“+3.3 MIIINUIHNIIIIIIIIHHIHIIIMHWI"WIHHIIUHII

P \ 193 01575 5725
“415:3

LIBRARY 1

Michigan State
University

 

 

 

 

This is to certify that the

dissertation entitled

RORSCHACH ASSESSDWT IN BORDERLINE PERSONALITY
DISORDER: A FOLLOW-UP STUDY

presented by

Laura Christine Baker

has been accepted towards fulfillment
of the requirements for

—_2h.D.— degree in —-Psyehei-egy

MW. M

Major professor

Date Mm '1 ‘7

MS U it an Affirmative Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

MTE DUE MTE DUE MTE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

in W14

RORSCHACH ASSESSMENT IN BORDERLIN E PERSONALITY DISORDER:
A FOLLOW-UP STUDY

By

Laura Christine Baker

A DISSERTATION
- Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Psychology

1997

ABSTRACT

RORSCHACH ASSESSMENT IN BORDERLINE PERSONALITY DISORDER:
A FOLLOW-UP STUDY

By

Laura Christine Baker

Although by definition borderline personality disorder (BPD) and other
personality disorders are characterized by long-term dysfunction (American
Psychiatric Association, 1994) follow-up studies using symptom-based
measures of borderline pathology have found that diagnostic stability is
relatively low for intervals less than 5 years (Barasch, Frances, Hurt, Clarkin,
8: Cohen, 1985; Links, Mitton, 8r Steiner, 1993). It is unknown whether the
reported changes in diagnostic status over time correspond to changes in
personality dynamics. The present study used the Rorschach to assess aspects
of personality structure in a group of inpatients diagnosed with BPD by the
Diagnostic Interview for Borderlines (DIB; Gunderson, Kolb, & Austin, 1981)
at two time periods: initial psychiatric hospitalization and follow-up 2-5 years
later. Four aspects of personality functioning as assessed by the Rorschach
were compared at both times: Object Relations, Defenses, Reality Testing, and
Ego Capacity. Reality Testing and Ego Capacity were operationalized using
Exner‘s (1986) Comprehensive System. Defenses on the Rorschach were
measured using the Rorschach Defense Scales (Cooper, Perry, & Arnow, 1988),
and Object Relations were assessed using the Human Experience Variable
(Perry & Viglione, 1991). Results indicate that changes in Defenses were
significantly correlated with changes in diagnostic status (; = .64, p < .001),
such that those who no longer met DIB borderline criteria at follow-up had

improved on the Rorschach measure of defensive functioning. Post-hoe
analyses revealed that the changes in Defenses were strongly correlated with
changes in the DIB subscales Affect (g = .51, p < .05) and Social Adaptation (g =
.41, 2 =06). Subjects who were no longer BPD by the DIB at follow-up had
better Time 1 Rorschach indices of Defenses, Object Relations, Reality
Testing, and Ego Capacity than those who retained a BPD diagnosis,
suggesting that those who improved diagnostically had better initial
personality functioning. Considerable variability in Rorschach personality

indices was observed among those receiving the BPD diagnosis.

ACKNOWLEDGMENTS

I would like to thank a number of people who have inspired and
encouraged me through this project. I owe infinite gratitude to the members
of the Coping Study, especially Dr. Ken Silk and Dr. Naomi Lohr, for taking
me under their wing and welcoming me into their research group. Their
enthusiasm for research and their genuine love of mentoring have left me
with a good internalized "research object" that I will pass on to others. I am
particularly grateful to Edna Pressler for her crucial role in initiating the
follow-up study. I wish to thank the entire Coping Study not only for their
valuable input which helped me to refine my rough concept into a viable
project, but also for volunteering their time and skills for data collection.
Without their work this study would not have been possible.

I am also grateful to the members of my dissertation committee - Dr.
Norman Abeles, Dr. Robert Caldwell, Dr. Albert Rabin, and Dr. Bertram
Karon - for their separate and unique contributions. As my dissertation chair
Dr. Abeles has been a dedicated and tireless supporter throughout, and his
remarkable accessibility in spite of his presidential obligations this year speaks
to his commitment to students. His consistently practical and realistic focus

was a tremendous asset to the dissertation process. I want to thank Dr.

iv

Caldwell for his thoughtful and scientifically rigorous criticisms. His feedback
always brought perspective to the issue at hand and stimulated me to think
more deeply about my project. Although less involved with the everyday
development of the project, both Dr. Rabin and Dr. Karon have provided me
with theoretical inspiration through their teaching and clinical wisdom.
Finally, I must thank my colleagues, friends, and my family for their
generous support throughout. Brian Tyl, Mark Reeves, Matt Novak, Jason
Dahn, Regina O'Connell, Alison Ward, Katari Brown, Kristina Hammerle,
and Jamie Jones volunteered their time to score Rorschachs, which was a
painstaking process. I would also like to thank my family, especially my
parents, for their unconditional support and for always encouraging me to

pursue my dreams.

TABLE OF CONTENTS

 

LIST OF TABLES ...................................................................................................... viii
Introduction"... - - 1
The Borderline Concept- Divergent Diagnostic Traditions .............. 1
Current Competing Definitions of "Borderline" ................................. 3
Descriptive / Syndrome Definitions ............................................. 4
Psychodynamic Formulations ...................................................... 7
Overlap between the BPD and BPO Constructs .................................... 11
The Question of Diagnostic Stability... ...13
The Use of the Rorschach 1n Assessing Personallty Structure .......... 21
Identity Diffusion ............................................................................ 24
Defensive Operations ..................................................................... 25
Reality Testing .................................................................................. 27
Ego Weakness ................................................................................... 29
Object Relations ................................................................................ 30
The Rorschach as a Measure of Personality Change ................ 33
Summary and Hypotheses ........................................................................ 37
Overview of the Current Study ................................................................ 40
Method ....................................................................................................................... 41
Index Study .................................................................................................. 41
Subjects ............................................................................................. 41
Measures .......................................................................................... 41
Procedure ......................................................................................... 47
Follow-up Study ........................................................................................... 48
Subjects ............................................................................................... 48
Measures ............................................................................................ 49
Procedure ........................................................................................... 49
Rorschach Scoring ........................................................................................ 50
Results ........................................................................................................................ 58
Discussion .................................................................................................................. 68
Footnote ..................................................................................................................... 80
Appendix A ............................................................................................................... 81
Appendix B
The Diagnostic Interview for Borderlines (DIB) Summary of
Borderline Criteria ......................................................................................... 82

vi

Appendix C

Rorschach Defense Scale: Scoring Notes .................................................. 83
Appendix D

Rorschach Variables and Scoring Systems ............................................... 86
LIST OF REFERENCES ........................................................................................... 89

vii

LIST OF TABLES

Table 1 - Inter-rater Reliability Coefficients between Raters and Exner's
scores for Exner Variables ..................................................................... 52

Table 2 - Subject and Follow-up Variables ......................................................... 59

Table 3 - Means and Standard Deviations of DIB and Rorschach
Variables ................................................................................................... 61

Table 4 - Pearson Correlation Coefficients Between Difference Scores
for DIB and Rorschach Variables ........................................................ 62

Table 5 - Rorschach Variables by Improvement Status at Time 1 and
Follow-up ................................................................................................. 65

Table 6 - Subjects Classified in Neurotic, Borderline, and Psychotic
Categories on Rorschach Variables ..................................................... 66

viii

Introduction

Follow-up studies of borderline patients have sought to answer the
question: what happens to borderlines over time? Competing definitions of
"borderline" make this question complicated and challenging. Studies using
symptom-based measures of borderline pathology have found that diagnostic
stability is lower than expected in intervals less than five years (Barasch,
Frances, Hurt, Clarkin, & Cohen, 1985; Links, Mitton, & Steiner, 1993;
Pressler, 1990). What is unknown is whether patients who show diagnostic
and symptomatic changes over relatively short intervals are also manifesting
structural personality changes in the psychodynamic sense. Lacking from the
literature are well-designed, prospective follow-up studies that use both
symptom-based and psychodynamic diagnostic criteria for defining
borderlines. Such studies would allow us to determine whether or not
diagnostic instability in the symptom-based measures corresponds to changes
in the underlying dynamics of borderline personalities. The present
investigation uses the Rorschach to assess the personality structure of a group
of inpatients diagnosed as borderline by the Diagnostic Interview for
Borderlines (DIB; Gunderson, Kolb, & Austin, 1981) at two time periods:
initial psychiatric hospitalization and follow-up 2 to 5 years post-
hospitalization.

The Borderline Concept - Divergent Diagnostic Traditions

Since the turn of the century, clinicians and diagnosticians have

attempted to name, define and understand a group of patients who could not

be classified as psychotic or neurotic, but had some characteristics of both.

1

Because this ambiguously defined group of patients were believed to function
better than psychotics but not as well as neurotics, they fell somewhere on the
'borderline' between the two. Since then the borderline diagnostic category
has been a controversial and constantly evolving subject for clinicians,
theorists, and researchers. The first borderlines, a heterogeneous collection of
patients, were identified simultaneously in hospital psychiatric and
psychoanalytic circles. Upon superficial examination these patients appeared
to function fairly well like neurotics, but when confronted with unstructured
situations, such as psychoanalysis and projective testing, they often
underwent psychotic-like regressions. As a result, unstructured testing such
as the Rorschach was useful in diagnosis due to their distinctly primitive
responses. At the time the differential diagnosis of these patients was
considered important for treatment recommendations, as psychoanalysis was
contraindicated for such patients due to the profound regressions and
primitive transferences that frequently occurred during treatment.

Early writings by psychoanalytic theorists relevant to borderline
personality are complicated by changing definitional criteria and different
diagnostic labels. The historical development of the borderline category is
characterized both by attempts to unite these early 'borderline' syndromes
under one nomenclature, as well as attempts to make finer discriminations
between subtypes within the general borderline category. The term
"borderline", believed to have been first used by Adolph Stern in 1938, was
applied to patients who showed the following characteristics: narcissism,
psychic bleeding, inordinate hypersensitivity, rigidity, negative therapeutic
reactions, feelings of inferiority, masochism, organic insecurity, projection

mechanisms, and difficulty in reality testing (Stern, 1938). These patients

shared characteristics with those described by later writers, such as Deutsch's
"as if" personality, and Zilboorg‘s "ambulatory schizophrenia". Other terms
applied to patients in indistinct clinically 'borderline' categories, but
eventually discarded, were "latent schizophrenia", "preschizophrenia"
or"schizophrenic character"(Rappaport, Gill, & Schafer, 1946), "psychotic
character" (Frosch, 1964), and "borderland" (Clark, 1919; Chessick, 1968). The
term "borderline" was eventually popularized by Knight (1953), and has
maintained it's popular status through the present.

Current Competing Definitions of "Borderline"

Currently, two main conceptualizations of borderline patients exist side
by side in the literature: (a) the descriptive, symptom-based syndrome
definition of Borderline Personality Disorder (BPD; American Psychiatric
Association, 1994), and (b) the broader, conceptual, psychodynamic definition
of borderline personality organization (BPO; Kernberg, 1975, 1985). These two
definitions are a source of continuing diagnostic confusion. According to
Sugarman and Lerner (1980): "Although the term borderline has become
common and has stimulated a vast array of clinical data, theoretical
formulations, and treatment variations, disagreement over its definition has
never really subsided." (p. 11). Contributing to current diagnostic
disagreements is the historical independence of the descriptive/ diagnostic
and psychodynamic literature on borderlines, which emphasize different
definitional criteria (Gunderson 8: Singer, 1975); while psychiatric writers
emphasize similarities in the overt behavioral symptoms, psychoanalytic
writers define borderlines according to similarities in underlying personality
conflicts and deficits. Sugarman and Lerner (1980) trace this divergence to the
simultaneous but independent historical development of the borderline

concept within hospital-based Kraeplinian classification traditions and more
theoretical psychoanalytic and psychological testing traditions.

In the following paragraphs these two distinct ways of defining
borderlines will be summarized and the overlap between the two will be
discussed. Afterward, the literature on longitudinal changes in the borderline
diagnosis will be reviewed and predictions about the concordance between
the two diagnostic approaches over time will be made.

Descriptive [ Syndrome Definitions.

In the 1960s and 1970's subtyping of borderline disorders based on overt
symptomology allowed for greater consistency in the descriptive
phenomenology of borderlines. In reaction to the inferential descriptions by
early psychodynamically oriented writers, Crinker, Werble, and Drye (1968)
conducted the first study to clarify this seemingly heterogeneous diagnostic
category. They distinguished four subtypes of borderline patients comprising
those typically identified by clinicians: Type I, the "Psychotic border",
characterized by lack of adaptive behavior and poor reality testing; Type II, the
"Core Borderline Syndrome", characterized by erratic involvement with
others, angry acting out behavior, and unstable self-identity; Type III, the
"Adaptive" subtype, marked by lack of affect or spontaneity, appropriate,
compliant behavior, and use of withdrawal for defensive purposes; and Type
IV, the "Border with the Neuroses", manifested by depression and anxiety
more similar to neurotic or narcissistic functioning.

Another attempt to identify a narrower borderline syndrome was
undertaken by Gunderson and his colleagues (Gunderson & Singer, 1975).
They developed a behavior-based diagnostic system for defining borderlines,
the Diagnostic Interview for Borderlines (DIB; Gunderson, KoIb, & Austin,

1981). This interview shows a considerable overlap of diagnostic criteria with
a measure designed by Spitzer, Endicott, and Gibbon, (1979). The criteria for
the borderline diagnosis on the DIB were derived from an extensive clinical
literature review and previous research studies, such as that of Grinker,
Werble, and Drye (1968). The measure assesses five areas of behavior
believed to be relevant to borderlines: social adaptation, impulsivity, affect,
psychotic experiences, and interpersonal relationships. Patients who received
a DIB diagnosis of borderline personality disorder were described by the
following characteristics: impairments in social adaptation, impulsivity,
self-destructive activity, emotional instability, intense feelings of depression,
adequate reality testing accompanied by minor thought disturbances, and
chaotic relationships with others. Certain questions, for instance those
relating to bipolar depressive and clearly psychotic phenomena, carried
negative weights in the scoring and therefore counted against the borderline
diagnosis in an attempt to exclude patients with these disorders. However,
the DIB definition of borderline encompassed more than one of Grinker's
subtypes, including some patients who were later classified by DSM-III under
Schizotypal Personality Disorder. Gunderson (1981) acknowledged that the
DIB definition of borderlines might eventually be refined into narrower
categories.

Grinker et al.‘s (1968) and Gunderson's early attempts to classify
borderline syndromes was followed by other efforts to create more specific
diagnostic typologies based on symptom profiles. Spitzer, Endicott, and
Gibbon (1979) researched the relationship between two subtypes of borderline
personalities, "unstable" and "schizotypal", and developed separate criteria
for them. These criteria resulted in the first standardized psychiatric

diagnostic criteria for Borderline Personality Disorder (BPD) that appeared in
DSM-III (American Psychiatric Association, 1980).

DSM-III required five of the following eight criteria to qualify for a
diagnosis of Borderline Personality Disorder: (a) impulsivity or
unpredictability in at least 2 areas that are potentially self-damaging, (b) a
pattern of unstable and intense interpersonal relationships, (c) inappropriate,
intense anger or lack of control of anger, (d) identity disturbance, (e) affective
instability, (f) intolerance of being alone, (g) physically self-damaging acts, and
(h) chronic feelings of emptiness and boredom. The DSM-III diagnostic entity
corresponded best to Grinker et al.'s subtype of the Core Borderline syndrome,
which they characterized as having vacillating involvement with others,
acted out expressions of anger, lack of identity, and lonely depression.
Another group of patients, similar to Grinker et al.'s subtype the Psychotic
Border, characterized by withdrawal from others, rage, and transient
psychoses, became Schizotypal Personality Disorder in DSM-III. Four of the
following eight criteria were required for a diagnosis of Schizotypal
Personality Disorder: (a) magical drinking (e.g., superstitiousness), (b) ideas of
reference, (c) social isolation, (d) recurrent illusions, depersonalization, or
derealization not associated with panic attacks, (e) odd speech, (f) inadequate
rapport in face to face interaction due to constricted or inappropriate affect, (g)
suspiciousness or paranoid ideation, (h) undue social anxiety or
hypersensitivity to real or imagined criticism. Although Schizotypal
Personality Disorder is considered by some as a relative of Schizophrenia,
some borderline patients meet DSM-III criteria for both Schizotypal and
Borderline Personality Disorder.

The diagnostic criteria for BPD in DSM-III-R were revised slightly from
DSM-III. The DSM-III item "intolerance of being alone" was replaced with
"frantic efforts to avoid real or imagined abandonment" in DSM-III—R, and
the DSM-lll item "physically self-damaging acts" was changed to "recurrent
suicidal thoughts, gestures, or behavior, or self-mutilating behavior" in DSM-
III-R DSM-lII-R also revised the diagnostic criteria for Schizotypal
Personality Disorder, adding the item "odd or eccentric behavior or
appearance" (American Psychiatric Association, 1987). DSM-III and DSM—III-R
diagnoses have been found to be narrower and less inclusive than diagnoses
based on the DIB or Spitzer et al's criteria.

The DSM-IV diagnostic criteria for Borderline Personality Disorder
retained the basic eight diagnostic criteria from DSM-llI-R, and added one
more: "transient, stress-related paranoid ideation or severe dissociative
symptoms" (American Psychiatric Association, 1994, p. 654). The DSM-IV
diagnostic criteria for Schizotypal Personality Disorder remained essentially
the same as in DSM-IlI-R. Nevertheless, the fine tuning of descriptive
characteristics over time represents a diagnostic trend toward narrower
definitions of "borderline", which are based on similarity of symptoms rather
than similarity of underlying personality organization. While this narrowing
of diagnostic criteria has increased reliability of diagnoses and increased
homogeneity, psychodynamically oriented theorists question whether the
narrower definitions are unnecessarily exclusionary.

Psychodynamic formulations.

The current psychoanalytic literature on borderline syndromes can be
roughly divided into those who view borderline pathology as a result of
structural personality deficits and those who view it as a result of structural

personality conflicts (Goldstein, 1990). Those who adhere to deficit models
generally see borderline pathology as a developmental arrest at pre-oedipal
stages that are critical for the formation of object constancy, separation-
individuation, and integrated representations of self and others. Authors
who fall into this theoretical camp include Blanck and Blanck, Kohut,
Stolorow and Brandchaft, and Adler and Buie (Goldstein, 1990). Other
theorists, such as Kernberg, see borderline behavior arising from personality
conflicts in addition to lack of development in these same personality
structures.

