z L6 L..r
J7. ,« ..t m .
1.111....2: . 1.” «$19 “an“.
r539". ... ._ d}: firmm‘
kwmn fig: .V a. Eruwfwnu
. . . ,- | A p , b .
, . 5...
EM??? .
mum, ..n V

.
1W2 .

r

«A:
.mw. ....r
.. ,. ..
Wm. «.....finv
abwar

4

.2...
_ -...nrsz.
...Ié ..
9 if}:
...:i.

... \
.... .Ja .
1: a .
“33.5,.
9 . its...

1

1... .

...» 13.3.3.
2 5:: :5..;....
s 4,. ....35lx

A.
1-15..
wilt). en a
3‘3 5‘! I
3 (in me: I)! .
41(4) . 11.5.»!
illlri; :- 13..
:xx ...: 1 ES
. 3.
1.3.

. .

1 a... .2

..-]. : ...: {L r .,

i} ...: $1.. 1.... 119.13.”. h an...

5...}...41 r1. ... laufl‘rllt-unxa ,7...Whnnfl

2...}: .. - ll . \
9mg.) 92,.“ ...... ”7.1.... ‘
r ,1.

it, x I 1!.
(nix; Jo
1.13.002... 4

.
.3723...» ,
.... z . 1.4:.
7!}...3:
:3 1.3..
., .5

 

gnaw. jadmmwwms

E. d ' ...LIA. '

l we:

2- LIBRARY
2% Michigan State
University

This is to certify that the
dissertation entitled

Personality Clusters and Family Relationships
in Women with Eating Pathology

presented by
Patrick Scott Perkins

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in PsychoLogy

S”

major Professor’s Signature

 

 

\

12/12/2005

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

W‘ 317 3098

 

[£000 4 5r
1 v

1
91.

R8

CO

 

 

 

 

 

 

 

 

 

 

 

 

2105 p:lClRC/DatoDuo.indd-p.1

 

PERSONALITY CLUSTERS AND FAMILY RELATIONSHIPS
IN WOMEN WITH EATING PATHOLOGY

By

Patrick Scott Perkins

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Psychology

2005

ABSTRACT

PERSONALITY CLUSTERS AND FAMILY RELATIONSHIPS
IN WOMEN WITH EATING PATHOLOGY

By

Patrick Scott Perkins
Objective: Recent research has identified three personality clusters in eating disordered
women that predict important clinical variables such as level of functioning and treatment
length and outcome better than DSM-IV eating disorder diagnoses. The purpose of the
present study was to expand upon previous research by comprehensively examining
personality clusters and family relationships in women with subclinical eating pathology
and by attempting to replicate previous links between clusters and level of functioning
variables. As an exploratory aim of the study, personality clusters of women without
eating pathology (i.e., control group) were also analyzed to determine if three similar
personality clusters were present and whether clusters differed on level of functioning
variables and family relationships. Method: Participants included 795 female
undergraduate students at Michigan State University who were screened for the presence
of disordered eating and classified into one of three groups: women who restrict food but
do not binge or purge (R; n = 20), women who binge and purge (B/P; n = 62), and
women with no eating pathology (CON; n = 109). Participants completed measures
assessing disordered eating attitudes and behaviors, personality, family relationships, and
psychosocial and clinical functioning. Results: Three personality clusters (i.e., Adaptive,
Rigid, Dysregulated) were found in women with eating pathology, and these clusters
differed on level of functioning and family relationships. In addition, clusters predicted

level of functioning better than initial disordered eating groups (i.e., R, B/P). Three

clusters were also found in the CON group, and were labeled Resilients, Overcontrollers,
and Undercontrollers. These clusters also differed on level of functioning and family
relationships. Discussion: Findings from this study increase understanding of personality
profiles of women with subclinical eating disorders and provide further validation that
personality clusters, and not DSM-IV eating disorder diagnoses, have more clinical utility
in predicting psychosocial functioning. Specific family relationships associated with
personality clusters in eating disordered women were also found that could be important
to target in treatment. Finally, results pointed to three personality clusters in the CON
group that highly resembled clusters in the disordered eating groups, and which may be

relevant for understanding other forms of psychopathology.

