experience with schizOphrenic patients readily reveals that somerego
functions are affected to a greater extent than are others, and that
the precise nature of the differential ego regressions varies widely
across individuals. For instance, disruption of normal, mature think-
ing processes may be a conspicuous feature in one patient's behavior
while disturbances in interpersonal relations may be paramount in anp
other patient's behavior. The second person's cognitive functioning,
however, may have remained relatively intact at its premorbid level
whereas the first patient's capacity for mature social relations reh
maimed comparatively intact. What is common to both individuals, if
the specific dysfunctions are severe enough, would be a general infan-
tilization of the ego (as judged by an external observer who is cone
sidering the varied functions of the ego as an integral whole).

Inasmuch as the schizophrenic state (and, more likely, all other
types of psychOpathology as well) may involve different degrees of

65

.regression in the various ego functions, recovery may entail different
degrees of progression in these processes. Thus, it can.be said that
there are many paths to pathology or to recovery, and that global clinp
ical status change need not involve alteration in every ego function.
To the extent that there is diversity in the dynamics of change, meas-
‘ures of Specific underlying processes are inadequate as signs of gener-
ic change.

A significant consequence of the differential change in the vari-
ous ego functions which is concurrent with changes in.mental health is
that in the present study the correlations between shifts in CS and in
Rorschach scores were quite low-even for the relationships which were
statistically significant. It is not surprising, then, that studies
'have encountered difficulty in obtaining positive findings.

The availability of replicated data is particularly valuable in
such a situation. Moreover, since the present investigation.was handi-
capped by the smallness of‘fi necessitated by analysis of the separate
drug groups, the combining of the Several tests of significance, made
possible by replication, provided a technique which enhanced the power
of the statistical tests, a most fortuitous circumstance.

If clinical status Change did involve unifbrm alteration in each
of the component ego functions, it is probable that every Rorsdhach
score which is a valid indicator of the deve10pmental status of an ego
function would be significantly correlated with overall changes in
adjustment.

Adflag g: Rorschach measures 9; Egg functionigg. The most ap-
prepriate measures of ego regression are therefore variables which
relate to gross disruptions of the ego rather than to disturbances of
specific, relatively circumscribed, functions. The success or failure
of certain Rorschach scores to vary in proportion to variation in clin-
ical status may thus depend upon their sensitivity to the more general-
ized aspects of ego functioning. A Rorschach measure which relates to
the overall effectiveness of the ego would be expected to correlate
with clinical status change Whereas a score which is specific to but
one or two of the several component ego functions that are instrumental
to adjustment might not be eXpected to do so, because that specific
function.may not be disturbed in.many persons.

66

The superiority of certain scores in reflecting clinical status
charge was illustrated in the present study by ES, 5, and Scorability,
all of which are nonspecific measures of‘S's capacity to adapt to his
environment. Each score conveys information about the ability to fbl-
low examiner's instructions or to modify behavior to fit the changing
demands of the environment, but little information can.be directly
gained from these three scores about relatively specific ego qualities,
such as degree of contact with reality, effectiveness of impulse con-
trol, dominant modality for expression of drives and affects, or atti-
tudes toward other persons. The predictions involving all three scores
were confirmed for the entire sample and (with the exception of Scora-
bility, which couldn't be tested in the separate drug groups because of
the extremely small'g) for the ND-ND condition as well.

The only other score to reflect CS change across the sample as a
whole was 5+1», which relates information about the reality-testirg
function of the ego and also represents an index of ego strength. The
latter function, being intimately related to the capacity to adjust to
the requirements of the situation, thus appears to be akin to the three
relatively nonSpecific measures of ego functioning discussed in the
preceding paragraph. The results of the present study therefore appear
to support the contention that Rorschach measures of generalized ego
functioning are more adept at correctly assessing CS changes than are
the measures of more specific ego processes.