Of the competing theories about the nature of borderline dynamics,
perhaps the most developed and influential recent conceptualization has
been that of Otto Kernberg. Kemberg's view is that borderline pathology
represents a level of personality organization rather than a specific behavioral
syndrome. According to Kernberg (1985), this borderline personality
organization is defined by particular intrapsychic "structures"; identity
integration, defensive constellations, quality of reality testing, ego strength,
and object relations. The constellation of borderline level defenses consists of
splitting, primitive idealization, projection and projective identification,
primitive denial, and omnipotence and devaluation. Furthermore, the
tendency toward primary process thinking, especially in unstructured
situations, without complete loss of reality testing is another hallmark sign of
borderline level pathology. Ego-weakness specifically includes lack of anxiety
tolerance, lack of impulse control, and lack of developed sublimatory
channels (Kernberg, 1985). Finally, object relations characterized by polarized
"good" and "bad" representations, due to defensive splitting, dominate the
borderline's interpersonal world.

Kernberg's concept of borderline personality organization (BPO) is a
broader construct than borderline personality disorder (BPD), encompassing
different personality types. His "infantile" personality type, which roughly
corresponds to the phenomenology of the borderline personality disorder of
other diagnostic systems, falls within the borderline range of personality
organization, as does schizotypal personality disorder and several other
personality types. Patients in other diagnostic categories, depending on their
capability for reality testing, defensive clusters, and maturity of object
relations, fall within the range of either Neurotic Personality Organization
(NPO) or Psychotic Personality Organization (PPO).

According to Kernberg, the overtly chaotic behavior of borderlines
results in part from strong aggressive conflicts, stemming from either pre-
oedipal maternal failures or excessive constitutional aggression. The
borderline attempts to cope with this aggression, which threatens to overtake
and destroy the good internal objects, with relatively underdeveloped and
ineffective defenses and ego controls. One way that the internal objects are
protected is through the defensive mechanism of splitting. Kernberg
describes how splitting may be manifested several ways, for instance in
contradictory expressions of complementary sides of a conflict, division of
objects and self-representations in to "all good" ones and "all bad" ones,
sometimes involving sudden reversals, and oscillation between contradictory
self-concepts. Other borderline defenses, such as primitive idealization,
omnipotence and devaluation, denial, and projective identification, are
closely related to splitting. While some believe that splitting results from the
inability to synthesize contradictory images, Kernberg sees splitting as an

active defense.

10

Furthermore, Kernberg sees these various borderline structural
indicators, identity diffusion, primitive defenses, ego weakness, and object
relations as mutually sustaining. For instance, splitting prevents
neutralization of aggression, which in turn leads to more splitting. This
vicious cycle is believed by Kernberg to be a main cause of ego weakness and
its behavioral manifestations in borderlines. Kernberg also sees object
relations and degree of primary process thinking as related in borderlines.

Kernberg (1975) designed an interviewing method for determining the
level of personality organization of individual patients. The interview
utilizes clarification, confrontation, interpretation, and transference
interpretation to elicit three primary characteristics of intrapsychic structure:
degree of identity integration, level of defensive operations, and reality
testing. Unlike neurotic patients who have more integrated identities and
higher level defenses, borderline and psychotic patients are presumed to have
fragmented identities and primitive defenses. Despite the borderline's
tendency toward greater primary process thinking, borderline and neurotic
patients should both demonstrate intact reality testing, while psychotic
patients will have grossly impaired reality testing (Kernberg, Goldstein, Carr,
Hunt, Bauer, & Blumenthal, 1981). These differences between levels of
personality organization have been organized by Acklin (1993), summarized
in Appendix A.

According to Kernberg, development in the underlying personality
organization, specifically defensive functioning and anxiety tolerance, would
allow the gradual integration of previously split off representations, which
would neutralize aggression, allowing in turn more energy to be devoted to

ego development. The development of these interrelated structures of the

ll

personality would therefore allow the individual to function in a more
cohesive, rather than fragmented and chaotic, manner. This would be
expected to have a stabilizing effect on the person's symptoms, relationships,
and general level of functioning.
Overlap between the BPD and BPO Constructs

There is evidence to support Kernberg's formulations of borderline
personality organization being a more inclusive category than the syndrome
borderline personality disorder. Kernberg et a1. (1981) conducted a study with
48 inpatients diagnosed as having either borderline (52%) or psychotic (48%)
personality organization by Kernberg's structural interview. Patients were
also diagnosed as either schizophrenic or borderline according to the
Diagnostic Interview for Borderlines (DIB; Gunderson, Kolb, 8: Austin, 1981)
psychological testing, and clinical diagnoses. Of the patients diagnosed as
having borderline personality organization (BPO), only 60% also met DIB
criteria for borderline personality disorder (BPD) (Kullgren 8r Armelius, 1990).
However, 75% of those who had borderline personality disorder were
diagnosed as having borderline personality organization, while the
remaining 25% had either psychotic personality organization or were
undetermined.

In a related study, Nelson, Tennen, Tasman, Borton, Kubeck, and
Stone (1985) studied 50 inpatients who were diagnosed with multiple
instruments. A subsample of 30 inpatients were given diagnoses of either
neurotic, borderline, or psychotic personality organization by Kernberg's
structural interview, as well as DSM-III and DIB diagnoses, and were rated
according to a checklist of Spitzer et al.'s criteria. These authors found a
higher overlap of BPO and BPD: while 67% of those diagnosed as BPO also

12

were BPD, 92% of those who were BPD were also BPO. However, the authors
point out that patients with BPD were over-represented in the subset of the
entire sample who undertook the structural interview, which may have
inflated the concordance between BPO and BPD in this study.

Stone (1987) also conducted a small pilot study on 18 inpatients who
were diagnosed according to three systems: Kernberg's structural interview,
the Diagnostic Interview for Borderlines (DIB), and the DSM-III criteria for
borderline personality disorder. They found that while not all patients
diagnosed borderline by Kernberg's classification met DIB or DSM—III criteria
for BPD, 5/ 6 of DIB diagnosed borderlines and 9 / 10 DSM-III diagnosed
borderlines met criteria for BPO. The one patient who was BPD by the DIB
but not by Kernberg's structural criteria had a psychotic personality
organization.

Finally, a similar study by Kullgren and Armelius (1990) attempted to
assess the concordance of different borderline diagnoses. Forty-four
inpatients were diagnosed according to Kernberg‘s structural interview, the
DIB, and DSM-III criteria. Twenty-four met criteria for borderline personality
organization (BPO). Although the authors did not give a breakdown
including DIB scores, they did report on the overlap between DSM-Ill and
Kernberg's structural diagnosis. Of the 15 patients who met DSM-III criteria
for BPD, 11 met criteria for BPO, 4 met criteria for neurotic personality
organization, and none met criteria for psychotic personality organization.

These studies support Kernberg's definition of borderline personality
organization as broader than the syndrome of BPD in both DSM-III and the
DIB. For the most part, behaviorally diagnosed borderlines are also
considered borderline by Kernberg's structural diagnosis. However, these

13

studies that have used the structural interview have found a small
percentage of patients who meet criteria for BPD but fall into a category of
either psychotic or neurotic personality organization.
The Qpestion of Diagpostic Stabiligg

One defining aspect of borderlines that both the symptomatic and
psychodynamic viewpoints seemingly agree on is the stability of the disorder.
According to the DSM-III (American Psychiatric Association, 1980), the Axis II
disorders, including borderline personality disorder, should be diagnosed
only when "the characteristic features are typical of the individual's long-
term functioning and are not limited to discrete episodes of illness."
(American Psychiatric Association, 1980, p. 305). The Diagnostic Interview for
Borderlines (Gunderson, Kolb, 8: Austin, 1981), the most widely used research
measure of BPD, requires that the symptoms be present consistently over a
period of months and sometimes years in order to contribute to the diagnosis.
In describing this measure, Gunderson et a1. (1981) state: "The fact that the
sections on social adaptation, impulse/ action patterns, and interpersonal
relations are largely based on historical information is in keeping with the
view that the borderline diagnosis represents enduring behavioral patterns,
or character structure, rather than a symptomatic reaction" (p. 896). Kernberg,
in describing a more inclusive group of patients with borderline personality
organization, wrote:

The term "borderline" should be reserved for those patients presenting

a chronic characterological organization which is neither typically

neurotic nor typically psychotic, and which is characterized (i) by typical

symptomatic constellations; (ii) by a typical constellation of defensive

operations of the ego; (iii) by a typical pathology of internalized object

14

relationships; and (iv) by characteristic genetic-dynamic features.

(1985, p. 5).

Thus, according to Kernberg, overt borderline symptomology as well as
underlying defensive and representational structures should remain stable
over time.

Research findings on the stability of borderline disorders has produced
mixed results. Longitudinal research to date on borderlines, divided into
short term studies (2-5 years) and long term studies (10 years and over), has
focused almost exclusively on symptomatic stability, without consideration
for underlying stability of borderline dynamics. Conclusions from reviews of
the follow-up research literature are that borderlines show relatively little
symptomatic change over short periods, but do manifest symptomatic
changes over longer periods of time. In reviewing the borderline follow-up
studies, Stone (1990a) wrote "...the life course of the typical hospitalized
borderline patient goes through a rather long period of poor functioning-
perhaps 5 to 8 or 10 years - followed by gradual mellowing and improvement,
especially as the patient enters the fourth decade of life." (p.2).

Although reviews of short-term follow-up studies have generally
concluded that the borderline diagnosis is stable for the first five years post-
index and continue to suffer from severe psychopathology (Links, Mitton, &
Steiner, 1990), a surprising number of studies have found a substantial
percentage of previously diagnosed borderlines who no longer meet
diagnostic criteria after only 2-5 years. For example, Barasch, Frances, Hurt,
Clarkin, and Cohen (1985) found at a 3 year follow-up that only 6 of 10 DSM-
III diagnosed borderlines and 7 of 13 DIB diagnosed borderlines retained their
diagnosis at follow-up. Pressler (1990), in an unpublished doctoral

14

relationships; and (iv) by characteristic genetic—dynamic features.

(1985, p. 5).

Thus, according to Kernberg, overt borderline symptomology as well as
underlying defensive and representational structures should remain stable
over time.

Research findings on the stability of borderline disorders has produced
mixed results. Longitudinal research to date on borderlines, divided into
short term studies (2-5 years) and long term studies (10 years and over), has
focused almost exclusively on symptomatic stability, without consideration
for underlying stability of borderline dynamics. Conclusions from reviews of
the follow-up research literature are that borderlines show relatively little
symptomatic change over short periods, but do manifest symptomatic
changes over longer periods of time. In reviewing the borderline follow-up
studies, Stone (1990a) wrote "...the life course of the typical hospitalized
borderline patient goes through a rather long period of poor functioning—
perhaps 5 to 8 or 10 years - followed by gradual mellowing and improvement,
especially as the patient enters the fourth decade of life." (p.2).

Although reviews of short-term follow-up studies have generally
concluded that the borderline diagnosis is stable for the first five years post-
index and continue to suffer from severe psychopathology (Links, Mitton, &
Steiner, 1990), a surprising number of studies have found a substantial
percentage of previously diagnosed borderlines who no longer meet
diagnostic criteria after only 2-5 years. For example, Barasch, Frances, Hurt,
Clarkin, and Cohen (1985) found at a 3 year follow-up that only 6 of 10 DSM-
III diagnosed borderlines and 7 of 13 DIB diagnosed borderlines retained their
diagnosis at follow-up. Pressler (1990), in an unpublished doctoral

15

dissertation, found that 1 1 of 24 (46%) previously DIB diagnosed borderlines
retained the borderline diagnosis at 2-5 year follow-up. Similarly, Links,
Mitton, and Steiner (1993) found that 26 of 65 (40%) of previously DIB
diagnosed borderlines no longer met diagnostic criteria at a 2 year follow-up.
Thus, the research literature has consistently found that just under half of
previously symptom— diagnosed borderlines do not meet diagnostic criteria in
less than 5 years after initial diagnosis. Taking into account sampling biases
inherent in follow-up studies and especially poor outcomes (such as suicide)
not included in follow-up samples, diagnostic stability seems lower than
expected.

Thus, the literature challenges the notion that borderline syndromes
are necessarily stable, even over short periods of time. The question raised by
these findings is: do symptomatic changes reflect real personality changes, or
are they merely temporary adjustments to situational life circumstances?

Unfortunately, the lack of carefully designed studies that assess the
stability of personality organization precludes definite conclusions. However,
results from long-term follow-up studies and case reports suggest that certain
borderline patients may make remarkable changes. In his careful and
exhaustive long-term study of previously hospitalized DSM-l'II diagnosed
borderline patients, Stone (1990b) reported considerable variability in the
outcomes of borderlines. While many borderlines remained severely
impaired at follow-up, some made substantial life adjustments and appeared
to function at a neurotic level of personality organization. Although Stone
reports mostly data on the symptomatic improvement of the borderlines, he
also gives his clinical impression that some of the borderlines at follow-up
had undergone true personality maturation. Likewise, McGlashan (1986)

16

reported that the borderlines in his long-term follow-up study were
distributed bimodally in terms of the quality of their relationships with
others. He stated:
One group was functioning well and and had managed to create and
maintain meaningful relationships with stability over time...The other
bimodal cluster of patients, however, essentially dealt with this
problematic area by studious avoidance of relationships. They
appeared to be people who concluded that their emotional equilibrium
required abstinence in object relations. (p. 29).
Thus, variability in the method and nature of interpersonal adjustment
seems to be expected in outcome studies of borderlines.
Findings from the one follow-up study by Kullgren and Armelius
(1990) on the long-term stability of structural diagnoses suggest that patients
who are diagnosed with BPO at index may change their structural diagnosis
over time. Thirty-four patients were diagnosed with Kernberg's structural
interview at both index hospitalization and 5 year follow-up. Of the 21
previously BPO patients, at follow-up 9 were still BPO, 5 were classified with
neurotic personality organization, and 7 were classified with psychotic
personality organization. Thus over half (57%) of the previously classified
BPO patients received a different structural diagnosis at follow-up. Notably,
the authors found that all of the index borderlines with a dual diagnosis of
affective disorder were diagnosed as having neurotic personality organization
at follow-up. Unexpectedly, half of those whose personality organization had
changed at follow-up went from a borderline to a psychotic personality
organization. Earlier follow-up studies of borderlines have not reported any
cases of borderlines becoming schizophrenic at follow-up (Carpenter 8r

17

Gunderson, 1977; Pope, Jonas, Hudson, Cohen, & Gunderson, 1983; Werble,
1970) . Such a dramatic worsening of intrapsychic structure is not predicted by
either Kernberg's theory or previous follow-up studies, and is difficult to
interpret.
Several methodological limitations of the Kulllgren and Armelius

(1990) study warrant caution in interpreting the results. Neither initial DIB
nor follow-up DIB and DSM-III diagnoses were reported to ascertain whether
or not the changes in structural diagnoses corresponded to changes in these
diagnostic measures as well. More specific diagnostic information about
Kullgren and Armelius' BPO subjects is needed in order to apply these
findings to the potential for change in patients with BPD. For example, in his
study on the overlap between Kernberg, DSM-III, and DIB diagnostic criteria,
Stone (1987) found that patients diagnosed BPO but not BPD were healthier
than those diagnosed with both BPO and BPD. As Kernberg (1985) has
pointed out:

Retuming to the diagnosis of the patient's character structure as an

important prognostic element in borderline conditions, it is the type of

character pathology which is prognostically important, in addition to

the general importance of ego syntonicity or ego dystonicity of the

character traits. It is not enough to diagnose a patient as presenting

"borderline personality organization." (Kernberg, 1985, p. 113).
Thus, the dramatic improvements and decrements seen in the personality
organization of the patients in this study may be related to the nature of their
particular character types, which are unknown.

Nevertheless, attempts to explain the symptomatic "recovery" of a

subset of borderline patients have led to longitudinal studies that have

18

assessed predictors of outcome. Contradictory or negligible results have been
obtained for the roles of most kinds of treatment (Paris, Brown, & Nowlis,
1987; Stone, 1990b); concurrent depressive disorders (Pope et al., 1983; Stone,
1990b) and a host of demographic and other factors (Stone, 1990b).

One predictor that has been consistently found to be a correlate of
symptomatic improvement in borderlines is age (Links, Mitton, 8t Steiner,
1993; McGlashan, 1986; Paris, Brown, & Nowlis, 1987; Stone, 1990b). One
conclusion that authors of long-term follow-up studies make about the
course of borderline pathology is that is that many borderline symptoms, for
instance impulsivity, dyphoria, and reckless behavior, decline naturally as a
function of aging. In line with this, Paris et a1. (1987) have remarked on the
"tendency of personality disorder symptomology in general, and BPD in
particular, to decrease in severity and prevalence into middle age" (p. 533).
The relatively young age of hospitalized borderline patients compared to
other patient groups has been reported by several studies (e.g., Stuart et al.,
1990). One explanation for the findings of dramatic short-term changes in
borderlines is that those patients who were no longer borderline at follow-up
were older at the time of initial assessment, and underwent greater
maturation and stabilization in line with the age trajectory described by Paris
et a1. (1987). According to this set of predictions, we would expect greater
improvements in personality organization in those who were older at index.
We would also expect to find changes in diagnostic status (BPD versus not
BPD) concordant with these changes in personality organization. Assuming
there are indeed real underlying changes taking place, it would be important

to consider the influence of age on follow-up outcome.

19

However, a more pessimistic explanation of these findings is that
fluctuations in behavior and symptoms, central to the borderline syndrome
diagnosis, may reflect situational factors more than intrapsychic ones.
According to this perspective, certain patients who maintain a stable
borderline personality organization may not manifest 'borderline' behavior
under optimal environmental circumstances. For example, Paris (1988) has
found that symptomatic "improvement" in some patients previously
diagnosed as borderline may reflect increased interpersonal isolation. In this
case, we would expect some patients to remain borderline in personality
organization at follow-up, while showing enough behavioral improvement
that they no longer meet BPD diagnostic criteria.