For my father

in memoriam

iv

ACKNOWLEDGNIENTS

I am very fortunate to have been surrounded by a cadre of highly talented,
insightful, and supportive individuals throughout this pleasurable and painful process.
My advisor and mentor, Dr. Kelly Klump, deserves a tremendous deal of credit for this
final product. Perhaps more importantly, I am indebted to her for taking an initially
insecure graduate student unsure of his capabilities and turning him into a confident,
mature clinician and researcher. It is also an honor to have been Kelly’s first doctoral
student, and I believe that the current body of work truly represents the culmination of
our strong collaboration over the years. Kelly’s influence as a scholar, teacher, and a
genuinely kind and caring person is sure to serve me well as I move forward. I also owe
thanks to my other committee members, Drs. Alytia Levendosky, Neal Schmitt, and
Richard Lucas. Each of them offered valuable perspectives that contributed to making
this study particularly rigorous and methodologically sound.

I am grateful to my parents, Barbara and Tom, for their unwavering support
during all of my schooling. Their desire for me to choose a path I was passionate about
has allowed me to realize my potential. They have generously given whatever possible
so that I could be successful. My wife, Kelley, has been a constant source of love from
the beginning, and my closest traveler on this journey. She has motivated me, challenged
me, and tolerated my worst moods. She is nothing short of an angel, and I am deeply
grateful for her presence in my life. I have dedicated this work to the memory of my
father. Although he was not able to directly witness the beginning or end of this process,

he has always been a guiding force while I have been completing my own dissertation.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ............................................................................................................ x
INTRODUCTION .............................................................................................................. 1
Personality ...................................................................................................................... 3
Personality Clusters ....................................................................................................... 6
Family Relationships, Eating Disorders, and Personality ............................................ 16
Specific Aims and Hypotheses .................................................................................... 23
Overall Purpose of Current Study ........................................................................... 23
Specific Aims .......................................................................................................... 25
Exploratory Aim ..................................................................................................... 27
METHOD ......................................................................................................................... 31
Participants ................................................................................................................... 3 1
Measures ...................................................................................................................... 32
Demographic Information ....................................................................................... 32
Disordered Eating Measures and Classifications .................................................... 32
Eating Disorder Inventory-2 .............................................................................. 32

Bulimia T est-Revised ......................................................................................... 34
Three-Factor Eating Questionnaire .................................................................... 34
Classifications .................................................................................................... 35
Personalin Clusters ................................................................................................ 37
Dimensional Assessment of Personality Pathology-Basic Questionnaire ......... 37

Parental Relationships ............................................................................................. 38
Parental Bonding Instrument ............................................................................. 38
Structural Analysis of Social Behavior .............................................................. 39

Level of Functioning ............................................................................................... 42
Social Adjustment Scale Self-Report ................................................................. 42

History of Treatment .......................................................................................... 43
Statistical Analysis ....................................................................................................... 44
Internal Consistencies, Variable Transformations, and Effect Sizes ...................... 44

Body Mass Index Across Groups ........................................................................... 45
Specific Aims .......................................................................................................... 45
Specific Aim #1: Replication of Personality Clusters ....................................... 45

Specific Aim #2: Personality Heterogeneity of R and B/P Groups ................... 46

Specific Aim #3: Level of Functioning ............................................................. 46

Specific Aim #4: Family Relationships ............................................................. 47
Exploratory Aim: Analysis of Control Group ........................................................ 47

vi

RESULTS ......................................................................................................................... 49