An alternative means of explaining the success of the above four
scores when drug and non-drug‘Ss are analyzed together is that certain
ego functions exert a more pervasive influence upon overall clinical
status than do others, i.e., there is greater tolerance for abnormality
(before adjustment is seriously affected) in some ego functions than in
others. Regression to a state of inflexibility of behavior, for example,
is generally more disabling than is regression in interpersonal behavior
or in the expression of emotions.

gig/:2, LI, 5, 2S, and 3 appear to be more representative of specif-
ic functions rather than of the general condition.of the ego. For the
entire sample, none of them correlated with CS change as predicted. It
thus appears that some of the more specific aspects of ego functioning
relate to CS change only in the absence of the drug effect whereas the

Q7

general aspects of ego functioning relate to CS change not only in the
absence of the drug effect but also in spite of the inclusion,of.Ss
exhibiting the drug effect.

It therefOre appears to be insufficient or inaccurate to talk a-
bout ego regression in a generic sense. Neither clinical impression
nor empirical study lends support to the idea that ego regression is a
monolithic mechanism in the deve10pment of sdhizophrenia or of_any
other mental disorder. Regression and progression are selective in the
functions that are affected during the course of pathogenesis and re-
covery.

The critics of regression theory (e.g., Buss & Lang, 1965) have
been answered. The ego regression theory of psychopathology is valid
insofar as it accurately predicts the changes which will occur in sev-
eral ego functions when schizophrenic patients show clinical status
gain or loss without the intervention of drug treatment.
Implications‘fgg future studies

Aside from the obvious need to investigate the total effect of
medication on every aspect of personality functioning, there are sever-
al other implications to be made with regard to other research endeavors.

Most studies of schizophrenia involve the use of'Ss who are on
medication. In light of the revelations about the effect of drugs on
the relationship between inner psydhological processes and overt behave
ior, the drug status of'Ss is a treatment variable with which to reckon.
Perhaps the seemingly contradictory findings of previous investigations
into clinical status and Rorschach performance reflects differences
among the samples in terms of the use of medication.

The potential value of a score which indicates clinical status
change was perhaps best expressed by Phillips and Framo (1954):

The severity of disorder could be determined in terms of
degree of perceptual regression. This scale could then be used
as an index of therapeutic success; if a patient were to be

tested before and after therapeutic intervention, movement along
this developmental scale...might represent the changes taking
place with therapy. To be meaningful, however, such indices
should first be related to psychiatrically relevant criteria--
to accepted measures of therapeutic change.

Their expectations have been realized with the demonstration of
longitudinal Rorschach and clinical status changes which are meaningful

68

in terms of ego regression theory. Although Phillips and Framo (1954)
had the Friedman 21; scores in mind when they voiced their anticipation
of an index of therapeutic progress, the present study suggests that
the slighting of §_§_ in deference to _QL_ in studies conducted since the
inception of the genetic mode of investigating psychopathology has been
unfortunate.

_P§_ was easily the most sensitive and reliable Rorschach measure of
CS change in the present study. It now needs to be used with other
populations to determine whether the psychological characteristics it
measures may represent a dynamic which is cannon to all instances of
clinical status change. No Rorschach score has yet been found in per»

sons who are not psychotic which does vary consistently with changes in
clinical status.

SUMMARY

The present study looked for Rorschach changes concurrent with
clinical status change from the perspective of the ego regression
theory of psychopathology . The concept of ego regression employed
herein is compatible with the independent formulations of psychoanaly-
tic and of Mparative-developmental theory. Both positions hold that
changes in the adequacy of overall functioning parallel changes in the
developmental status of the ego, i.e. , the precipitation of mental dis-
order is accompanied by retreat to less mature forms of functioning;
and , conversely , recovery involves movement toward more mature modes of
ego functioning.