Nevertheless, the results of the long-term longitudinal studies,
combined with the preliminary results of the study by Kullgen and Armelius
(1990), provide some evidence that some borderlines may be undergoing
significant structural changes over relatively short periods of time. Although
contrary to most theoretical notions of borderlines, which assume stability,
the suggestion that personality organization in borderlines might change is
compelling. This prediction is in line with the statement of Sugarman and
Lerner (1980): "The conceptualization of borderline personality organization
as a stable configuration perhaps represents a premature closure in
understanding this group of patients noted for unpredictable and dramatic
shifts, variability, and fluid ego states" (p. 19).

To summarize, the question of whether the apparent diagnostic
changes in borderlines over short-term follow-up periods reflects changes in
personality organization remains to be answered. Given that the Kullgren
and Armelius (1990) study is a single report with significant flaws, this study

20

endeavors to answer the same question while improving on the
methodology.

In addition to the methodological factors mentioned previously,
several others require attention. First, it seems likely that the reliability of the
method for assessing the level of personality organization, the structural
interview, is subject to fluctuations in the skill level and personality
characteristics of the interviewer. Given that there is no data available on the
test-retest reliability or inter-judge reliability of the structural interview, it is
difficult to judge whether these apparent changes in personality organization
are real or whether they are an artifact of problems in measurement.
Furthermore, considerable subjectivity in the rating of the patient, given the
unquantified nature of the material, may also lead to decreased reliability. It
is impossible to assess the relative weight the interviewers give to each
behavioral indicator of structural organization. In addition, the interpersonal
nature of the structural interview, in which the patient is pressured and
challenged by the interviewer, may be more subject to distortions, rather than
clarifications, of the patient's functioning. It is worth questioning whether
the measurement of personality structure would be as heavily influenced if
the method of structural diagnosis was less interpersonally determined, more
quantifiable, and reliably assessed.

Another shortcoming of studies that use the structual interview to
determine personality organization is that the structural interview yields
only global information: neurotic, borderline, or psychotic level of
personality organization. Although this global categorization is conceptually
relevant, we would want to know something about changes in the specific
components that contribute to the determination of personality organization

21

(Murray, 1993). For example, it would be important to know whether the
borderline indicators change, if at all, uniformly or unevenly, and whether
they cluster together according to Kernberg's theoretical predictions. It would
be also be helpful to determine whether particular components of BPO
correspond with changes in BPD diagnosis better than others. The
measurement of personality organization with a more reliable and less
subjective method might clarify these intriguing questions.

The Use of the Rorschach in Assessr_11g' Personality Structure

As mentioned earlier, the role of psychological testing with borderlines
occupied a central place in the historical development of the borderline
concept. Coincident with the identification of borderline patients in
psychiatric and psychoanalytic settings, pioneers in psychological testing
identified a group of patients who appeared to have a combination of
neurotic and psychotic features. These patients performed relatively well on
the structured WAIS, but showed profound regressions toward primary
process thinking on unstructured tests such as the Rorschach. Indeed, the
formulation "good WAIS, bad Rorschach" became almost pathognomonic for
the borderline personality in the psychological testing literature.
Convergence has been obtained between clinical diagnoses based on the
Rorschach and diagnosis using Kernberg's structural interview that assesses
the level of personality organization in borderline, psychotic, and neurotic
patients (Kernberg et al., 1981).

Although there is continuing debate about the exact nature of the
Rorschach task and the information that it yields, most earlier Rorschach
theorists agree that it goes beyond being a purely perceptual task to reflect
aspects of the respondant's personality and intrapsychic world. Schafer (1954)

22

has been influential in shaping the widely held view that test pulls for a
combination of primary and secondary process material that may be useful in
the assessment of both unconscious derivatives and conscious processes.
Similarly, Mayman (1967) has stated:
When a person is asked to spend an hour immersing himself in
a field of impressions where amorphousness prevails and where
strange or even alien forms may appear, he will set in motion a
reparative process, the aim of which is to replace formlessness
with reminders of a palpably real world...A person's most readily
accessible object-representations called up under such
unstructured conditions tells much about his inner world of
objects and about the quality of relationships with these inner
objects toward which he is predisposed. (p. 19).
More recently, Leichtrnan (1996) has described the response process as
involving a "representational" process, which is an active construction of the
Rorschach percepts. Regarding this he writes:
...Representation is unmistakably a social process. It begins and ends as
a form of communication, even if the audience is internal. Just as
artists' needs, wishes and intentions and the character of their real or
imagined audiences affect the form and content of their work, so too do
these factors enter into the creation of Rorschach images...1n addition,
when the audience is viewed as the self, psychodynamic processes such
as defenses can be encompassed by conceiving of them as factors
shaping communication within a system of internal representation

(p. 486).

23

Even Exner, who has been a proponent of the more perceptual view of
Rorschach responses, has recently advocated a more interpretive
understanding of the projective value of personal material in the responses
(Exner, 1996). However, Exner has contributed to a better understanding of
the Rorschach response process as a complex interplay of perceptual factors as
well as conscious and unconscious selection of reponses, which include both
deliberate censorship and selective reporting of responses which are
concordant with personality traits (Exner, 1986a, 1996).

Many theorists and clinicians consider the Rorschach to be an
indispensable tool in detecting borderline pathology. According to Acklin
(1993) "...when used in a manner that integrates empirically based nomothetic
and theory-saturated idiographic analysis [Acklin 1992], the test is
unparelleled in graphically assessing and displaying the underlying
structural, affective, and representational features of the borderline's inner
world" (p. 338). Arnow and Cooper (1984) have gone beyond stating that the
Rorschach captures the regressed features of the borderlines inner world to
argue that it actually provokes a regression because of the disorganization
generated by the unstructured testing situation.

The clinical observations of psychological testing with borderlines have
been largely supported by research that has found that borderlines as a group
exhibit identifiable characteristics on projective tests, in particular the
Rorschach. A recent resurgence of interest in psychological testing with
borderlines has led to a growth of research aimed at operationalizing and
identifying borderline characteristics on projective tests. The development of
scoring systems to identify domains of functioning has at last permitted
researchers to test psychodynamic theories of underlying borderline

24

personality structure. For the sake of clarity, I will review the literature
specifically on those aspects of borderline functioning that are major
components of the structural personality organization discussed by Kernberg:
Identity Diffusion, Defensive Operations, Reality Testing, Ego Weakness, and
Object Relations.

Identity Diffusion.

Identity diffusion has been one of the indices of borderline personality
organization that has been least studied with regard to Rorschach variables.
As with the interpretation of dream material, the responses may reflect a
combination of self and other representations which may not be easy to
distinguish. Identity diffusion as a construct may also be expressed in
contradictory representations of self across time, which may or may not be
captured in a single administration of the Rorschach. Nevertheless, Carr
(1987) has stated that the fabulized combination response of Exner's (1986b)
scoring system is "probably the best single index of the borderline's identity
diffusion" (p. 352). Exner (1986b, 1990) derived his scoring of FABCOM from
Schafer (1954), which refers to percepts in which there is an implausible
relationship between two discrete and separate details of the blot. While
some studies assessing this score in borderlines have found a much higher
rate of fabulized combination in borderlines than schizophrenics (Thaler
Singer 81: Larson, 1981), others have found no difference between borderlines
and other patient groups (Berg, 1990; Buttenheim, Lohr, 8r Kerber, 1985).
Furthermore, FABCOM has been seen in less than one half of DSM-III
diagnosed borderlines (Exner, 1986a). To date, no study has specifically set out
to validate a Rorschach measure of identity diffusion, and therefore identity
diffusion appears to be unreliably measured by the Rorschach at this time.

Defensive Qperations.
Several empirical coding systems for the Rorschach have been devised

to measure specifically borderline defenses, such as splitting, primitive
devaluation, idealization, denial, projection, and projective identification
(Cooper, Perry, 8r Arnow, 1988; Grala, 1980; Lerner, Sugarman, & Gaughran,
1981). Two of the most widely used systems used for scoring borderline
defenses are the Rorschach Defense Scale (RDS) developed by Cooper and
colleagues (Cooper, Perry, 8: Arnow, 1988), and the Lerner Defense Scale (LDS)
developed by Lerner and colleagues (Lerner, Sugarman, 8r Gaughran, 1981).
The RDS assesses both primitive and higher level defenses from human and
animal responses on the Rorschach, specifically: isolation, intellectualization,
reaction formation, rationalization, repression, devaluation, primitive
idealization, projective identification, splitting, omnipotence, projection,
pollyannaish denial, higher level denial, massive denial, and hypomanic
denial. The LDS uses only human responses on the Rorschach and assesses
only primitive defenses: splitting, denial, devaluation, idealization, and
projective identification.

Although Cooper and colleagues conceptualize the RDS from a
Kohutian perspective and Lerner and colleagues conceptualize the LDS from
a Kernbergian perspective, Carr (1987) has been critical of making theoretical
distinctions in the absence of confirmatory data. Theoretical criticisms aside,
preliminary validity studies have shown some evidence for the usefulness of
both scales in discriminating between DIB or DSM-HI diagnosed borderlines
and other groups, such as neurotics, anorexics, and schizophrenics (e. g.
Lerner, 1990). In support of the validity of the LDS, borderlines have been

shown to overlap with narcissistic groups in their use of some defenses,

26

particularly devaluation, idealization, and denial, while they were
significantly different on defenses of splitting and projective identification
(Lerner, 1990). However, in line with Kernberg's conceptualization of
borderline personality organization, other studies using the LDS have not
found borderlines to differ grossly in their defenses from other personality
disordered groups, such as those thought to share an underlying borderline
personality organization (Lerner, 1990).

Lerner (1990) compared the LDS and the RDS in their respective ability
to differentiate four groups: outpatient neurotics, outpatient borderlines,
inpatient borderlines, and schizophrenics. Using composite defense scores,
Lerner found that both the RDS and LDS discriminated different patient
groups. He found that the RDS effectively discriminated between neurotics
and inpatient borderlines, outpatient from inpatient borderlines, and
outpatient borderlines from schizophrenics; the RDS did not discriminate
between neurotic and outpatient borderlines, or outpatient borderlines from
schizophrenics. The LDS effectively discriminated neurotics from inpatient
borderlines, outpatient borderlines from schizophrenics, and inpatient
borderlines from schizophrenics; the LDS did not discriminate between
neurotics and outpatient borderlines, or between inpatient and outpatient
borderlines. The author concludes: "The Cooper Scale is more effective in
distinguishing between healthier outpatients, whereas the Lerner & Lerner
Scale better discriminates more seriously disturbed inpatients" (p. 42).

With regard to particular borderline defenses, the Lerner Scale found
that all five borderline defenses significantly distinguished borderlines from
neurotics and schizophrenics. Splitting, devaluation, idealization, and denial

all differentiated the groups, while projective identification was found

27

exclusively in the borderline groups. The defenses on the RDS were also
examined separately, yielding a slightly different pattern of discrimination:
splitting differentiated inpatient borderlines from neurotics and
schizophrenics, devaluation and omnipotence differentiated outpatient
borderlines from neurotics and schizophrenics, while idealization and
projective identification did not discriminate between groups.

In summary, the two systems designed to detect borderline defensive
functioning as expressed on the Rorschach have been shown to have validity
in discriminating patients with borderline personality from other patient
groups. Thus, there seems to be a sound basis for the notion that borderlines
as a group employ certain identifiable defenses on the Rorschach.

Reality Testing.

In line with the early observations of borderline Rorschachs, transient
psychotic features and specific types of thought disorder have been found to
be characteristic of patients diagnosed with borderline personality disorder in
several studies (Gartner, Hurt, 8: Gartner, 1989). Most studies have found
that while borderlines have more illogical, bizarre, and confabulated
responses on the Rorschach compared to neurotic and normal populations,
they are able to maintain better form quality than schizophrenics. For
example, Thaler Singer and Larson (1981) used Friedman's system for scoring
the form quality of the Rorschach, and found that a sample of clinically
diagnosed borderlines (prior to DSM-III and Gunderson's criteria) showed
greater fabulized combinations as well as greater decline in the form quality of
responses across cards than the neurotic and normal comparison groups.
However, the borderlines in this study showed fewer signs of serious thought
disorder than the schizophrenics. Notably, the authors suggest that the

28

borderline group contained some subjects who would later qualify for both
schizotypal and borderline DSM-III diagnoses; thus, we would expect their
thought disorder scores to look somewhat worse than a pure BPD group.

Patrick and Wolfe (1983) used Holt's scoring system for primary process
thinking in a group of DSM—III diagnosed borderlines and found that they had
evidence of bizarre ideation and thinking disturbance and elevated aggressive
and libidinal content. However, form quality indices in this sample were not
suggestive of disturbance.

Another study by Exner (1986a) used the Comprehensive System for
scoring Rorschachs, comparing DSM-III diagnosed borderlines, schizotypals,
and schizophrenics, with non-overlapping diagnoses. He found a distinct
pattern for each group on the Rorschach, with the schizotypal patients
resembling the schizophrenics more than the borderlines. When the specific
scores related to thought disorder were compared, borderlines were found to
have fewer special scores (specifically WSum6) than both schizotypals and
schizophrenics. Furthermore, an X-% score greater than 15% in each
protocol, indicating perceptual problems of significant magnitude to interfere
with functioning, were found significantly more often in the schizotypal and
schizophrenic groups compared to the borderlines; 87% of the schizophrenic
group, 63% of the schizotypal group, and only 27% of the borderline group
scored above 15% on this variable. Finally, a score of 4 or 5 on the
Schizophrenia Index (SCZI), indicating serious thought disorder, was found
in 82% of the schizophrenic patients, 37% of the schizotypal patients, and only
13% of the borderline patients. These results suggest that while borderlines
do show evidence of thought disorder, it is less severe and less frequently

encountered than in schizophrenic or purely schizotypal groups.

29

In a review of the borderline testing literature, Gartner et a1. (1989)
conclude that borderlines across studies can be discriminated from normals
and schizophrenics in the form quality of their responses. They report that
the average F+%, indicating good perceptual accuracy, ranges from 67% to
70% in DSM-III and DIB diagnosed borderlines across studies, with
borderlines scoring "approximately 10% to 15% lower than normals and 10%
to 15% higher than schizophrenics" (p. 427). They further conclude that
while borderlines are more likely to have minus and weak form responses
than normals, schizophrenics are more likely to have more minus form
responses than weak form responses, while borderlines show relatively more
weak than minus responses.

In summary, the empirical literature supports the commonly held
notion that borderlines do manifest thought disturbances on the Rorschach.
However, it appears that borderlines as a group can be distinguished from
both normals, neurotics, and schizophrenics in the severity and quality of
their thought disturbance indices. A review of the literature suggests that
borderlines show more fabulized combinations and a greater decline in
response quality than both normals and schizophrenics, although they show
less severe scores on Exner's Schizophrenia Index (SCZI), Special Scores
(WSum6), and determinants of form quality (F+, Fo, Fw, or F-) than
schizophrenics.

Ego Weakness.

Kernberg's definitional criteria for patients with borderline personality
organization includes non-specific manifestations of ego weakness, which can
be broken down into components: lack of anxiety tolerance, lack of impulse

control, and lack of developed sublimatory channels (Kernberg, 1985). In

30

general, the ego capacity of borderlines is characterized by problems
containing affect and impulses which interferes with deliberate problem
solving and delaying of gratification. Also taken as an indication of ego
weakness in a previous study of borderlines is their demonstrated progressive
decline of response quality across cards (Thaler Singer 8r Larson, 1981). In
Exner‘s (1986a) study, described above, he found that borderlines showed
greater indices of impulsivity (FC: CF+C and D) than schizophrenics or
schizotypals. He also found that borderlines scored significantly lower than
schizophrenics on indices of stress tolerance (D, Ade), and higher than
schizophrenics on an index of internally perceived stressors (es).
Furthermore, the borderlines averaged a score of 0.66 on EA, a measure of
coping ability, which indicates as a group they show limited resources for
deliberate coping strategies. Berg (1990) has also looked at the ego weakness of
clinically diagnosed borderlines using the Rorschach. She ‘operationalized
impulse control and affective regulation by looking at Exner's variables CF, C,
FC, Afr, and Lambda, and found no differences between the borderline
sample and a comparison group of clinically diagnosed narcissists on these
specific variables. However, due to the uncertain diagnosis of this sample it is
difficult to apply these findings to other groups of borderlines. Although
there has not been a scale or coding system developed to measure ego
characteristics, it does appear that in general there is evidence to suggest that
borderlines show global deficits in variables that reflect impulse control and
affective regulation on the Rorschach.

Object Relations.

A great deal has been written about borderline object relations as they

are manifested on projective tests. Several scoring methods for assessing

31

object relations on the Rorschach have been developed, including the
Empathy-Object Relations Scale (Pruitt 8t Spilka, 1964 ), the Blatt
Developmental Object Relations Scale (Blatt, Brenneis, Schimek, 8: Click,
1976), the Mutuality of Autonomy Scale (Urist, 1977), and the Human
Experience Variable (Perry 8: Viglione, 1992). Several studies of borderline
object relations have been conducted using the Blatt et a1. (1976) scale, and
have consistently found that DSM-III and DIB borderlines score higher on
this scale than schizophrenics (e.g., Lerner 8r St. Peter, 1984; Spear 8r Lapidus,
1981; Spear 8r Sugarman, 1984). In a study by Lerner and St. Peter (1984),
inpatient and outpatient borderlines were distinguishable from
schizophrenics and neurotic outpatients based on their accuracy of human
perceptions and developmental scores.

In line with Kernberg's conceptualization, borderlines have been
found to consistently portray objects as more malevolent than other patient
groups on a range of projective tests (e.g., Gacono, Meloy, Gr Berg, 1992;
Lerner & St. Peter, 1984; Segal, Westen, Lohr, Silk, 6: Cohen, 1992; Stuart et
al., 1990) and scoring systems. For example, using the Blatt Object Relations
Scale, Stuart et a1. (1990) found that DIB diagnosed borderlines portrayed
objects as more actively malevolent than did either depressives or normals.
Another study by Gacono, Meloy, and Berg, (1992) used a different scoring
system for primitive object relations developed by Kwawer (1980). These
authors found that DSM-III-R diagnosed borderlines showed more
malevolent representations than either DSM-III-R diagnosed antisocial
personality and narcissistic personality disordered patients. Furthermore, the
borderlines displayed percepts that were characterized specifically by

"malignant internal processes (78%) and violent symbiosis and reunion

32

(67%)." (p. 41). In addition, the study by Lerner and St. Peter (1984) described
above found that the percepts of both inpatient and outpatient borderlines in
the sample were characterized by malevolent representations, which
distinguished them from the neurotic and schizophrenic comparison groups.