Research with the Rorschach has thus far demonstrated that the
responses of adults suffering from serious psychological disorders
resemble in many ways the inkblot performance of young children
rather than that of normal adults. Every study purporting to have
confirmed the existence of ego regression in psychopathology can be
criticized, however, for not offering evidence from longitudinal re-
search. Much more convincing would be demonstration that as a person
becomes increasingly maladjusted, his ego operates at a more infantile
level, and as he improves, his ego again operates at a more mature
level .

The present study thus represented the first known attalpt to
empirically investigate the role of ego regression in psychOpathology
through the use of longitudinal Rorschach data. It was hypothesized
that whenever a retest Rorschach shows movement toward or away from a
more mature mode of ego functioning, there is a concomitant improvement
or decment in overall adjusment level. It was predicted that changes
in ten separate scores or indices, which served as measures of the
maturity level of the ego, are correlated with change in clinical status.

Four times over a period of twenty months, thirty-five _S_s diag-
nosed as clearly schizophrenic were aministered both a psychiatric

69

70

interview to ascertain clinical status (CS) and the Rorschach. Meas-
uresofclinical statuschangeandofRorschachscorechangewere
obtained for three separate test-retest intervals—initial testing to
six months, six months to twelve months, and twelve months to twenty
months-uso that data was provided for three independent replications
of the study. The delta regressed score transformation was used to
free each measure of change from the effects of regression error.
Product-moment correlation coefficients were computed for each of the
three tests of the ten predictions. The significance of the findings
was determined by combining the probabilities of the three tests of
each prediction.

‘ Four of the ten measures—Perseveration Score (ES) , Scorability,
5+ %, and gin-correlated with CS change in accord with prediction. When
gs were divided into three groups for analysis according to drug treat-
ment status, the results were striking. £3 and g correlated with CS
charge only for _Ss not on medication at any time during the test-retest
interval (ND-ND), while 33% related to CS change only in _S_s remaining
on medication throughout the interval (D-D). Furthermore, 11, _P_, and
the two Friedman Developmental Level scores (95;), although not indica-
ting CS change for the entire group, did correlate with CS charge, as
predicted, in gs not on drugs. In addition, g and [-1 changes correlated
with CS change in gs withdrawn from medication (D—ND) between the ini-
tial test and the retest.

Regression and progression were not uniform across all the ego
functions examined. The data thus corroborated the position established
on the basis of clinical experience that there are many ways for in-
proth or decline to occur (i.e., CS change need not involve the
same dynamics in every case). What most, if not all, instances of clin-
ical status charge do hold in' cannon appears to be a broadly-defined
shift in the developmental quality of ego functioning, i.e., the abil-
itytoadapttothedemandsoftheenviroment. Somescoresaremore
adept at reflecting clinical status change across individuals than are
others , apparently in direct proportion to their ability to indicate
general rather than specific aspects of ego functioning. Thus, charges
in g, _R_, _1:+% , and Scorability, all relatively nonSpecific measures of
ego adequacy, correlated with CS change in accord with prediction. The

71

remaining scores appeared to relate to more specific and circmnscribed
functions of the ego.

Drug-mediated CS change was shown to differ in its dynamic quali-
ties from CS change not accompanied by the use of medication. The
former condition was associated with charges in reality testing and in
ego strength; the latter with changes in several general and specific
aspects of functioning , including adaptability to the enviroment ,
reality testing, complexity of perceptual-cognitive processes, cone
trolled expression of drives, and degree of socialization. Furthermore,
the use of medication leads to clinical status charge characterized by
greater heterogeneity in the accompanying ego changes than is the case
when medication is not used.

It was therefore concluded that the ego regression theory of psy-
chopathology is valid in that it accurately describes changes in the
quality of important aspects of ego functionirg which occur upon charge
in mental health. It was further concluded that the specific ego
charges which accompany charges in overt adjustment are dependent upon
the nature of the treatment conditions to which persons are exposed
during the course of change.

Also discussed were the implications of the findings for research
with psychotics and for the development of a Rorschach index for evalu-
ation of the effectiveness of therapy.