Another aspect of borderline object relations that has been
operationalized and studied on the Rorschach is the quality of the borderline
patient's boundaries. In support of the psychodynamic conceptualization of
incomplete differentiation of self from other, several studies have found that
borderlines display boundary confusion on projective tests. For example,
Coonerty (1986) developed a scale to measure Separation-Individuation
themes on the Rorschach, which she divided into preseparation-
individuation and separation-individuation themes. She found that while
DSM-III diagnosed borderlines and schizophrenics do not differ in the
combined amount of separation-individuation themes on the Rorschach,
borderlines have more separation-individuation themes, while
schizophrenics have more pre-separation-individuation themes. In the study
by Gacono et a1. (1992) previously mentioned, Kwawer's (1980) scale of
primitive object relations was used. They found that 67% of the borderlines
produced Rorschach themes of "violent symbiosis separation and reunion"
(p. 41), compared with 45% of antisocial personalities, and 33% of narcissistic
personalities. Urist's (1977) Mutuality of Autonomy Scale was also developed
to assess the developmental progression of the human representations on the
Rorschach, ranging from representations of autonomous but interactive
pe0ple to enmeshed, hostile and controlling relationships. This measure has
been shown to be positively related to independent measures of relationship
functioning in an inpatient setting (Urist, 1977).

33

There is no shortage of validated measures of object relations that have
been developed in recent years. Studies of borderline object relations on the
Rorschach have shown that borderlines tend to have distorted, malevolent
representations characterized by boundary problems.

In summary, the increase in Rorschach research on borderlines has led
to the conclusion that borderlines as a group are distinguishable from other
diagnostic groups on a number of personality variables. There is established
validity for a number of measures of reality testing, defensive functioning,
ego strength, and object relations, which have largely supported the
theoretical predictions about borderline intrapsychic functioning.

The Rorschach as a Measure of Personalig Change

Although the Rorschach has become a more popular research
instrument, only a few studies have used it to assess longitudinal change in
intrapsychic functioning, and none to date have used the Rorschach to assess
changes in reliably diagnosed borderlines. Exner (1986b) has produced test-
retest data from several studies using the Comprehensive System Rorschach
variables in various clinical and non-clinical populations across different
time intervals: 7 days, 30 days, 60 days, 90 days, 180 days, and 35-38 months.
He found that most Rorschach variables showed high test-retest correlations,
reflecting the general stability of personality across relatively short time
periods. He also found test-retest correlations over a three-year period to be
higher in non-patients than inpatient populations. This finding is in line
with the prediction that non-patients should show more solidified
personality structure than patients "subject to psychopathological
disorganizations or the influences of treatment" (p. 65). He found that two
variables related to situational stress (m and sum of shading) showed the

34

lowest retest correlations, lending validity to the idea that they are more
situational and less predictable, while variables reflecting characteristic styles
and structure (Affective ratio, EB, Z frequency, Egocentricity Index, and X+%)
showed the highest retest correlations. Exner (1986b) summarized: "The
matrix of these seven studies offers substantial support for the contention
that the majority of Rorschach variables underpinning interpretation are
stable over time" (p. 79).

To date, of the studies using the Rorschach as a measure of change only
one has focused on borderline patients exclusively, and all are primarily post-
treatrnent studies aimed at assessing the degree of change brought about by
treatment. However, I will briefly review these studies because of their
application to the methodology of the current study.

Frieswyk and Colson (1980) used the Rorschach to predict longitudinal
changes in 28 borderline inpatients, who were broadly diagnosed according to
clinical judgment. Patients were assessed at index hospitalization and at
discharge (approximately one year later), and again at a two year follow-up.
The authors developed their own Rorschach scale to measure the following
variables: object depreciation, oral rage, malevolence, fear of interpersonal
contact, fluidity, lack of boundaries, and object constancy. They also used the
Empathy—Object relations scale developed by Pruitt and Spilka that assesses
the degree of realism versus fantasy in human representations, and Krohn
and Mayman‘s scale for assessing level of object relations in dream imagery
applied to the Rorschach (cited in Frieswyk 8t Colson, 1980). Clinical ratings
were used as measures of outcome at post-treatment termination and follow-
up. No attempt was made to determine DSM-III or DIB diagnosis at either

termination or follow-up. Rorschach variables were only assessed at time 1,

35

so we cannot know what changes, if any, would occur in these particular
variables.

Results indicated that the index Rorschach variables that were positive
predictors of termination outcome were: the number of human responses,
human movement responses, and object relations; however, oral receptivity
and oral aggression were negative predictors of termination outcome.
Surprisingly, the only Rorschach variable that was a predictor of positive
outcome at follow-up 2 years post-hospitalization was expressed malevolence
on the Rorschach at initial testing. Regarding this paradoxical finding the
authors suggest: "...those who produce even a few percepts characterized by
intense malevolence may be those in whom there is a greater propensity for
fighting and struggling rather than passive resignation" (p. 253). However, it
is unclear why indices of malevolence or aggression would predict outcome
differently at termination and follow-up assessment.

Another post-treatment study by Weiner and Exner (1991) assessed the
Rorschach changes in long- and short-term outpatient psychotherapy patients
at several intervals: initial testing, 12 months, 27-30 months, and 46-50
months. The patients in this study were diagnostically heterogeneous, and
were not rated on any measures of functioning other than the Rorschach.
Rorschachs were scored according to 27 indices of adjustment using the Exner
Comprehensive scoring system. The authors concluded that both short-term
and long-term patients improved somewhat on most of the Rorschach
variables, although the long-term group improved the most. In general, both
groups were found to: (a) manage stress better (decreased D<0, Ade<O, EA<7,
and CDI>3); (b) deal more effectively with experience (decreased Ambitence,
Zd<-3.0, Lambda>.99, X+%<70, X-%>20); (c) modulate emotional experience

36

better (decreased SumSh>FM + m, DEPI = 5, Afr<.50, and CF+C > FC+1); (d)
become happier and less negatively preoccupied with themselves (decreased
3r + (2)/ R >.43, and <33); and (e) improve in their comfort with interpersonal
relationships (less p>a+1, T=0, T>1, Pure H<2, and H<[(H)+Hd+(Hd)]).

In another post-treatment study, Blatt and Ford (1994) assessed 90
general psychiatric inpatients using the Rorschach along with other
projective and non-projective measures to compare changes in functioning
over a 10 - 12 month hospitalization. Although their sample included
borderline patients, analyses of the data were not broken down by diagnostic
category. Overall they found few changes in the Rorschach over the course of
the year following psychiatric admission. Of the Rorschach variables
measured that changed significantly, several indices of thought disorder were
found to decrease, while a non-significant trend (p < .10) was found for a -
decrease in malevolence. No changes were found in Blatt, D'Affliti, and
Quinlan‘s (1976) Object Relations Scale or the Mutuality of Autonomy scale
(Urist, 1977) in the following variables: developmental index, the
developmental level of the concept of the object, or mutuality of autonomy.
However, due to the large variability in diagnostic status of the patients in
this study, it is not surprising that consistent pre-and post-treatment changes
did not emerge. Paradoxically, increased thought disorder and expressed
malevolence at admission were prognostic of better functioning at
termination assessment. The unusual finding that malevolence is a positive
treatment predictor is along the lines of the follow-up finding of Frieswyk
and Colson (1980).

Although these three studies demonstrate the usefulness of the

Rorschach in assessing personality changes, and provide valuable insights

37

into some Rorschach predictors of post-treatment functioning, none were
designed to address the question of the stability of underlying borderline
dynamics. First, none of the samples were comprised of reliably diagnosed
borderlines, and none of the studies attempted to determine BPD diagnosis at
termination or follow-up. Second, because all of the studies included only
patients voluntarily engaged in ongoing, controlled treatment, they may not
reflect the naturalistic changes that occur in borderlines, who are known for
their treatment instability. Nevertheless, these studies provide interesting
clues to the predictors of outcome.
Summgry and Hypotheses

To summarize, several issues are raised by the research on borderline
stability. The first regards the surprising number of DIB and DSM-III
borderlines who are no longer borderline at follow-up. The fact that a
proportion of borderlines no longer meet BPD diagnostic criteria in short-
term follow-up studies raises diagnostic and conceptual questions related to
the borderline construct: (a) Are the components of personality organization
more or less stable over time? (b) What is the relationship between
syndrome-based diagnosis of BPD and indicators of borderline personality
organization? Although Kernberg's theoretical conception of borderline
personality organization predicts stability in the underlying dynamics,
findings raised by the follow-up literature suggest that some borderlines may
undergo significant personality changes over time, and these changes may be
related to age. A single longitudinal study by Kullgren and Armelius (1990),
addressing the stability of personality organization, found that borderline
personality organization may change over time. Although this study must be

viewed critically for methodological reasons, the findings are nevertheless

38

worth replicating. We predict that improvement in the components of
personality organization should correspond to changes in BPD diagnosis
between index and follow-up. Furthermore, we expect more improvement
in the symptoms of patients who were older at follow-up assessment.

Because the use of Kernberg's structural interview only yields global
information about personality structure, the Rorschach was chosen to
operationalize the components of personality organization. Four of the five
indices of borderline personality organization were selected for the current
study: defensive operations, reality testing, ego-weakness, and object
relations. These four were chosen based on the theoretical centrality of the
personality components, and their demonstrated usefulness and scoring
validity in previous Rorschach studies. Identity diffusion, although
theoretically important, was excluded due to the lack of studies establishing
validity and the low base rates of the Rorschach variables believed to measure
this construct.

Finally, an exploratory question is raised regarding the validity of
Kernberg's construct of BPO. Previous studies of personality organization
have used Kernberg's structural interview, which only yields global
information about the level of personality organization. It is important to
know whether the four components of personality organization measured in
this study (Defenses, Reality Testing, Ego Weakness, and Object Relations)
cluster according to the pattern predicted by Kernberg (1975) and Acklin (1993)
for patients with borderline personality organization. Murray (1993) cautions
against a simplistic approach that assumes that borderlines will show
consistent indicators of BPO, stating "it is unrealistic to believe that such a

diverse group of patients would present with any specific or particular type of

39

Rorschach pattern - despite their having a common level of personality
organization" (p.343). However, this notion has not been empirically tested
to date.

The following hypotheses will be tested in the present study:

1. Changes in overall indices of Borderline Personality Organization
will be positively correlated with change in the degree of BPD symptoms
between time 1 and follow-up.

2. Changes in specific components of Borderline Personality
Organization (Defenses, Ego Weakness, Reality Testing, and Object Relations)
will correspond to change in BPD symptoms between time 1 and follow-up.

3. In line with the follow-up findings of Stone (1990b) and others, age
at follow-up should correlate positively with symptom change between index
and follow-up.

Ex lorato estion.

1. Do we find distinct configurations of Defenses, Reality Testing, Ego
Capacity, and Object Relations on the Rorschach in line with the predictions
of Kernberg's construct of Borderline Personality Organization in patients
who are diagnosed with fairly specific criteria for BPD? All time 1 subjects,
who carry a diagnosis of BPD, will be assessed for the following pattern of
borderline personality organization: Defenses: poor, Reality Testing:
moderate to good, Ego Capacity: poor, Object Relations: poor.

Overview of the current study
The current follow-up study aims to clarify the relationship between

symptom-based borderline diagnosis and Rorschach indicators of borderline
personality structure over time. Patients in this study were assessed at two
time periods: index hospitalization and follow-up approximately 2-5 years
later. At index, all patients received a diagnosis of Borderline Personality
Disorder (BPD) by the Diagnostic Interview for Borderlines. Using the
Rorschach allowed more certainty in the assessment of personality
organization than Kernberg's structural interview, in that the situation is
better standardized, and the borderline indicators can be quantified and
subjected to reliability tests. Furthermore, the Rorschach allows us access to a
wealth of information, both global and specific, about intrapsychic
functioning. Measuring the components of personality organization
separately makes it possible to test them individually as well as in
summation. Unlike previous Rorschach studies which have focused
primarily on one personality dimension, this study looks at several
personality dimensions at the same time, and also tests the presumption that
borderlines will show a distinct pattern across Rorschach variables.
Furthermore, whereas most Rorschach studies collapse scores across subjects
and compare means of groups, the present methodology looks at ipsative
change across time to preserve the integrity of the individual protocols as

much as possible.

Method

Index study
Subjects.

Subjects considered for the initial phase of the study were inpatients on
two psychiatric units at the University of Michigan Medical Center. Criteria
for exclusion were chronic psychosis or a medical condition that would
preclude a two-week medication free period while in the hospital. Patients
who were identified by senior clinicians as meeting two or more DSM-III
criteria for either Borderline Personality Disorder or Schizotypal Personality
Disorder were administered the Diagnostic Interview for Borderlines (DIB;
Gunderson, Kolb, 8: Austin, 1981). Subjects who scored 7 or higher on the
DIB were considered borderline and included in the current study. All
subjects were medication-free at the time of assessment].

Measures.

The Diagnostic Interview for Borderlines (DIB; Gunderson, Kolb, 8:
Austin, 1981) is a 50-90 minute semi-structured diagnostic interview that is
designed to gather information about 5 areas of borderline functioning:
Social Adaptation, Impulse Action Patterns, Affects, Psychosis, and
Interpersonal Relations. Twenty-nine statements overall receive a score of 0,
1, or 2; a score of 2 indicates presence of the item, 0 indicates absence, and 1 is
scored for ambiguous items. Items for each subsection are totaled, and
subsection totals are converted into scores of 0,1, or 2 depending on the
number of items endorsed in that subsection. The five resulting subsections
are totaled to yield an overall score 0-10. Scores of 7 and above are considered

borderline (Gunderson, Kolb, & Austin, 1981). In addition to symptoms and

41

 

42

behaviors, the DIB also gathers information about historical events, for
example, incest and previous psychiatric hospitalizations, as well as
subjective information, such as frequent disappointments with significant
others. (See Appendix B for a summary of the DIB diagnostic criteria).
Because all but three items are based on recent experiences and events (past 24
to 36 months), the DIB can be used as a measure of change for intervals of two
years or longer.

The authors of the DIB report adequate interrater reliability, with total
DIB scores showing greater reliability than individual section scores.
Interrater reliability for the presence or absence of BPD according to the DIB
was previously established by the researchers involved in the current study as
acceptable, K = .8 (Cornell, Silk, Ludolph, 8r Lohr, 1983). Validity studies have
shown that the DIB distinguishes borderline personality disorder from
schizophrenia and other personality disorders (Barrash, Kroll, Carey, 8: Sines,
1983; Frances, Clarkin, Gilmore, Hurt, 8r Brown, 1984) and correlates
significantly with DSM-III diagnoses. However, the DIB and DSM III
borderline constructs are slightly different, with the DIB appearing to be more
inclusive than DSM-1H.

The Rorschach Inkblot Method was used to operationalize the
personality variables derived from Kernberg's (1985) criteria for structural
diagnosis, and Acklin‘s (1993) discussion of the application of Rorschach
variables to Kernberg's diagnostic criteria. The major categories chosen to
define the particular aspects of borderline functioning were Defenses, Reality
Testing, Ego Capacity, and Object Relations. Each of these four sub-categories
of the structural diagnosis was separately considered, and Rorschach variables

43

that best operationalized each one and showed reasonable validity were
applied to form a separate scale for each.

Three primary coding systems for the Rorschach were used in the
current study. Exner's (1986b) Comprehensive System was used to code
Rorschach responses, and structural summaries were calculated for each
protocol. Exner's Comprehensive System is currently considered the most
researched and best validated coding system for the Rorschach with good
interrater reliability (Weiner, 1997).

Object relations was assessed with the Rorschach Human Experience
Variable (HEV) devised by Perry and Viglione (1991). The HEV is one
component of the Ego Impairment Index (EII; Perry & Viglione, 1991), a more
inclusive measure of functioning as measured on the Rorschach. The HEV
has been shown to account for the majority of the variance on the EII (Perry 8:
Viglione, 1991), and has since been validated separately. The HEV coding
scheme, like other Rorschach scales of object relations (e.g. Blatt et al., 1976;
Urist, 1977), is based on the theory that the quality of internal object relations
are expressed in the quality of Rorschach human percepts. In this scheme
good object relations are characterized by whole humans, lack of distortions,
benign affect, and cooperative interpersonal representations, while poor
object relations are characterized by fictionalized and partial humans,
distorted percepts, and malevolent, destructive interpersonal representations.
The HEV uses Exner's Comprehensive System scores to operationalize a set
of decision criteria which are used to divide human responses into "good"
and "poor" humans, based on the features of the human percept. After each
human percept is categorized as either "good" or "poor", the number of each

is entered into a a fixed decision algorithm to arrive at a z score that

44

corresponds to the ratio of good to poor human percepts. For the HEV,
negative 2 scores indicate a better "good" to "poor" human percept ratio,
while positive 2 scores indicate more "poor" humans. The HEV was chosen
because, unlike other Rorschach scales of object relations, it integrates several
aspects of object relations into a single dimensional score. The HEV has been
shown in research to be a strong predictor of poor treatment response (Perry
6: Viglione, 1991), has shown temporal stability (Perry, McDougall, 8r
Viglione, 1995) and has successfully differentiated between groups rated to
have high and low quality of interpersonal relationships (Burns 8: Viglione,
1996).

Another Rorschach coding system, The Rorschach Defense Scales
(Cooper, Perry, & Arnow, 1988) was used for operationalizing the subsection
Defenses. This measure uses all responses on the Rorschach to code for 15
defenses that span a range of defensive functioning, from developmentally
lower-level to higher-level defenses. RDS scoring is based on the content of
the responses (including random verbalizations about the testing) rather than
the formal aspects of the percepts. Each defense has between 6 and 14
exemplars which qualify for a coding of that particular defense. The authors
divide the defenses into three developmental categories: Psychotic (massive
denial and hypomanic denial), Borderline (splitting, primitive idealization,
devaluation, omnipotence, projective identification, and projection), and
Neurotic (denial, intellectualization, isolation, reaction formation,
repression, rationalization, and pollyannish denial). A summary of the types
of Rorschach responses coded for the various defenses is included in

Appendix C.