LISTOPREPMS

LIST OF REFERENCES

Allport, G. W. The general and the unique in psychological science.
Journal _o_f_ Personality, 1962, 39, 405-422.

Ames, L. B. Changes in Rorschach responses throughout the human life
span. Genetic PM No a he, 1966, 14, 89-125.

Ames, L. B., Learned, J., Metraux, R. w., & Walker, R. N. Child Ror-
schach resEnses: developmental trends from two 1:3 ten yeg,
New York: Hoeber, 1952.

Arieti, S. The interpretation _o_f_ schizophrenia. New York: Brunner,
1955.

Barry, J. R., Blyth, D. D., & Albrecht, R. Relationships between Ror-
schach scores and adjustment level. Journal 9: Consultigg gm-
ggy, 1952, 16, 30-36.

Beck, S. J. Rorschach's test. Vol. II. A variety 93' personality pic-
tures. New York: Grune 8: Stratton, 1945.

Bedc, S. J. Rorschach's test, Vol. III. Advances in interpretation.
New York: Grune & Stratton, 1952.

M, So Jo, M, A. Go, Leutt, E. E0, & "011811, Ho Be maCh's
test. Vol. I. Basic processes. (3rd ed.) New York: Grune&
Stratton, 1961.

Becker, w. C . A genetic approach to the interpretation and evaluation
of the process-reactive distinction in schizophrenia. Journal 93;
Abnormal and Social Ps chol , 1956, _53, 229-236.

Bellak, L. Research on ego function patterns: a progress report. In
L. Bellak& L. Loeb (Ed.), The schizophrenic M. New York:
Grune 8: Stratton, 1969. Pp. 11-65.

BUhler’ Ce, mer, Keg & Lefever, D. We Develogpgnt 9-f- the 8351‘:
Rorschach Score with manual g_f_ directions. Basic Rorschach Stand-
ardization Studies, No. 1, 1948.

Buss, A. H. Psychgpathology. New York: Wiley, 1966.

Buss, A. H., &Lang, P. J. Psychological deficit in schizophrenia:

I. Affect, reinforcement, and concept attairment. Journal 9;
Abnormal Psmolggy, 1965, 19, 2-24.

'72

73

Cadman, w. H., Misbach, L., & Brown, D. V. An assessment of round-

table psychotherapy. Psmlgical Nogggraphs: General and Ag-
lied, 1954’ _62, (“1018 NO. 384).

Cameron, N. Personalig develoggent ;_ar_u_d_ psmopatholggy. Boston:
Houghton Hifflin, 1963.

Carr, A. C. Evaluation of psychotherapy by the Rorschach. Journal _o_f_

Cronbach, L. J. Statistical methods applied to Rorschach scores: a
“Vi-We Psxgl-glgical mating 1949, fig, 393-429.

Draguns, J. 6., Haley, E. M., a Phillips, L. Studies of Rorschach con-
tent: a review of the research literature. Part I: Traditional
content categories. Journal g_f_ Pro jective Techniges 2:}; Person-
ality Assessment, 1967, _71, 3-32.

Dudek, S. 2. Effects of different types of therapy on the personality
as a whole. Journal _o_f Nervous and Mental Disease, 1970, 150, 329-
345.

Edwards, A. L. Statistical methods for the behavioral sciences. New
York: Rinehart, 1954.

Fenichel, 0. 3h; psychoanalxgic £32922 _og neurosis. New York: Norton,
1945.

Fiske, D. W., & Baughman, E. E. Relationships between Rorschach scoring
categories and the total number of responses . Journal 2: Abnormal
and Social Psycholggr, 1953, ;4_8_, 25-32.

Flynn, A. P. Relationship between developmental levels and pathology
on perceptual and conceptual tasks. Unpublished doctoral disser-
tation, Catholic Univ. of America, 1965.

Fox, J. The psychological significance of age patterns in the Ror-
schach records of children. In Klopfer, B., Develoaggts 2.3 the
Rorschach technigg, Vol. II. Fields 9; Application. New York:
World Book, 1956, 88-103. '

Freud, A. _T_1_:_e_ ing _t_h_e_ mechanisms of defence. London: Hogarth
Press, 193 .

 

Friedman, H. Perceptual regression in schizophrenia: an hypothesis
suggested by the use of the Rorschach test, Journal g_f_ Genetic

W: 1952. 9.1.. 63-98-

Gaylin, N. L. Psychotherapy and psychological health: a Rorschach
function and stmcmre analysis. Journal _o_i: Consultig W,
1966, _32, 494-500.

Goldman, A. E. A comparative-developmental approach to schizophrenia.
Psflglgfical Bulletin, 1962, 52, 57-69.

'74

Goldman, R. Charges in Rorschach performance and clinical improvement
in schizophrenia. Unpublished doctoral dissertation, Boston Uni-
ver., 1955. Published in abridged form in Journal 2; Consulting

m, 1960, 24, 403-407.

Guilford, J. P. fundamental statistics 22 psychology and education.
(3rd ed.) New York: McGraw-Hill, 1956.

Haimowitz, N. R., &Haimowitz, M. L. Personality changes in client-
centered therapy. In W. Wolff & J. Precker (Eds.), Success in
psychotherppy, (Psychological Monographs, Vol. 3), New York:
Grune & Stratton, 1952. Pp. 63-93.

Halpern, F. Rorschach interpretation of the personality structure of
schizOphrenics who benefit from insulin therapy. Psychiatric
Quarterly, 1940, .12, 826-833.

Hamlin, R. 14., 8: Albee, G. W. Muench's tests before and after nondi-
rective therapy: a control group for his _S_s. Journal 2; Consult-
_i_9_g Psychology, 1948, _1_2_, 412-416.

Hamlin, R. 14., Berger, B., & Cunmirgs, S. T. Changes in adjustment
followirg psychotherapy as reflected in Rorschach signs. In W.
Wolff & J. Precker (Eds.), Success i2 psychotherapy (Psmolggical
Monographs, Vol. 3). NewYork: Grune 8zStratton, 1952. Pp. 94-111.

Hemmendinger, L. Perceptual organization and development as reflected
in the structure of Rorschach test responses. Journal 9_f_ Projec-
tive Technigues, 1953, 11, 162-170.

Hmendinger, L. Developaental theory and the Rorschach method. In H. A.
Rickers-Ovsiankina (Ed. ) , Rorschach W. New York: Wiley, 1960.

Hertz, H. R. W tables for scorigg Rorschach responses. (4th
ed.) Cleveland: Western Reserve Univer. Press, 1961.

Hollingshead, A. B., & Redlich, F. C. Social class and mental illness.
New York: Wiley, 1958.

Jacobson, E. Psychotic conflict and realig. New York: International
Universities Press, 1967.

Karon, B. P., & O'Grady, P. Intellectual test changes in schizophrenic
patients in the first six months of treatment. Psmtheram:
Theo , Research, .3; Practice, 1969, _6_, 88-96.

Karon, B. P., & O'Grady, P. Quantified judgments of mental health from
the Rorschach, TAT, and clinical status interview by means of a
scaling technique. Journal 21; Consultim 299. Clinical W,
1970, 24, 229-235.

Kisker, G. W. A projective approach to personality patterns during
insulin-shock and metrazol-convulsive therapy. Journal 93-; Abnor-
mal and Social Psychology, 1942, _31, 120-124.

75

Klopfer, B., Ainsworth, H. D., Klopfer, W. G., & Holt, R. R. Develop-
ments in. the Rorschach technifie. Vol. I. Techng and theog.
New York: Harcourt, Brace, & World, 1954.

Klorman, R., a Chapman, L. J. Regression in schizOphrenic thought dis-
order. Journal p_f_ Abnormal W, 1969, 14, 199-204.