45

Preliminary validity has been established (Lerner, Albert, and Walsh,
1987), with the type of defenses significantly discriminating between
outpatients, DSM-III diagnosed inpatient borderlines, and DSM-III diagnosed
schizophrenics. Furthermore, another RDS study by Cooper, Perry and
Arnow (1988) found borderline defenses to be significantly correlated with a
scale of borderline diagnostic criteria, but not scales of bipolar or antisocial
diagnostic criteria.

RDS scores were determined by tallying the number of occurrences of
each of the 15 defenses per protocol. For the purposes of the current study,
the sum of the defenses in each category was calculated and weighted
according to the pathology of the defense (Neurotic = 1, Borderline = 2, and
Psychotic = 3). Next, in order to control for overall productivity, the weighted
sums were divided by the total number of defenses that were scored to yield
an overall weighted average. This weighted average was used to indicate the
overall level of defensive functioning, with higher scores pointing to more
pathological defenses.

Reality Testing was measured using Exner's variable Weighted Sum of
Special Scores (WSum6; Exner, 1986b). This variable is a weighted sum of the
special scores that indicate odd verbalizations, cognitive slippage, peculiar
logic, and unrealistic or bizarre combinations, and weights are assigned
depending on the severity of the thought disturbance. WSum6 is
dimensional, with higher scores indicating more thought disorder, and it has
been shown to distinguish schizophrenics from borderlines and other
character disordered groups (Exner, 1986a).This variable was chosen
specifically because it operationalizes the types of thought disorder which

46

have been described historically in the early Rorschach literature on
borderlines. According to Kernberg (1985):

...on projective testing, and especially in response to unstructured
stimuli, primary process thinking tends to appear in the form of
primitive fantasies, in a decrease in the capacity to adapt to the formal
givens of the test material, and particularly in the use of peculiar
verbalizations" (p. 24).

Another Exner measure of thought disorder, the Schizophrenia Index (SCZI),
was not chosen because it was developed specifically for detecting thought
disorder in schizophrenics, is much more restricted in its range than the
WSum6, and has been shown to have poor specificity for schizophrenia in
clinical populations (Exner, 1986b).

Ego Capacity was operationalized for the current study with several
variables from Exner's Comprehensive System. Although an index of poor
coping ability, the Coping Deficit Index (CD1) has been developed by Exner,
research on the CDI indicates that it may have very poor sensitivity for
detecting borderline coping deficits (Carlson, Kula, 6: St. Laurent, 1997).
Furthermore, the CDL despite its name, was also originally developed as a
depression scale, and the items measure affective constriction, passivity, and
dependency. Although these characteristics may indicate poor ego
functioning, the deficits in ego functioning typical of borderlines are
somewhat different. Ego capacity was conceptualized for the purpose of the
current study as those aspects of functioning that would allow one to regulate
impulses and emotions, delay action in favor of thought, and resist
decompensation under stress. Six Rorschach conditions that operationalized
poor ego capacity (poor emotional control, impulsivity, and poor stress

47

tolerance) were chosen in consultation with a senior clinical psychologist
who teaches Rorschach interpretation at the graduate level (see Appendix D
for all scoring criteria). Five of the six conditions chosen were single variables
or combinations of variables for which norms were available (e.g., FC: CF+C,
C, D, Lambda, and WSumC: M). Exner's (1990) normative sample of
nonpatients was used to determine the percentage of normal individuals
who met each of the criteria, and in each case only a small minority of non-
patients met each one. The Ego Capacity score was calculated as the sum of
the criteria met. Based on the non-patient norms, it was decided that the sum
of the 5 ego capacity scores for most non-patients would be 0 or 1, that 2 to 3
indicated moderate difficulty, and scores higher than 4 suggested extremely
poor ego functioning.

Procedure.

All time 1 potential inpatient research subjects were approached for
participation in the study by a research nurse within several days of
admission. Those who agreed to participate were given a verbal explanation
of the study and signed a consent form indicating their voluntary
participation. First, the DIB was administered by an advanced clinician who
was also a member of the research group, who determined whether the
patient met criteria for a borderline diagnosis (DIB score 2 7). Interrater
reliability for the diagnosis of BPD according to the DIB was previously
established by the group as acceptable, _I_(_ = .8 (Cornell, Silk, Ludolph, 8r Lohr,
1983).

Subjects in the study were administered a battery of psychological tests
by an experienced graduate student or PhD. level clinician, not necessarily
blind to diagnosis. These tests included the Wechsler Adult Intelligence

48

Scale-Revised (WAIS-R), the Thematic Apperception Test (TAT), Mayman's
Early Memories Test, and the Rorschach. The Rorschach was administered
according to the Rapaport-Schafer instruction, with all 10 cards presented first
followed by the inquiries. The verbatim instructions for the Rorschach given
in all testings were: "What does this look like?". Instructions were not
repeated once the subject clearly understood the task.

All subjects in the study also completed a battery of self-report
measures not used in the current study. Due to the ongoing data collection

for this study, some subjects did not complete all measures.

Follow-up Study

Slime—eta
For the follow-up phase of the study, a subset of the original pool of

subjects diagnosed as BPD on the DIB were reassessed between 1988 and 1994.
Follow-up subjects included only those from the original sample who had
been hospitalized at index less than 10 years previously. Although
researchers were aware that the study was on patients with BPD, several non-
BPD subjects from the original research pool were included in the follow-up
sample to limit the researcher's assumptions about diagnosis during the
follow-up phase.

Of the 55 eligible BPD subjects from the original pool, 31 were
successfully contacted and reassessed at follow-up. Four of the original 55
patients contacted for the follow-up study were excluded because they were
found to have Major Depression at follow-up, and the investigators were
interested only in non-depressed borderlines in the early stages of the project.
Four subjects who were contacted declined to participate in the study, and 2
agreed to participate but had moved out of state and could not travel in time

49

for the scheduled testing days. Two subjects had died during the follow-up
period, and the remaining 12 subjects could not be located. Of the 31
successfully recontacted, 28 had Rorschachs done at time 1.

Measures.

Follow-up subjects were again administered the same interviews,
measures, and psychological testing that they completed at time 1, including
the Rorschach.

Procedure.

All potential follow-up subjects were recontacted by letter, which
explained that they would be contacted by phone regarding the follow-up
study and offered them $50 to participate. Subjects were then contacted by
phone within 2 weeks by a research assistant who explained the study in
more detail and scheduled the follow-up testings. Subjects who consented to
the study were sent the packet of self-report measures, and were asked to
bring the completed measures to the medical center on the day of follow-up
testing.

Follow-up subjects came to the University of Michigan Medical Center
for testing, and were administered the DIB by a senior clinician. Subjects were
given a 60-90 minute sernistructured interview by another interviewer who
was blind to Time 2 DIB diagnosis. This interviewer gathered information
about the follow—up period, including demographic information, type and
length of treatment sought during the follow-up interval, suicide attempts,
prescription and non-prescription drug use, and social relationships. During
this interview, the clinician also scored the subject on several measures not

used in the current study. Other Ph.D. or graduate level clinicians, blind to

50

DIB and RDC diagnoses, administered the Rorschach and other psychological
tests. Follow-up subjects also completed the same set of self-report measures
they completed at index.

Rorschach Scoring
Six graduate students in clinical psychology previously trained in

Exner's Comprehensive System (1986) scored all protocols. Coders were all
blind to the purpose of the study and the diagnoses of the sample.
Preliminary interrater reliability was first established by having all coders
score 20 sample responses given by Exner in his Rorschach Workbook for the
Comprehensive System, Third Edition (Exner, 1990). Reliability estimates
were calculated by computing kappa coefficiants between each coder's set of
scores and the correct scores given by Exner in Appendix A of the Workbook
for the Comprehensive System (Exner, 1990). Kappas were calculated
separately for each of the following categories of scores: Location,
Developmental Quality, Form Quality, Pair, Populars. Reliability estimates
were calculated using percent agreement for the categories Determinants,
Contents, and Special Scores because of the frequent asymmetries in the
number of categories scored, because many responses receive multiple scores,
and there is a greater number of possible scores, and because there is a greater
degree of scoring difficulty. For Determinants, percent agreement was
calculated for both exact agreement and "partial" agreement between the
coder's determinants and Exner's answers. Partial agreement for
determinants was calculated to allow for some variation in the determinants
as long as the difference would only minimally affect the overall structural

summary. A determinant was considered partially correct under three

51

circumstances: the determinants in a blend were identical but the order of
determinants was different, some part of a blend was exactly correct but
another was slightly different (e.g. "ma.FV" when the correct answer was
"mp.FV"), or determinants were very similar ("PC when the correct answer
was "CF"). Exact agreement for Determinants ranged from 40% to 90%, while
"partial" agreement ranged from 90% to 100% for all coders.

Reliability estimates for Contents and Special Scores were also
calculated using percent agreement. For both of these categories, reliabilities
were calculated by dividing the number of scores that agreed exactly with the
correct answer, divided by the highest number of scores given. Exact
agreement for Contents ranged from 57.6% to 92.9%, while exact agreement
for Special Scores ranged from 81.8% to 95.5%. Reliability coefficients for
Exner variables are reported in Table 1.

Despite generally good reliability estimates between coders and Exner
scores on practice items, it was decided that for the study each protocol would
be scored independently by two coders and then coders would be asked to
reach consensus in their final scoring. Consensus scoring was done to reduce
potential variability and errors in scoring, and to arrive a single set of scores
for each protocol. Furthermore, coders were assigned to pairs on a rotating
basis to prevent coder drift. After all scoring was completed, Exner protocols
were entered into Exner's computer scoring program which calculated the
structural summaries.

For the Rorschach Defense Scale four coders, two undergraduate
psychology majors and two graduate students in clinical psychology were
specifically trained in the coding system. Coders were trained in the scoring

rules and then scored several practice protocols together and discussed their

 

52

338m .mmummm use .mEmucoU damp—«8.860 cow 633mm.“ mm

 

 

 

 

 

 

EoEmmuwm “sushi $03 1:208 258— PB segmen— vcm Ema .Om .00 A8380..— .How Vetoes." 356$on .302
mm. R. ma. mo. 3. so. am. we. 84 3 <

mm. mm. mm. ow. 3. mo. mm. 3. 8A mM

3. mo. 2: mm. cog we. ow. em. 2: EU

ma. co. mm. mm. 84 mo. mm. «3. 3w. Om

ow. am. 2: 8. B. as. am. mm. no. a

as. aw. om. cm. 84 3. mm. 8. 84 E

museum Euomm 3:855 Rwan— Eumxm an mn— Om OD vor— ~83—

muclmEEcBmD

 

 

H 2an

m2flr€~> Baxm new
8.8% Mcmcxm 98 Essex 53253 $56580

 

253mm 8%.325

53

answers. Most discrepancies were resolved through ongoing discussions, and
those practice scoring items that continued to be unclear were resolved by
contacting the author of the scoring systems directly (S. Cooper, personal
communication, May 1996). Interrater reliability was initially established
between all four scorers on 13 sample responses. Reliability estimates were
calculated as the percent agreement between the defense scoring for each of
the 6 pairs of coders, based on whether the score for each coded response fell
into the same general Neurotic, Borderline, or Psychotic defense category.
Percent agreement ranged from 50% to 83%, which was comparable to the
interrater reliability reported by the authors (Cooper, Perry, & Arnow, 1988).
As with the Exner scoring system, it was decided to have two coders score
each protocol and reach consensus scoring for the RDS to reduce scoring
errors. Furthermore, as with the Exner coding, pairs were rotated to avoid
coder drift.

In order to run analyses in which the 4 Rorschach variables (Defenses,
Reality Testing, Ego Capacity, and Object Relations) were combined to yield an
overall score, each was recoded so that they were similarly scaled. Rorschach
variables in each area were recoded as 1, 2, or 3 depending on the severity of
the score, with 1 coded for the neurotic (healthy) range, 2 coded for the
borderline (moderately severe) range, and 3 coded for the psychotic (most
severe) range.

The determination of cutoffs for each of the four variables was made
separately to take into account empirical norms available for each scale as
well as theoretical considerations. For Primitive Defenses, the categorization
of defenses as neurotic, borderline, or psychotic were predetermined by the

authors of the RDS. Cutoffs for the defense scores, described above, were

54

made by deciding on appropriate ratios of neurotic, borderline and psychotic
defenses (a score of 1 meaning all neurotic defenses and 3 meaning only
psychotic defenses). Because norms for non-patient, borderline, and psychotic
patients were not available, it was decided that conservative cutoffs would be
used. A defense score of 1.33 marked the upper cutoff for the neurotic range,
indicating mostly neurotic defenses with a ratio of two neurotic defenses to
each borderline defense. A defense score greater than 2, indicating the
defenses were mostly borderline and psychotic, was the cutoff applied for the
psychotic range. Scores between 1 and 1.33 were determined to fall in the
neurotic range, scores between 1.34 and 2.0 were determined to be in the
borderline range, and scores greater than 2.0 were determined to be in the
psychotic range. Scores were then recoded 1, 2, or 3 corresponding to the
appropriate range.

Cutoffs for the Object Relations HEV scores were partially determined
using the z scores and norms provided by Perry and Viglione (1991) and
Burns and Viglione (1996) for inpatients and non-patients with different
levels of functioning, respectively. However, because there were not specific
norms available for borderline and psychotic patients, cutoffs were also
partially determined by calculating the z-scores that corresponded to certain
ratios of the I-IEV's "good" and "poor" humans. The author's original
formula, devised on a sample of depressed inpatients (Perry 6: Viglione,
1991), yielded 2 scores in which negative z-scores indicate more "good"
humans and positive 2 scores indicate more "poor" humans. However, in
this original study a z score of 0 was based on a ratio of 2.6 "good" to 3.8
"poor" humans; thus, the original distribution is slightly skewed in that 0
corresponds to more ""poor human percepts. As reported by Burns and

55

Viglione (1996), high interpersonally related non-patients were found to have
a mean HEV of -1.7 and standard deviation of 2.0, and on average had a ratio
of 2 "good" humans for each "poor". Taking the skewed nature of the
distribution into account, it was decided that ratios of "good" to "poor"
humans would be used to decide the z score cutoffs rather than using strict
statistical guidelines based on percentiles. For the purpose of determining
cutoffs, a predominance of "good" humans was considered to be a sign of
neurotic level functioning, a mixture of "good" and "poor" humans signalled
the borderline level of functioning, and a predominance of "poor" humans
was considered to be a sign of severe, psychotic level functioning. The
neurotic cutoff was set at -.70, which corresponded to a ratio of 1 "good" to 0
"poor" humans, and fell at the 70th percentile of the healthy non-patient
sample presented by Burns and Viglione (1996). The severe/ psychotic cutoff
was set at .54, which corresponded to a ratio of 1 "poor" to 0 "good" human
percepts, and fell at the 87th percentile of the healthy sample . Scores below
-.70 (indicating a higher ratio of "good" humans) were categorized in the
neurotic range, scores between -.70 and .54 were categorized in the borderline
range, and scores higher than .54 (indicating a higher ratio of "poor" humans)
were categorized in the severe / psychotic range.

For the Rorschach variable Reality Testing, norms for WSum6 for non-
patients and schizophrenic patients are given by Exner in his Rorschach
Workbook for the Comprehensive System, Third Edition (1990), and means
and standard deviations for a sample of DSM-III diagnosed borderlines are
reported in another study by Exner (1986a). The data from these sources were
used to determine cutoffs. Exner (1986b) has given general clinical guidelines
about the ranges for WSum6, suggesting that scores between 6 and 9 indicate

 

56

likely thought disturbance, and scores greater than 9 indicate severe thought
disturbance. Nevertheless, since thought disorder is typical of the Rorschachs
of borderlines, the cutoff for the upper end of the borderline range was set
higher to reflect this. In Exner's (1990) nonpatient population norms, the
scores for WSum6 ranged from 0 to 15, and had a mean and standard
deviation of 3.21 i 2.73. DSM-III diagnosed borderline patients had a mean
WSum6 score of 7.57 i 5.96, and DSM-III diagnosed schizophrenic patients
had a mean of 19.21 i 9.81. The borderline mean of 7.57 was found to be at
the 91 st percentile of the nonpatient distribution, and the schizophrenic
mean was higher than any score in the nonpatient distribution. For the
purposes of the current study 12 was chosen as a cutoff for the upper limit of
the borderline range, since 12 fell approximately one standard deviation
above the mean for the borderline reference group and fell below the mean
for the reference schizophrenic group. Thus, the WSum6 scores of 0 to 5 were
categorized as neurotic (consistent with Exner's guidelines), scores of 6 to 12
were categorized as borderline, and scores higher than 12 were categorized as
psychotic.

Ego Weakness cutoffs were determined using Exner's normative data
for nonpatients and schizophrenics reported in the Third Edition
Comprehensive System Workbook (Exner, 1990), and data for DSM-III
diagnosed borderlines was reported in another study by Exner (1986a).
Because this scale was constructed for the purpose of the current study, norms
for the cumulative scale were not available. Instead, data for 5 of the 6
separate criteria on the Ego Weakness scale were used to estimate the
percentage of nonpatients, borderlines, and schizophrenics meeting each one.

 

57

The resulting scores for the four subscales (Defenses, Object Relations,
Reality Testing, and Ego Weakness) were then summed, yielding a summary
borderline Rorschach score between 4 and 12. Higher values on this
summary score indicate more pathological overall functioning as measured
by the Rorschach. Next, the difference between the time 1 and follow-up
Rorschach summary scores were calculated, and each subject was categorized
as either "improved", or "no different", or "worse". Likewise, index and
follow-up DIB scores were compared for each subject, and each subject was

categorized as either (a) "improved" or (b) "not improved" on the DIB.