Krout, J., Krout, H. H., & Dulin, T. J. Rorschach test-retest as a
gauge of progress in psychotherapy. Journal 9_f_ Clinical m-
a, 1952’ §, 380.384.

Ledwith,N . Rorscha___________c_:_h_re resEns es o__f_ elementgw school children: a
normative study. Pittsburgh: Univer. of Pittsburgh Press, 1959.

Lipton, E., & Ceres, H. Correlation of clinical improvement of inten-
sively treated psychoneurotics with changes in consecutiVe Ror-
schach tests. Psychiatric Q_u_arter1y Supplement, 1952, _2_§, 103-117.

McNemar, Q. L. Pmlogical statistics. (3rd ed.) New York: Wiley,
1962.

Maslow, A. Toward _a_ W 93 balm. Princeton: VanNostrand, 1962.

McFate, R., 8: Orr, F. Through adolescence with the Rorschach. Journal
p_f_ Projective Techrugues, 1949, 1_3_, 302-319.

Meile-Dworetzki, G. The development of perception in the Rorschach. In
Klopfer, B. DeveloEnts ip the Rorschach Technigg. Vol. II.
Fields 93 Application. New York: World Book, 1956. Pp. 104-176.

Meltzoff, J ., a Kornreich, H. Research _ip Psflotherapy. New York:
Atherton, 1970.

Miale, F. R., & Harrower-Erickson, H. R. Personality structure in the
psychoneuroses. Rorschach Research Exchamg, 1940, 4, 71-74.

Mosak , H . H . Evaluation in psychotherapy: a study of some current
measures. Unpublished doctoral dissertation. Univer. of Chicago,
1950.

Muench, G. A. An evaluation of non-directive psychotherapy by means of

the Rorschach and other indices. Applied W Moggrams,
1947, 13, 1-163.

Munroe, R. L. Prediction of the adjustment and academic performance of
college students by a modification of the Rorschach method.

Applied Psmlggcal W, 1945, No. 7.

Neiger, S., Slemon, A. G., & Quirk, D. A. Rorschach scales of regres-
sion in psychosis. Genetic W W, 1965, 11, 93-136.

Oppenheim, S., & Brower, D. The effects of electric shock therapy as

revealed by the Rorschach technique. Psmatric Qu_a_pt_e_r;ly m-
ment, 1948, 32, 318-325.

76

Peterson, A. O. D. A comparative study of Rorschach scorirg methods in
evaluating personality charges resulting from psychotherapy.
Journal 21;: Clinical W, 1954, 12, 190-192.

Phillips, L., & Elmadjian, F. A Rorschach tension score and the di-
urnal lymphocyte curve in psychotic subjects. Psychosomatic Hedi-
cine, 1947, 2, 364-371.

Phillips, L., It Framo, J. L. Develoynental theory applied to normal

and psychopathological perception. Journal 2; Personalii_:y, 1954,
_2_2_, 464-474. _

Phillips, L., 8: Saith, J. G. Rorschach intflretation: advanced tech-
n_1g_e_. New York: Grune 8: Stratton, 1953.

Piotrowski, z. A. Rorschach manifestations of improvement in insulin
treated schizophrenics. Psychosomatic Medicine, 1939, 1, 508-526.

Piotrowski, z. A. The movwent score. In H. A. Rickers-Ovsiankina
(Ed.), Rorschach w. New York: Wiley, 1960. Pp. 130-153.

Piotrowski, Z. A.,& Schreiber, M. Rorschach perceptanalytic measure-
ment of personality charges durirg and after intensive psychoana-
lytically oriented psychotherapy. In G. Bychowski 8: J. L. Despert

(Eds.), S alized techniggg am. New York: Basic
Books, 1952. Pp. 537-361.

Pitman, E. J. G. Significance tests which may be applied to samples
from any population, II. The correlation coefficient test.
nal 9__f _t_h_e Royal Statistical Socigty Slappiwe went, 1937, 4, 225-25 .

Rabin, A. I., 8: Beck, S. J. Genetic aspects of some Rorschach factors.
American Journal _o_f_ Orthopsggagy, 1950, _2_9_, 595-599.

Rickers-Ovsiankina, H. A. Longitudinal approach to schizophrenia
through the Rorschach method. Journal of Clinical and W
1:21.. W: 1954, 1.12:1 107-118-

Rioch, H. J. The use of the Rorschach test in the assessment of change
in patients under psychotherapy. PM, 1949, 13, 427-434.

Rogers, C. R. 92 w _a_ Epsom Boston: Houghton Hifflin, 1961.

Rogers, C. R., 8: Dymond, R. (Eds.). Pgflptheram 9gp W
chggge. Chicago: Univer. of Chicago Press, 1954.

Rorschach, H. PMaflstics (5th ed.). Bern: Huber, 1951.

Sanders, R., & Cleveland, S. E. The relationship between certain exam-
iner personality variables and subjects' Rorschach scores. Jour-
p_a_l_ pf Prolective Techni es, 1953, 11, 34-50.

Schafer, R. Ppflganalytic intgpretation y; Rorschach team. New
York: Grune & Stratton, 1954.

77

Schafer, R. Psychological tests in clinical research. Journal 31: con-
sultig PM, 1949, 12, 328-334.

Schafer, R. Psychological test evaluation of personality change during
intensive psychotherapy. Psychiatry, 1955, 1Q, 175-192.

Schafer, R. 0n the psychoanalytic study of retest results. Journal of
Projective Techniques, 1958, _2_2_, 102-111.

Shapiro, D. A perceptual understanding of color response. In 1!.
Ri'ckers-Ovsiankina (Ed.), Rorschach W. New York: Wiley,
1960. Pp. 154-201.

Sidman, H. Tacticsip scientific research. New York: Basic Books, 1960.

Siegel, E. L. Genetic parallels of perceptual structuralization in
paranoid schizophrenia: an analysis by means of the Rorschach
technique. Journal _o_f_ Projective Techniques, 1953, 11, 151-161.

Smith, N. C. Replication studies: a neglected aspect of psychological
research. American Psychologist, 1970, 35, 970-975.

Thetford, W. N., Molish, H. B., & Beck, S. J. Developmental aspects of
personality structure in normal children. Journal pf Projective
Techniques, 1951, 15, 58-78.

Tudcer, J. E. Investigation of criteria for evaluating non-directive
psychotherapy with college students. Unpublished doctoral disser-
tation, Pennsylvania State College, 1951.

Van Pelt, W. P. Perceptual-cognitive development as reflected by Ror-
schach test content . Unpublished doctoral dissertation. Syracuse
Univer., 1961.

Ward, A. J. The meaning of the movement reSponseandits changes during
therapy: areview. Journal p_f_ Projective Technigpgg ing Personal-
m Assessment, 1966, _32, 418-428.

Watkins, J. G. Evaluating success in psychotherapy. American P§yc_hol-
pgist, 1949, _4, 396 (Abstract).

Werner, H. Comparative m 9; mental development, (rev. ed.) .
Chicago: Pallet, 1948.

Werner, H . The concept of developnent from a comparative and organis-
mic point of view. In D. E. Harris, The concept _o_i: develognt:
an i_s__sue in th__e_s tug__ of human behavior. Minneapolis: Univer. of
Minnesota— Press, 1957.

Zamansky, H. S., 8: Goldman, A. E. A comparison of two methods of ana-
lyzing Rorschach data in assessing therapeutic charge. Journal of
Projective Techniques, 1960, 22! 75-82.

Zax, N., Stricker, 6., 81 Weiss, J. H. Some effects of non-personality
factors on Rorschach performance. Journal o__i: Pro lective Techniques,
1960, 24, 83-93.

HICHIGAN STRTE UNIV. LIBRARIES
1|WMWIWillWNWWHIIHWHIIIW"
31293104649995