Results
Twenty-eight subjects were successfully reassessed at follow-up.
Demographic and Follow-up information is presented in Table 2. Length of
interval between initial assessment and follow-up ranged from 16 months to
78 months, with an average of 38 months. The follow-up sample was
primarily female (n = 24, 86%), Caucasian (n = 26, 93%), and unmarried ~'
(54%). The average age of the follow-up sample was 33.9, the median age was 1

33.3, and the range was 21.6 to 53.3. At follow-up, 11 (39%) retained the BPD

 

diagnosis by the DIB, and 17 (61%) had improved and no longer met the DIB
criteria for BPD. i
Analyses of differences between the follow-up sample and the patients
who were not successfully recontacted were performed to see if the dropouts
were significantly different in age and sex. Data on these variables was only
available for 15 of the dropouts; nevertheless, the dropout sample was
significantly younger (M = 25.3, t = 2.22, df = 35 , p <.05), and had significantly
more males (60%, X2 = 3.93, df = 1, p <.05) than the follow-up sample.
Of the 28 follow-up subjects, 3 had Rorschachs with either missing data
or missing location charts. Of the remaining 25 subjects, 3 subjects had one or
both protocols that had fewer than 12 responses and were considered invalid
for the purpose of the current study (Exner, 1986b). Only the 22 remaining
subjects who had valid protocols (R 2 12 and Lambda < 1.2 and > .32 ) at both
time 1 and follow-up were included in the statistical analyses for Rorschach
variables. The average number of responses for the Rorschach protocols at
time 1 and follow-up were 22 :1; 9.5, and 22 _+_ 7.0, respectively. Means and

58

59

Table 2

Subject and Follow-up Variables (N = 28)

 

Age at Age at Follow-Up
Time 1 Follow-up Interval (months)
Mean 30.4 33.9 38.1
SD 8.3 8.3 13.8
Min 19 21 16
Max 50 53 78
5g: Race:
Female: p = 24 (86%) Caucasian p = 26 (93%)
Male: 11 = 4 (14%) African-American n = 1 (3.5%)
Asian a = 1 (35%)
Marital Status (Time 1):
Single = 54%
Divorced = 21%
Married = 14%
Separated = 4%
Widowed = 7%
Inpatient Treatment During Follow-up Interval
Percent Psychiatrically Hospitalized: 57%
Number of Hospitalizationsa: M = 2.3
Percentage of interval hospitalizedb: _M_ = 6.6%
Percent in one or more type of
outpatient psychotherapy: 100%
Percent of interval in some type
of outpatient psychotherapy: 100%
Percent taking psychiatric medication
some time during follow-up interval: 89%

 

Npte. a,b Calculated using only subjects who were hospitalized during

follow-up.

 

"Hut “‘2'!“ ' f"

60

standard deviations of the DIB and Rorschach variables are presented in Table
3.

The first hypothesis stated that changes in the overall index of
borderline personality organization would correspond to changes in the DIB.
This was tested by calculating a Pearson product-moment correlation
coefficient between the difference scores for the summary Rorschach scale and
the DIB. Results of this analysis are included in Table 4. This test yielded a
statistically non-significant correlation (r = .26, p = .24) ; thus, the hypothesis
was not supported.

Nevertheless, inspection of the data revealed that the method for
computing the overall Rorschach score had resulted in restricted variance,
which may have contributed to the lack of significant findings. Therefore, a
post-hoc method for computing the overall score was introduced, with the
expectation that the new method would result in greater variance. The
component Rorschach scores (Defenses, Object Relations, Reality Testing, and
Ego Capacity) were transformed into standard (2) scores, and these standard
scores were summed to produce the overall Rorschach scores. Next, the
difference score was obtained between the time 1 and follow-up overall
scores. This new overall Rorschach difference score was then correlated with
the DIB difference score, using a Pearson correlation. Although the
correlation between the DIB and the new overall Rorschach scores was
stronger (; = .37, p > .07), the trend was non-significant.

The second hypothesis, that changes in the specific Rorschach indices
will correspond to changes in the overall BPD diagnosis, was tested by
computing Pearson correlations between the individual Rorschach scales and
the DIB score. Results of the correlations are reported in Table 4. The

 

61

Swo—ofimmosuzmm
.88me 38:65 .82.; 8.32 can wficonuca .833 38:35 Big grams: use .832 35 em nous—=23
«.8 230,—: was bvmmmu 0mm $553: bzmum .mcnfifimm “66.50 @3539 8338‘, sumsuflom as; .26 Z

 

 

HN + max «A H de «.mm H wcém N6. + S. vm. H no.“ “gauge—Em

w; + MNN m4 H we; Q2 H 8.: wd + 3. 9. H mné ”H 95B

 

 

:33. Ba «U o m Emma. 28% 9832mm away—NO momaowom

smug
gang/549m

 

 

R. u Rs R. H 9: cm. H use a. H Rd 3. H e: as a men ”assess
om.Hom.~ 93.1254 inflow; equmA mv. Hwné :AHoow ”H 08E.
823er massing Erma «.2 =36:
Gasmaogfifi 383‘ mmmonuzmn— 8395 360m 988 ED
am u 7|:
Baum 88% egg

 

3% so a assets sausages.

m 2an

62

Table 4

Pearson Correlation Coefficients Between Difference Scores
for DIB and Rorschach Variables

 

(IN. = 22)
Object Reality Ego
Defepg Belatigns Jeann' g Qapgg'ty [93a]
DIB Total .64""’* .23 .14 .08 .26
DIB Impulse .05 .07 .01 .28 .15
DIB Interpers. .11 .33 .14 -.10 .06
DIB Psychosis .30 .33 .45" -.07 .36
DIB Affect .51* -.25 -.22 .09 .04
DIB Social Ad. .41 .19 -.02 -.21 -.14

 

* p < .05. *“p < .001

63

hypothesis was partially supported, in that change scores for one of the four
Rorschach indices - Defenses - was significantly correlated with the changes
on the DIB. Changes in Rorschach Defense scores were highly correlated (r =
.64, p <.001) with the change in the total DIB score, indicating that as DIB
scores improved, Defense scores improved. Difference scores for the other
three scales, Object Relations, Psychosis, and Ego Capacity, were not
significantly correlated with DIB difference scores. ..
To further test the relationship between Rorschach variables and :
specific areas of improvement on the DIB , Pearson product-moment

correlations were run between the 3 Rorschach indices and 3 DIB section

 

scores that based on conceptual similarity would be expected to correspond
(Ego Capacity vs. DIB Impulse Action Patterns, Reality Testing vs. DIB
Psychosis, and Object Relations vs. DIB Interpersonal Relations). Results of
these post-hoc analyses are also reported in Table 4. Because several
correlations were run in this analysis, the Bartlett chi-square test of overall
significance was calculated and was found to be significant (X 2 = 28.7, p = .02),
indicating that individual correlations within the matrix are less likely to be
spurious. Only one of the three correlations tested emerged as predicted:
changes in the Rorschach reality testing variable was significantly correlated
with changes in the DIB section score Psychosis ( g = .45, p <.05). Contrary to
the hypothesis, changes in the Rorschach variables Ego Capacity and Object
Relations were not significantly correlated with changes in the DIB section
scores Impulse Action Patterns and Interpersonal Relations, respectively.

In a post-hoc exploration of the data, the scores for the Rorschach
variables were broken down by outcome group: "BPD" or "Non-BPD" based

on their DIB diagnostic status at follow-up. This allowed for a more detailed

64

explication of the relationship between diagnostic outcome and change in
Rorschach variables over time. Means and standard deviations for the
overall group, the "BPD" group, and the "Non—BPD" group at both time 1 and
follow-up are reported in Table 5. From the breakdown of the data, it is
apparent that differences between the BPD and Non-BPD groups exist at both
Time 1 and Follow-up.

The third hypothesis, that age at follow-up will be related to
improvement in total DIB scores was tested using a Pearson correlation. Age
at follow-up was correlated with the DIB difference score. The hypothesis was
not supported, in that age was not found to be significantly related to changes
in the DIB (r = -.34, p < .12).

In order to test the exploratory hypothesis regarding the configurations
of personality organization, each of the four Rorschach components
(Defenses, Reality Testing, Ego Capacity, and Object Relations) was categorized
as either good/ neurotic, moderate/ borderline, or severe/ psychotic. Psychotic
scores indicated extremely pathological indices, borderline scores indicated
moderate pathology or inconsistent blends of severe and medium pathology,
and Neurotic scores indicated mild pathology. Each protocol was categorized
Neurotic, Borderline, or Psychotic on each of the four scales, and the
breakdown of subjects that fell into each category was calculated and is
reported in Table 6. From this pattern, each configuration was categorized
into one of four categories based on the theoretically predicted configuration
of the four personality indices (see Appendix A): Neurotic, Borderline,
Psychotic, or Other. The Other category was designated for protocols that
showed incongruent combinations that are not predicted by Kernberg's
model, for example poor reality testing and good ego capacity, or poor object

 

Time 1
Follow-up

Time 1
Follow-up

Time 1
Follow-up

Time 1
Follow-up

65

Table 5

Rorschach Variables by Improvement Status
at Time 1 and Follow-up

Object Relations

 

 

 

Total Group (N = 22) BPD Grpup (n = 10) Non-BPD Group (11 =12)

M S_D M 52 M .52
E: .

.16 2.8 1.41 3.7 -.89 1.2

.11 3.2 1.66 3.9 -1.18 1.6 .
E
1

Reality Testing

Total Group (N = 22) BPD Group (11 = 10) Non-Em Group (11 =12) B;

M 5.12 M SD M E

17.36 15.8 19.80 17.5 15.33 14.8

21.68 23.4 28.00 30.4 16.42 15.0

m

Iota] Group (N = Q) BPD Group (3 = 10) Non-BPD Group (11 =12)

M E M SD M Q

1.75 .18 1.72 .20 1.77 .17

1.68 .34 1.86 .28 1.52 .31

Ego Capacig

Total Group (N = 2) BED Gimp (n = 10) Non-BEL) Camp (3 =12)

M SD M E M it.)

1.68 1.5 2.10 1.8 1.33 1.1

2.32 1.4 2.70 1.6 2.00 1.0

66
Table 6

Means and Standard Deviations of Rorschach Variables by
Improvement Status and Time

Time 1
Follow-up

Time 1
Follow-up

Time 1
Follow-up

Time 1
Follow-up

 

 

 

 

Object Relations

Total Group BPD gr_'oup N on-BPD Grow

m=22> (_Ir=10) m=12)

M S_D M SD. M SD

.16 2.8 1.41 3.7 -.89 1.2

.11 3.2 1.66 3.9 -1.18 1.6
Reality Testing

Total Group BPD ggoup Non-BPD Gror_rp,

(N=22) (n=10) (n=12)

M S_D M SD M 52

17.36 15.8 19.80 17.5 15.33 14.8

21.68 23.4 28.00 30.4 16.42 15.0
[lemmas

Total Group

BPD gr_oup Non-BPD Group

 

(M=22) (n=10) (n=12)
M S_D M SD. M 5.12
1.75 .18 1.72 .20 1.77 .17
1.68 .34 1.86 .28 1.52 .31
Ego Capacity

Total Group BPD Group Non-BPD Group
(N=22) (n=10) (n=12)
M E M SD M SD
1.68 1.5 2.10 1.8 1.33 1.1
2.32 1.4 2.70 1.6 2.00 1.0

 

67

relationships and good defenses. Results of this pattern analysis are the
following: Of the 22 time 1 subjects, all of whom received a diagnosis of BPD,
none fit the predicted Borderline pattern exactly (Borderline level Defenses,
Object relations, Ego Capacity, and Reality Testing). One subject was
categorized in the Neurotic range on all 4 Rorschach indices, and no subjects
were categorized in the Psychotic range for all 4 Rorschach indices. Of the 22
subjects tested, 21 fell into the "Other" category. Seven (30%) of the 22 time 1 f ....
subjects had some blend of all three levels of functioning, 11 (48%) had a
combination of Neurotic and Borderline ranges, and 4 (17%) had a
combination of Borderline and Psychotic ranges. Thus, the

 

exploratoryhypothesis that the borderline group would show predictable l
configurations of personality indices was not supported.

Discussion
The current study set out to determine if changes in borderline
diagnostic status over time reflect changes in underlying personality
dynamics as assessed by the Rorschach. Due to the correlational nature of the
current data, we cannot make the assumption that the relationships between
the Rorschach and the symptom variables are causal. Nevertheless, the
findings do suggest that some aspects of personality functioning do improve

 

with time, and these changes correspond to changes in borderline symptoms.
The patients in the current study, almost all women, were all
diagnosed with borderline personality disorder, and were all disturbed
enough to require inpatient hospitalization. All patients reported receiving
at least one type of outpatient treatment during the follow-up interval, and
many had individual psychotherapy, group psychotherapy, and a psychiatrist.
Over half (57%) were re-hospitalized at some point between the initial
assessment and follow-up study. However, of the 28 patients followed a
substantial number had improved: less than half (39%) met DIB diagnostic
criteria at follow-up. The sample of borderlines successfully recontacted were
slightly older and had fewer males than a comparsion sample of dropouts.
The first hypothesis, that changes in overall indices of personality
organization on the Rorschach would correspond to changes in borderline
symptoms, was only partially supported. No significant relationship was
found when the data was analyzed according to an absolute scale of severity.
The effort to standardize the component variables by recoding them all in the
same range may have reduced the variability to such an extent that the real

changes in the component variables affected the overall index very little.

68

69

Therefore a second analysis was performed using standard scores, which
reflected severity relative to the group, and resulted in a score that
maintained the variance of the component scores. This analysis yielded a
non-significant trend (p > .07), indicating that there is some relationship
between the changes in the DIB and changes in the overall Rorschach score.

Several findings occurred when the separate personality components
were analyzed to determine if they changed in accordance with changes in
borderline symptoms. One particularly strong relationship emerged: positive
changes in defenses corresponded with a decrease in borderline symptoms.
Upon post-hoc examination, the "improved" group had decreased the

number of borderline and psychotic defenses and had a slight increase in

 

neurotic defenses between time 1 and follow-up, while the "not improved"
group showed a decrease in neurotic defenses and an increase in psychotic
defenses between time 1 and follow-up. Thus, as the defenses expressed on
the Rorschach became less pathological, the borderline diagnostic symptoms
decreased.

In order to understand more precisely which borderline symptoms
changed in accordance with the improvement in defenses, post-hoc
correlation analyses were performed between the difference scores for
Rorschach defenses and the difference scores for the DIB subsections (Social
Adaptation, Impulse Action Patterns, Affects, Psychosis, and Interpersonal
Relations). This set of analyses was not included in the original hypotheses
due to the fact that there was no clear prediction of the relationship between
defenses and these DIB section scores. Pearson correlations between the
difference scores for Rorschach Defenses and the five section scores were

performed. This yielded a significant correlation between defenses and the

70

DIB section score for Affect (; = .51, p < .05), and a non significant trend
between defenses and DIB Social Adaptation (; = .41, p = .06). Thus,
improvements in dyphoric affects (depression, anger, emptiness, lability) and
social functioning (occupational and social stability) were the main factors
reflected in the significant correlation between improvement in defenses and
improvement in overall DIB score.

The hypothesis that improvements in the other three aspects of '9'
personality organization - Reality Testing, Ego Capacity, and Object Relations -
would be significantly correlated with improvements in borderline
symptoms was only partly supported. Table 6 provides the means and

 

standard deviations for these variables broken down by time and outcome ;.
group ("BPD" or "Non-BPD" based on their DIB diagnostic status at follow-
up). Although differences between the Non-BPD and BPD groups were not
statistically significant due to the small sample size, important trends in the
data were observed. For Object Relations, the Non-BPD group scored better
than the BPD group at time 1 and follow-up, indicating that those who went
on to improve symptomatically started out with better object representations
than those who remained BPD. Nevertheless, the Non-BPD group showed a
substantial improvement in object relations between time 1 and follow-up,
while the BPD group actually showed a slight overall worsening of object
relations. Although the correlation between changes in total DIB score and
changes in object relations was not significant, the current data do suggest that
there are meaningful differences in object relations between those patients
who improved diagnostically and those who did not. When correlations
were performed between time 1 object relations and time 1 DIB scores no

significant correlation emerged (; = .37, p = ns) , but there was a significant

71

correlation between follow-up object relations and follow-up DIB scores (1 =
.54, p < .01). This relationship between object relations and borderline
symptoms does not show up when difference scores are used in the analyses.
The finding that those who went on to improve at follow-up had better object
relations at time 1 makes sense, in that patients who have more mature
object representations may adapt better and therefore make more gains over
short intervals. For example, these patients may have fewer disruptive r-
interpersonal crises, and may be more amenable to psychotherapy than
borderlines who have more primitive object representations.

Regarding Reality Testing, the "improved" group showed somewhat

 

less thought disorder than the "not improved" group at time 1, indicating
that those who went on to improve diagnostically were also less thought
disordered to begin with. At follow-up, while the "improved" group changed
little overall, the "not improved" group actually showed an overall
worsening of thought disorder. Nevertheless, the changes in thought
disorder on the Rorschach did correlate significantly with the changes in the
DIB subscale Psychosis, indicating that there was a relationship between these
two measures of disordered thinking. The correlation between the Rorschach
thought disorder measure and the overall DIB score, however, was not
significant. This most likely reflects the fact that Psychosis was only one
aspect of the DIB that changed over time.

For Ego Capacity, the correlation between change scores and DIB change
scores was not significant. Table 6 shows that both the "improved" and "not
improved" groups showed a worsening of ego indices at follow-up, with the
"improved" group showing slightly better ego capacity at both times than the

"not improved" group. It is unclear why this index would show a decrement

72

for both groups at follow-up. One hypothesis might be that the structure and
"hold" of the hospital setting may have influenced the responses at time 1,
which may have been reflected in indices of stress tolerance, impulsivity, and
affective regulation. However, DIB items reflecting impulsivity contradict
this idea, in that the group showed a decrease in impulsivity at follow-up.
Because this Rorschach scale was created for the current study and it has not
been validated, it is more likely that it may simply not accurately reflect FT
changes in ego capacity. Further validation of this scales would be necessary
before conclusions could be drawn about the meaning of this result.

The third hypothesis, that age at follow-up would be positively related

 

to borderline symptomatic improvement was not supported. No significant ,9
relationship bewteen age and changes in diagnostic status was found for
subjects in this sample. Although age has been related to improvement in
other studies, these follow-up intervals have been much longer and have
extended into the 4th and 5th decade for the majority of patients (Stone, 1990).
Although several patients in the current study were in their 408, the majority
were still in their 208 and 303 at follow-up. Improvement occurred in our
sample at a range of ages; however, the borderlines who did not improve
may be more like the patients that Stone (1990) describes who don't improve
until they reach a a later age. If at all possible it would be instructive to
recontact the patients in the current study who were still borderline at follow-
up in another ten to fifteen years to see if their adjustment has improved.
Although we did not measure potential correlates of improvement in the
current study, it would also be interesting to know in our sample which
demographic, family or other factors contribute to the differences between

those who improved early on versus later in life.

73

The robust relationship between defenses and symptom improvement
is compelling in that it is in line with the theoretical prediction of Kernberg
(1985) and others. As mentioned earlier, Kernberg sees the borderline's
primitive defensive functioning as a main obstacle to the neutralization of
angry affect, as well as interfering with the development of ego strength and
sublimatory channels. Although we cannot infer causality from correlational
data, the current study found a significant relationship between
improvement in defenses and affect, and a marginal relationship between
improvement in defenses and social functioning as predicted.

Furthermore, although these differences were not statistically
significant, it is notable that the group who at follow—up no longer met DIB
borderline diagnostic criteria looked less pathological in their Rorschach
variables of Reality Testing, Ego Capacity, and Object Relations at time 1.
Thus, even though they all met the borderline diagnostic criteria at time 1,
some borderlines initially looked less disturbed than others on the Rorschach,
and these borderlines were more likely to show symptomatic improvement
two to five years later. The results therefore support the use of the Rorschach
in predicting which borderlines are more likely to show symptomatic
improvements even over short periods of time. In the long-term outcome
study on borderlines conducted by Stone (1990b), he found that borderlines
with histories of parent-child incest, rape, and emotional abuse tended to
have worse outcomes. It seems fair to speculate that extreme psychological
damage of this sort would be reflected in Rorschach responses, which are in
turn related to differences in outcome. Despite the fact that all of the subjects
received outpatient therapy during the follow-up interval, perhaps
intrapsychic differences in the capacity to make use of treatment contibuted to

____.__ _"_‘J

74

the improvement of some patients more than others. Similarly, patients
with better defensive functioning and better internal models of relationships
may also be more likely to have "therapeutic" social relationships (e.g. family,
partners) that facilitate their improvement.

Another possible explanation of the differences in Rorschach pathology
between the "improved" and "not improved" groups is that there may be
undetected diagnostic differences between these two sets of borderline
patients. In a study by Kullgren and Armelius (1990), all of the borderlines
who had a concomitant major affective disorder at index improved to
neurotic personality organization at follow-up. The authors conclude:

"There are two possible explanations: A concomitant affective disorder either
a) produces structural changes in personality organization, or b) it influences
the measurement of personality structure"(p. 211). Similarly, Nelson et a1.
(1985) have suggested that a subset of patients who are not truly borderline
may be diagnosed borderline because of an affective disorder that produces
borderline features. For example, they point out that some patients may
experience more borderline-like features "such as chronic feelings of
emptiness, problems with being alone, and impulsivity" (p. 858) when
depressed. They suggest that in some depressives: "rather than depression
superimposed on a pre—existing personality disturbance, those features which
clinicians consider characterological are actually mood-state dependent"(p.
858). The current study did not assess the presence of affective disorders in the
subject population, and therefore this intriguing question should be
addressed in a subsequent investigation.

Likewise, another possibility is that the borderlines who did not
improve at follow-up were borderlines with more schizotypal features, who

75

tend to have especially severe interpersonal difficulties and more thought
disorder. The DIB is more inclusive than the DSM-IV criteria, and it is
possible that at least some of the patients in this borderline sample had
schizotypal features. One finding in the current data that may support this
idea was that the subgoup of borderlines who did not improve on the DIB at
follow-up showed a dramatic worsening of thought disorder over time. This
decline is consistent with the clinical picture of people with more schizotypal
traits. Unfortunately, independent measures of affective or schizotypal traits
were not obtained in the current study to address these diagnostic issues, but

they may be important variables to consider in future follow-up studies of

1* ‘.-

borderlines.

The findings from the exploratory hypothesis that borderlines do not
show a predictable pattern of Defenses, Reality Testing, Object Relations, and
Ego Capacity point to the wide variability in borderline presentations on the
Rorschach. This poses a challenge to the conservative, widely used
nomothetic approach to understanding borderline dynamics on the
Rorschach. When group data are used to describe borderlines and to compare
their Rorschachs to those of other diagnostic groups, much information is
lost about the potential variability within the groups. Although much effort
has been put into finding a single borderline profile on the Rorschach, this
line of thinking has had critics (Murray, 1993; Rosegrant, 1995). Rosegrant
(1995) has cautioned against using the Rorschach in isolation to diagnose.
The data from this study support this position: while the Rorschach is useful
to clarify personality dynamics, assess severity of pathology, determine areas
of strength and weaknesses, and even. give prognostic information, it is most

useful in combination with objective diagnostic measures. Within this

76

sample of borderlines, all of whom carried the same diagnosis and functioned
poorly enough to be hospitalized, most patients showed a range of
functioning with combinations of different levels of functioning on the four
Rorschach indices. This finding highlights the importance of using
idiographic methods to approach the study of borderline dynamics.

Perhaps one of the most important findings of the current study is that
Rorschach variables, considered to reflect intrapsychic processes, correlated
with reported behavior. Defenses in particular showed a very robust
correlation (; = .64) with the patient's self-reported borderline symptoms.
This is a both a theoretically important finding, given the centrality of
defenses in the creation and maintenance of borderline pathology (Kernberg,
1985), and a methdologically important finding for those who seek to validate
Rorschach scoring systems. Given the infinite number of situational
circumstances which act as mediator variables, such high correlation between
intrapsychic tendencies and behavior is unusual. The likelihood that this
correlation is spurious is low given the theoretically explained pattern of
intercorrelations between Rorschach variables and DIB subscales. Notably,
many of the Rorschach variables did not correlate highly with each other,
suggesting that they are measuring somewhat separate constructs. The
Rorschach indices did, however, show a somewhat predictable pattern of
correlations with the DIB subscales: for example, Ego Capacity correlated most
highly with DIB Impulsivity (g = .28), Reality Testing correlated most highly
with DIB Psychosis (; = .45), and Object Relations correlated highly with DIB
Interpersonal (; = .33). Although some of these correlations were non-

significant due to the small sample size, they do suggest non-random

 

77

associations between certain constructs on the Rorschach and reported
behaviors.

A number of methodological shortcomings limit the conclusions that
can be drawn from this study. The relatively small sample size limited the
power to detect real differences, which may have resulted in type 11 error on
some tests. The sample size also prevented the use of certain statistical tests,

 

such as multiple regression, which would have been more suited to assessing j”
the relative predictive power of Rorschach variables on DIB outcome. if
Furthermore, the follow-up sample may not have been representative of the ;
larger time 1 sample, which most likely included a number of patients who i
had extremely poor outcomes (e.g., homelessness, suicide). The subjects who I,

were able to be followed up all had some way of being contacted, either
through family or through a permanent residence of their own, suggesting
that they may have been a more socially connected and less symptomatic
subsample. However, the author happened to learn of the suicide of one
follow-up subject after our call back, indicating that the subjects in this sample
were not immune to tragic outcomes.

Another methodological shortcoming of the current study is the
absence of multiple outcome measures. Other measures of global functioning
as well as measures of comorbid diagnoses, such as affective disorders, in
addition to the DIB would have provided more information regarding the
reliability of the DIB as an outcome measure. Multiple outcome measures
would have also have allowed us to determine whether the Rorschach
variables predicted improvement across different domains of functioning.

A final caution about the results of the current study bears repeating:
the correlational nature of the data prevent making causal attributions;

78

therefore, we cannot conclude that intrapsychic changes produced
symptomatic ones or vice versa. We also tend to assume that the Rorschach
is measuring "trait" variables rather that "state" variables. Indeed, there is
some evidence from the literature that most Rorschach variables tend to
remain stable over time, and are not easily affected by situational factors
(Exner, 1986b). Nevertheless, the best test of whether the Rorschach changes
really reflected fundamental improvement in personality would be to do a F
second follow-up on the same subjects to see if their gains continue to be
stable over time.

Future research on the stability of Rorschach personality indices in

..1- and 38.. .' ‘A‘; '. 7'

 

 

borderlines would be improved by the addition of multiple outcome
measures, measures of premorbid (pre-hospital) functioning, and the
assessment of familial and social variables, such as abuse, which have been
shown to be predictive of outcome in other longitudinal studies. This would
allow us to see whether aspects of the Rorschach, such as object relations, are
correlated with these predictors. Another area of future research would be to
look at the covariance of certain Rorschach variables as they change over
time. For example, Murray (1993) has suggested that although borderline
dynamics will change over time, they will not change uniformly. He suggests
the first aspects of personality that will change are aspects of personality style
(such as cognitive style, affective control and experience, defensive
orientation, and flexibility), followed by structural factors, such as deficits in
handling anxiety, problems in experience of the self, and ideational
difficulties. Although beyond the scope of the current study, future
longitudinal studies might address whether the Rorschach shows changes
that conform to predictions of personality growth over time.

79

In summary, the current study found a significant association between
Defenses on the Rorschach and DIB improvement over time, such that
improvement in borderline diagnostic status correlated significantly with
improvement in the quality of defensive functioning on the Rorschach. This
finding strengthens the argument for the use of the Rorschach in detecting
and predicting outcome in future studies. Furthermore, the wide variability
found between Rorschach variables within the group of subjects receiving the
borderline diagnosis suggests that the search for a single "borderline" profile
is not realistic and may be misleading. Future studies of longitudinal course
of borderline pathology using the Rorschach would be improved by a careful
formulation about the progression of personality maturation as manifested
on the Rorschach, as well as the inclusion of multiple outcome measures and
additional diagnostic information. Idiographic rather than nomothetic
approaches to data are also recommended for future studies using the

Rorschach.

 

80

Footnote

lCopies of data for the current study can be obtained by writing The Coping
Study, c/ 0 Dr. Ken Silk, Department of Psychiatry, University of
Michigan Medical Center, 1500 E. Medical Center Drive, Ann Arbor,
zMI, 48109-0120.

 

APPENDICES

 

APPENDIX A

Pm "L'

”LID. . . ED .. 1’) ...acél.

80.858855

8502.88. 0: $8.. .0030 08:08 .5585. 35:5 508.888: 80:88.0. x878 .00: £00,885 .88. 830. 3.5.8. 3.8:.— 80: 8d. «...—dd...

.:0.8..88...5 850.5% 8.8:? $8.. .850

8208.9: 0.2.0.8 ......85. 88:5 58880.08: 50.8.3.0. 80.8 .00: 8088.0. x8528 .00.. $3885 .88. 830. $5.8. 3.88.. 88.5.; .... u a

.80.. .830 958:. 58 5.85. 588.88. 80.8.3.0. .088 500w 80:58.0. 30.58 500». “880.05 88.52%: 55.8. 3.88.. .05.... n . u

58.2.8855 850 585.8...

 

885 5:8 .2. acsru .5358:

 

 

8.8..88 .5 80: 8280.88,.
585.882. . 8.800 8.58:3 :. 8.85.8855 8502.88. 8.858855 8502.8

.830 0: 585.888: 88.800 .850 0: 50:88.8: $08.88 .830 .8288 92035—0:
588525808208 .50882:.:: 0.2.0.8 58588885 58898.80 .9030
.08—.00
$13.:—
.818... 82.888: .818... 82.8.3. 308850E 88185:. 62.8.8 haw... mo m8..—
.8. a 38.. 5:8. 5.2 means...
.8. 0. 3.898 98:28.0. :. 80.8.8.8 5388.: m. .988. .8. 0. 5.880 .349...“

 

8.82885 :5 808.0558 8.85 £085.88. 02.8.08

 

8.85.85. 8225...: .m:.....% ”88:85 88730..

 

 

:0.8~.8:.8..8.:.
60.8.5888. 5505:: 60.8.8. mGO-uauun—o
8.86.0. 8.88.. 88885 .88.-.. .... Erma—awwoh—

 

 

mg 3585:3—

 

“4.3432

< 035.893.

81

APPENDIX B

 

APPENDIX B

THE DIAQNQSI'IQ INTERVIEW FQR BQRDERLINE§ SD13!
§QMMARY OF ERDERLINE QRITERIA

I, fixlAL ADAPTATION
Unstable work history

Periods of achievement effectiveness
Active social life
Appears socially appropriate

11. IMPULSE ACI'IQN PATTERN§
Self-mutilation

Manipulative suicide threats
Pattern of serious substance abuse
Promiscuity or sexual deviance
Other impulsive behaviors

III. AFFE
Appears depressed or reports recent clinical depression
Feels angry and short-tempered
Demanding or entitled attitude
Chronic feelings of boredom, emptiness, or anhedonia
Absence of hypomanic or manic behavior

IV. Y 515
Derealization
Depersonalization
Fleeting drug-free psychotic depressed or paranoid experiences
Drug-related psychotic symptoms
Absence of true psychotic experiences

V. IN P R NAL R A N

Needs social activity/ avoids being alone

Acts as a caretaker/ conflicted about giving and receiving care

Has intense, unstable relationships

Devaluation, manipulation, and hostility emerge in
relationships

Masochism and dependency are problems in relationships

Has engaged in staff splitting/ evoked significant
countertransference in therapists

Absence of "loner" behavior

82

APPENDD( C

 

APPENDIX C

R R C HDEFEN E E: IN N TES

Defense scoring includes all responses, human, animal, and inanimate, as
well as comments made to the examiner. The following is a short-cut to
locating the general defense used in each percept. More than one defense
may be used in a single percept - please record as many as apply.

NEQRQ’I’IQ DEFEN§§§z

REACTIQN FQRMATIQN = bending over backward to appear the opposite of how
you feel

-expressions of extreme helpfulness in the testing situation, over 40 responses
-discomfort at messiness or aggression in card, anti-sentimentality,

- themes of obedience, cleanliness, benevolence

INTELLEQJAngATIQN = overly intellectual approaches to task

-technical or scientific language, textbook illustrations and artistic references,
- describing blots as"interesting", overly systematic approach to responding,
-probabilistic statements "maybe" with non-drive laden responses.

BATIQNALIZATIQN = justification

-using socially acceptable customs to justify drive content,
~reference to "chance" or "accidental" events of aggressive or sexual nature,
-hybrids with justification, personal experience used to justify percept

PQLLYANIfiH DENIAL = efforts to see excessively positive, cheerful content
-"nice" content, pleasure described, playful or childlike content (e.g. holidays),
- superficial positive remarks about the card.

I§QLATIQN = affect and ideas are disconnected

disconnected or separated objects, formal or stylized activity, distance in time
or place used for sexual or aggressive content

- machine content, , maps, pictures or sculptures, shadows,

- use of "object" or "image" to describe percept, reports of own thought
processes during response, lack of feeling described in reaction to drive-laden
percepts

- "cold" content, descriptions of precarious balancing

83

 

84

APPENDIX C
NE TI
W = mild denial, minimizing emotional significance

-disavowals ("not a bat"), minimizations, qualifications, retractions,
-tiny areas used in aggressive content,

-humor, cartoon characters

-use of "maybe" or "could be" with aggressive or sexual content

W

QEVALQAILQN= derogatory descriptions
- things described in negative or critical terms, missing parts, hybrid creatures,
mythological monsters, negative comments about self or testing process.

W = own objectionable feelings / experiences attributed to another
- seeing card as source of percepts - not owned as own

- suspiciousness of testing process, uncertainty about percepts' motives,

- spy and police content, surveillance, eyes watching, traps,

- impending harm or danger in content, victimization

PROIEQIVE IDENTIFICATIQN = projection with identification

- one object "injects" a feeling or substance into another to control or destroy
it, poisonous substances, objects placed inside others for safekeeping, empathy
with feared objects, texture used with aggressive objects, paranoid content

§PLIITIN§ = division into all good and all bad objects

- same person seen as having opposite attributes or body parts, love and hate
coexisting, transformations from good to bad or vice versa,

- one figure described as good and one as bad

- all good and all bad responses alternating within / between responses

PRIMITIVE IDEA IZATI N = belief in unrealistic, all-good, powerful images

- percepts described in excessively positive terms, famous or grandiose figures,
objects related to famous or grandiose figures,

- idealized remarks about the tester

W = inflated sense of self, grandiosity

- self flattery, percepts identified as self, use of "we" to describe blot,
- instructions on improving tester's technique, permission given to
examiner.

 

85

APPENDD( C

W

mgMANIg DENIAL = distortions used to avoid experiencing strong emotion
- justification of response based on wish to see something rather than actual
percept,

- attempts to deny aggressive content by criticizing blot construction,

- pollyannish concepts (goodness)

- aggressive characters with incongruent nice or cute features.

- idiosyncratic justifications for percepts, confabulations,

- hypomanic mood described, euphoric affect during testing

- repetition of affective words (e.g. mean witches are mean)

W = massive distortions of blot without- recognition ‘
-incongruous hybrids without justification, figures engaged in impossible
activities, merged figures,

-rejections on cards I, HI, VI, and VIII.

-pure F responses

-absence of populars

..-!

 

APPENDIX D

1. Defenses:

APPENDIX D

BQRSQHAQH VARIABLES AND SQQRINQ SYSTEMS

Rorschach Defense Scale, (Cooper, Perry, 6: Arnow, 1988).

Neurotic defenses: denial, intellectualization, isolation, reaction
formation, repression, rationalization, and pollyannish
denial

Borderline Defengs: splitting, primitive idealization,
devaluation, omnipotence, projective identification, and
projection

Psychotic Defenses: massive denial and hypomanic denial

Scoring formula:
[1 x (# Neurotic defenses) + 2 x (# Borderline Defenses) + 3 x (#

of Psychotic Defenses)] / R = weighted average

Scoring cutoffs:
Neurotic = < 1.33

Borderline = 1.34 to 2.00
Psychotic = > 2.00

2. Realig Testing:

Exner's Comprehensive System (Exner, 1986b)

WSUM6 (weighted sum of 6 special scores):

Deviant Verbalizations, level 1 or 2 (DV or DV2),

Deviant Response, level 1 or 2 (DR, DR2)

Incongruous Combinations, level 1 or 2 (INCOMl,
INCOMZ)

Fabulized Combination, level 1 or 2 (FABCOMI,
FABCOMZ)

Contamination (CONTAM)

Inappropriate Logic (ALOG)

86

 

87

APPENDIX D

2. Reality Testing; (Cont'd)

Scoring fgrm_u_la:

(1) x DV 4» (2) x DV2 + (2) x INCOM + (4) x INCOM2 + (3) x
DR1 + (6) x DR2 + (4) x FABCOM + (7) x FABCOMZ + (5) x
ALOG + (7) x CONTAM F

Scoring Cutoffs :
Normal, no thought disturbance: 0—5

Borderline, likely thought disturbance 6-12
Psychotic, severe thought disturbance 12+:

 

 

3. Ego Capacim

Exner Comprehensive System (Exner, 1986b)

1 . Pure C (overly intense, poorly
modulated emotions)
2. CF + C > PC (poor impulse control )
3. CF + C > PC a_n_d any of the following:
X-% > .29 (potential for decompensation)
AG > 1 (aggression)
S > 2 (anger, oppositional tendencies)
Afr > .85 (overresponsivity to emotional
stimuli)
4. Lambda > 1.2 or < .32 (emotional overinvolvement
or constriction)

5. WSumC 2 M + 2 (lack of capacity for delay,
reflection)
6. D < 0 (poor stress tolerance)

Scorin f rm a:
sum of items 1 - 6

Neurotic / good: 0 - 1
Borderline/ moderate: 2 - 4
Psychotic] severe: 5 - 6

4. Object relations:

88

APPENDIX D

Human Experience Variable (HEV), from the Ego Impairment

Index, (Perry 8: Viglione, 1991). Derived from
Exner's scoring:

""Good humans: Good form quality, no special scores,
populars, responses with COP

"Poor" humans: Poor form quality, special scores, AC or
MOR, (H), (Hd), Hd or Hx.

Scoring formula:
2 score = .51 (# of Poor humans) - .75 (# of Good humans)

+ .04.
Scoring Cutoffs:
Neurotic / good: 2 g -.70
Borderline/ moderate z > -.69 and < .54

Psychotic/ severe: z 2 .55

- 1
h. ' _-—

 

'u‘pnsmn :3. ' ..“.~ 4...

LIST OF REFERENCES

 

LIST OF REFERENCES

Acklin, M. W. (1993). Psychodiagnosis of personality structure 11: Borderline

personality organization. Iournal of Personality Assessment, 61(2), 329-
341.

American Psychiatric Association (1980). Diagnostic and Statistical Manual of
Mental Disorders Third Edition. Washington, DC: American
Psychiatric Association. ,

 

American Psychiatric Association (1987). Diagnostic and Statistical Manual of
Mental Disorders, Third Edition, Revised. Washington, DC: American
Psychiatric Association.

American Psychiatric Association (1994). Diagnostic and Statistical Manual of
Mental Disorders Fourth Edition. Washington, DC: American
Psychiatric Association

 

Arnow, D., 8r Cooper, S. H. (1984). The borderline patient's regression on the
Rorschach test: An object-relations perspective. Bulletin of the

Menninger Clinic, 48(1), 25-36.

Barasch, A., Frances, A., Hurt, S., Clarkin, 1., 8: Cohen, S. (1985). Stability and
distinctness of borderline personality disorder. American Iournal of

Psychiagx, 142(12), 1484-1486.

Barrash, 1., Kroll, 1., Carey, K., 8r Sines, L. (1983). Discriminating borderline
disorder from other personality disorders: cluster analysis of the
diagnostic interview for borderlines. Archives of General Psychiag,
49, 1297-1302.

Berg, 1. L. (1990). Differentiating ego functions of borderline and narcissistic
personalities. Iournal of Personaligg Assessment, 55(38z4), 537-548.

Berg, M. (1983). Borderline psychopathology as displayed on psychological
tests. Iournal of Personality Assessment, 47(2), 120-133.

89

 

Blatt, S. 1., Brenneis, B., Schimek, 1. G. 6: Click, M. (1976). Normal
development and the psychopathological impairment of the concept

of the object on the Rorschach. Iournal of Abnormal Psychology, 85(4),
364-373.

Blatt, S. 1., D'Afflitti, 1. P., & Quinlan, D. M. (1976). Experiences of depression
in normal young adults. loumal of Abnormal Psychology, 85, 383-389.

Blatt, 5.1., & Ford, R. Q. (1994). Theragutic Change: An Object Relations
Persngctive. New Yorlc Plenum Press.

Blatt, S. 1., 8r Lerner, H. (1983). The psychological assessment of object
representation. Iournal of Personalig Assessment, 47(1), 7-28.

Blatt, S. 1., Tuber, S. B., 8t Auerbach, 1. S. (1990). Representation of
interpersonal interactions on the Rorschach and level of

psychopathology. Iournal of Personality Assessment, 54(3&4), 711-728.

Burns, B., & Viglione, D. 1. (1996). The Rorschach human experience variable,
interpersonal relatedness, and object representation in nonpatients.

Psychological Assessment, 8(1), 92—99.

Buttenheim, M. C., Lohr, N. E., & Kerber, K. (1985, August). The impact of
depression on ego functioning. In Lohr, N. E., (Chair) Borderlines and
depressives: A cormgarative study of ego functions. Symposium
presented at the annual convention of the American Psychological
Association, Los Angeles, CA.

Carlson, C. F., Kula, M. L., 8: St. Laurent, C. M. (1997). Rorschach revised DEPI
and CD1 with inpatient major depressives and borderline personality
disorder with major depression: validity issues. Iournal of Clinical

Psychology, 53(1), 51-58.

Carpenter, W. T., 8r Gunderson, 1. G. (1977). Five year follow-up comparison
of borderline and schizophrenic patients. Comprehensive P§ychiat_ry,
1_8, 567-571.

Carr, A. (1987). Borderline defenses and Rorschach responses: A critique of

Lerner, Albert, and Walsh. Iournal of Personality Assessment, 51(3),
349-354.

 

91

Carr, A. C., Goldstein, E. (3., Hunt, H. F., Kernberg, O. F. (1979). Psychological

tests and borderline patients. Iournal of Personality Assessment, 43,
582-590.

Carsky, M., 8: Bloomgarden, 1. W. (1981). Subtyping in the borderline realm
by means of Rorschach analysis. Psychiatric Clinics of North America,
4, 101-116.

Chessick, R. D. (1968). The "crucial dilemma" of the therapist in the
psychotherapy of borderland patients. Ar'nerican Iournal of

Psychotherapy, 22(4), 665-666.

Clark, L. P. (1919). Some remarks upon the use of modified psychoanalysis
on the treatment of borderland neuroses and psychoses. Psychoanalytic
Review 6, 306-308.

 

Clarkin, 1. F., Widiger, T. A., Frances, A., Hurt, S. W., & Gilmore, M. (1983).
Prototypic typology and the borderline personality disorder. Iournal of

Abnormal Psychology, 92(3), 263-275.

Cohen, 1. (1968). Weighted kappa: Nominal scale agreement with provision
for scaled disagreement or partial credit. Psychological Bulletin, 70 (4),
213 - 220.

Coonerty, S. (1986). An exploration of separation-individuation themes in

borderline personality disorder. Iournal of Personality Assessment,
50(3), 501- 511.

Cooper, S. H., Perry, 1. C., & Arnow, D. (1988). An empirical approach to the
study of defense mechanisms: I. Reliability and preliminary validity of
the Rorschach defense scales. Iournal of Personaling Assessment, 52(2),
187-203.

Cooper, S. H., Perry, 1. C., Hoke, L., 6: Richman, N. (1985). Transitional
relatedness and borderline personality disorder. Psychoanalytic

Psycholggy, 2(2), 115-128.

Cornell, D. 6., Silk, K. R., Ludolph, P. 5., & Lohr, N. E. (1983). Test-retest
reliability of the diagnostic interview for borderlines. Archives of

General Psychiatry, 40, 1307-1310.

Exner, 1. E. (1986a). Some Rorschach data comparing schizophrenics with
borderline and schizotypal personality disorders. Iournal of

Personality Assessment, 50(3), 455-471.

92

Exner, 1. E. (1986b). The Rorschach: A Comprehensive System, Volume 1:
Basic Foundations Second Edition. New York: John Wiley 8: Sons.

 

Exner, 1. E. (1990). A Rorschach Workbook for the Comprehensive System,
Third Edition. Asheville, NC: Rorschach Workshops.

Exner,1.E. (1996). Critical bits and the Rorschach response process. Iournal of
Personality Assessment, 67(3), 464-477.

Frances, A., Clarkin, 1.F., Gilmore, M., Hurt, S.W., 8r Brown, R. (1984).
Reliability of criteria for borderline personality disorder: a comparison
of DSM-III and the diagnostic interview for borderline patients.

American Iournal of Psychiatry, 141(9), 1080-1084.

Frieswyk, S. & Colson, D. (1980). Prognostic considerations in the hospital
treatment of borderline states: the perspective of object relations theory
and the Rorschach. In 15. Kwawer, H.D. Lerner, P.M. Lerner, 6: A.
Sugarman (Eds), Borderline Phenomena and the Rorschach Test (pp.
229 - 255). New York, New York: International Universities Press, Inc.

Frosch, 1. (1964). The psychotic character: Clinical psychiatric consideration.
Psychiatric Qnarterly, 38, 81-96.

Gacono, C. B., Meloy, 1. R., 8: Berg, 1. L. (1992). Object relations, defensive
operations, and affective states in narcissistic, borderline, and antisocial

personality disorder. Iournal of Personality Assessment, 59(1), 32-49.

Gartner,1., Hurt, S. W., 8: Gartner, A. (1989). Psychological test signs of
borderline personality disorder: A review of the empirical literature.

lourn al of Personality Assessment, 53(3), 423-441.

Goldstein, E. G. (1990). Borderline Disorders: Clinical Models and
Techniques. New York: The Guilford Press.

Grala, C. (1980). The concept of splitting and its manifestations on the
Rorschach test. Bulletin of the Menninger Clinic, 44(3), 253-271.

Grinker, R. R., Werble, B., & Drye, R. C. (1968). The Borderline Syndrome.
New York: Basic Books.

Gunderson, 1. G., 8: Singer, M. T. (1975). Defining borderline patients: An
overview. American Iournal of Psychiatry, 132(1), 1-10.

 

93

Gunderson,1. G., Kolb, 1. E., 8: Austin, Y. (1981). The diagnostic interview for
borderline patients. American Iournal of Psychiag, 1g, 277-288.

Kernberg, O. (1975, 1985). Borderline Conditions and Pathological Narcissism.
Northvale, N1: Jason Aronson, Inc.

Kernberg, O. F. (1979). Two reviews of the literature: an assessment.

Schizophrenia Bulletin, 5(1), 53-58.

Kernberg, O., Goldstein, E. G., Carr, A. C., Hunt, H. F., Bauer, S. F.,
Blumenthal, R. (1981). Diagnosing borderline personality: A pilot
study using multiple diagnostic methods. The Iournal of Nervous and
Mentgl_Diseases 169 225-231.

 

Knight, R. P. (1953). Borderline states. In R. P. Knight 81: C. R. Friedman

(Eds), Psychoanalytic Pachianry and Psycholggy, (pp. 97-109). New
York: International University Press.

Kullgren, G., & Armelius, B. (1990). The concept of personality organization:
A long-term comparative follow-up study with special reference to

borderline personality organization. Iournal of Personalin Disorders, 4
(2), 203-212.

Leichtman, M. (1996). The nature of the Rorschach task Iournal of
Personaligg Assessment, 67(3), 478-493.

Lerner, H. D., Albert, C., St Walsh, M. (1987). The Rorschach assessment of
borderline defenses: A concurrent validity study. Iournal of

Personality Assessment, 51(3), 334-348.

Lerner, H. D., Sugarman, A., 8: Gaughran, 1. (1981). Borderline and
schizophrenic patients: a comparative study of defensive structure.

The Iournal of Nervous and Mengl Disease, 169(11), 705-711.

Lerner, H. D., at St. Peter, S. (1984). Patterns of object relations in neurotic,
borderline, and schizophrenic patients. Pachiag, 47, 77-92.

Lerner, P. M. (1990). Rorschach assessment of primitive defenses: A review.

Iournal of Personality Assessment, 54(18z2), 30-46.

Links, P. S., Mitton, 1. E., St Steiner, M. (1990). Predicting outcome for

borderline personality disorder. Comprehensive Psychian'y, 31(6), 490-
498.

94

Links, P. S., Mitton, 1. E., & Steiner, M. (1993). Stability of borderline
personality disorder. Canadian Iournal of Psychiany, 33, 255-259.

Mayman, M. (1967). Object representations and object relationships in

Rorschach responses. Iournal of Prpjective and Personality
Assessment 31 17-24.

 

McGlashan, T. H. (1984). The Chestnut Lodge follow-up study: Follow-up

methodology and study sample. Archives of General Psychian'y, 41,
573-585.

McGlashan, T. H. (1986). The Chestnut Lodge follow-up study: 111. Long-
term outcome of borderline personalities. Archives of General

Psychiagy, 43, 20-30.

Murray, 1. F. (1993). The Rorschach search for the borderline holy grail: An
examination of personality structure, personality style, and situation.

Iournal of Personality Assessment, 61(2), 342-357.

Nelson, H. F., Tennen, H., Tasman, A., Borton, Kubeck, M., St Stone, M.
(1985). Comparison of three systems for diagnosing borderline

personality disorder. American Iournal of Psychiatry, 142, 855-858.

Nigg, 1. T., Silk, K. R, Westen, D., Lohr, N. E., Gold, L. 1., Goodrich, 5., Gt
Ogata, S. (1991). Object representations in the early memories of
sexually abused borderline patients. American Iournal of Psychiatry,
1413(7), 864-869.

Paris, 1., Brown, R., & Nowlis, D. (1987). Long-term follow-up of borderline
patients in a general hospital. Comprehensive Psychiag, 28, 530-535.

Paris, 1. (1988). Follow-up studies of borderline personality disorder: A
critical review. Iournal of Personality Disorders, 2(3), 189-197.

Patrick, 1 8t Wolfe, B. (1983). Rorschach presentation of borderline
personality disorder: primary process manifestations. |o urnal of

Clinical Psycholngy, 39(3), 442-447.

Perry, 1. C. (1988). A prospective study of life stress, defenses, psychotic
symptoms, and depression in borderline and antisocial personality
disorders and bipolar type II affective disorder. Iournal of Personality

Disorders, 2(1), 49—59.

95

Perry, 1. C., & Cooper, S. H. (1986). A preliminary report on defenses and
conflicts associated with borderline personality disorder. Iournal of the
Americgn Psychoanalytic Association, 34, 863-893.

Perry, W., & Viglione, DJ. (1991). The ego impairment index as a predictor of
outcome in melancholic depressed patients treated with tricyclic

antidepressants. Iournal of Personality Assessment, 56(3), 487-501.

Perry, W., McDougall, A., 8: Viglione, D. (1995). A five-year follow-up on the
temporal stability of the ego impairment index. Iournal of Personality
Assessment 64(1), 112-118.

Plakun, E. M. (1991). Prediction of outcome in borderline personality
disorder. Iournal of Personality Disorders, 5(2), 93-101.

Pressler, E. 1. (1990). Short-term Ontcome of Patients with BPD. Unpublished
Doctoral Dissertation, University of Michigan, Ann Arbor, MI.

Rappaport, D., Gill, M., 8: Schafer, R. (1946). Diagnostic Psychological Testing,
Volume 2. Chicago: Year Book Publishers.

Rosegrant, 1. (1995). Borderline diagnosis in projective assessment.
Psychoanalytic Psychology, 12(3), 407-428.

Schafer, R. (1967). ProEctive testing and psychoanalysis. New York:
International Universities Press, Inc.

Segal, H. G., Westen, D., Lohr, N. E., Silk, K. R., 81: Cohen, R. (1992). Assessing
object relations and social cognition in borderline personality disorders
from stories told to the picture arrangement test of the WAIS—R.
lourn_al of Personality Discfllers. 6(4), 458-470.

Silk, K. R., Lohr, N. E., Ogata, S. N., 8: Westen, D. (1990). Borderline
inpatients with affective disorder: Preliminary follow-up data. Iournal

of Personality Disorders, 4(2), 213-224.

Spear, W. E., 8: Lapidus, L. B. (1981). Qualitative differences in manifest
object representations: Implications for a multidimensional model of

psychological functioning. Iournal of Abnormal Psychology, 90, 157-
167.

Spitzer, R. L., Endicott,1., 8t Gibbon, M. (1979). Crossing the border into
borderline. Archives of General Psychiatr_'y, 36, 17-24.

96

Stern, A. (1938). Psychoanalytic investigation and therapy in the borderline
group of neuroses. Psychoanalgn'c (garterly, 7, 467-489.

Stone, M. H. (1987). Systems for defining a borderline case. In 1. S. Grotstein,
M. F. Solomon, 8: 1. A. Lang, (Ed.s)., The Borderline _Pgtient:

Emerging Concepts in Diagnosis, Psychodynamics, and Treatment, Vol
1, (pp. 13-35), Hillsdale, N1: The Analytic Press.

Stone, M. H. (1990a). Multimodal therapy: Applications to partial
hospitalization in the light of long-term follow-up of borderline

patients. International Iournal of Partial Hospitalization, 6, 1-14.

Stone, M. H. (1990b). The Fate of Borderline Patients: SgccessflOntcome
Ed Psychiatric Practice. New York: The Guilford Press.

Stuart, 1., Westen, D., Lohr, N., Benjamin, 1., Becker, 5., Vorus, N., 8: Silk, K.
(1990). Object relations in borderlines, depressives, and normals: An
examination of human responses on the Rorschach. Iournal of

Personality Assessment, 33, 296-318.

Sugarman, A., 8: Lerner, H. D. (1980). Reflections on the current state of the
borderline concept. In 1. S. Kwawer, H. D. Lerner, P. M. Lerner, & A.
Sugarman, (Ed.s), Borderline Phenomena and the Rorschach Test (pp.
11-37), New York: International Universities Press.

Thaler Singer, M., 8: Larson, D. G. (1981). Borderline personality and the
Rorschach test. Archives of General Psychiag, 38, 693-698.

Urist, 1. (1977). The Rorschach test and the assessment of object relations.
Iourn_al of Personality Assessment, 41, 3-9.

Urist, 1. (1980). The continuum between primary and secondary process
thinking: Toward a concept of borderline thought. In]. S. Kwawer, H.
D. Lerner, P. M. Lerner, 8: A. Sugarman, (Ed. 5), Borderline
Phenomena and the Rorschach Test (pp. 133-154), New York:
International Universities Press.

 

Weiner, 1., & Exner, 1. (1991) Rorschach changes in long term and short term

psychotherapy. Iournal of Personality Assessement, 56, 453-465.

Werble, B. (1970). Second follow-up study of borderline patients. Archives of
General Psychiat_ry, 23, 3-7.

nrcnrcnu STATE UNIV. LIBRARIES
"WWI“WWW“llll“ll”llllllllllll
31293015755725