Internal Consistencies .................................................................................................. 49
Body Mass Index Across Groups ................................................................................ 49
Specific Aim #1: Replication of Personality Clusters ................................................. 50
Personality Factor Structure .................................................................................... 50
Cluster Analyses of Disordered Eating Groups ...................................................... 54
Disordered Eating Symptoms Across Clusters of Eating Pathology ...................... 56
Specific Aim #2: Personality Heterogeneity of R and B/P Groups ............................. 56
Specific Aim #3: Level of Functioning ....................................................................... 57
Comparisons Across Clusters ................................................................................. 57
Predictive Relationships between Original Disordered Eating Groups, Personality
Clusters, and Functioning ....................................................................................... 59
Specific Aim #4: Family Relationships ....................................................................... 60
PBI Scores ............................................................................................................... 60
SASB Scores ........................................................................................................... 62
Exploratory Aim: Analysis of Control Group ............................................................. 65
Cluster Analysis of Control Group ......................................................................... 65
Level of Functioning ............................................................................................... 66
Family Relationships .............................................................................................. 68
DISCUSSION ................................................................................................................... 7 1
Personality Clusters in Women with Subclinical Eating Pathology ............................ 72
Heterogeneity of Personality in Women with Subclinical Eating Pathology .............. 75
Level of Functioning in Women with Subclinical Eating Pathology .......................... 76
Comparisons Across Clusters ................................................................................. 76
Comparisons of Clusters versus R and B/P Groupings .......................................... 77
Family Relationships in Women with Subclinical Eating Pathology .......................... 78
Analysis of Control Group ........................................................................................... 81
Conclusions and Implications ...................................................................................... 82
Limitations ................................................................................................................... 88
REFERENCES ................................................................................................................. 90
APPENDICES ................................................................................................................ 101
Appendix A: Tables ................................................................................................... 101
Appendix B: Figures .................................................................................................. 139

vii

LIST OF TABLES

Table 1: Descriptions of DAPP—BQ Dimensions ............................................................ 101
Table 2: Internal Consistencies of the Eating, Personality, Family Relationship, and Level

of Functioning Variables ................................................................................... 102
Table 3: Correlations of 18 DAPP-BQ Subscales (n = 82) ........................................... 104

Table 4: Direct Oblimin Rotated Principal Components Factor Loadings of 18 DAPP-BQ
subscales Across R (n = 20) and B/P (n = 62) Groups: Four-Factor Solution . 105

Table 5: Direct Oblimin Rotated Principal Components Factor Loadings of 18
DAPP-BQ subscales Across R (n = 20) and B/P (n = 62) Groups: Three-Factor
Solution ............................................................................................................. 106

Table 6: Direct Oblimin Rotated Principal Components Factor Loadings of 14
DAPP-BQ subscales Across R (n = 20) and B/P (n = 62) Groups: Three-Factor
Solution ............................................................................................................. 107

Table 7: Mean Differences in DAPP-BQ Subscales Across Adaptive (n = 40), Rigid
(n = 35), and Dysregulated (n = 7) Clusters and CON (n = 109) ..................... 108

Table 8: Mean Differences in Eating Disorder Symptoms Across Adaptive (n = 40),
Rigid (n = 35), and Dysregulated (n = 7) Clusters ............................................ 112

Table 9: Mean Differences in SAS-SR Subscales Across Adaptive (n = 40), Rigid
(n = 35), and Dysregulated (n = 7) Clusters and Control (n = 109) Women.... 113

Table 10: Hierarchical Multiple Regression of R and B/P Groups and Personality Clusters
on SAS-SR Student Work (N = 80), Social and Leisure Activities (N = 82),
Relationships with Extended Family (N = 82), and Total Score (N = 82) ...... 115

Table 11: Logistic Regression of R and B/P Groups and Personality Clusters on History
of Psychiatric Medication (N = 79) ................................................................ 116

Table 12: Mean Differences in PBI Subscales Across Adaptive (n = 40), Rigid (n =
35), and Dysregulated (n = 7) Clusters and CON (n = 108) Women ............. 117

Table 13: Mean Differences in SASB Clusters Across Adaptive (n = 40), Rigid (n =
35), and Dysregulated (n = 7) Clusters and CON (n = 108) Women ............. 119

Table 14: Mean Differences in DAPP-BQ Subscales Across Resilient (n = 53),
Overcontroller (n = 43), and Undercontroller (n = 12) Clusters ..................... 128

viii

Table 15: Mean Differences in SAS-SR Subscales Across Resilient (n = 53),
Overcontroller (n = 43), and Undercontroller (n = 12) Clusters ..................... 130

Table 16: Mean Differences in PBI Subscales Across Resilient (n = 53), Overcontroller
(n = 43), and Undercontroller (n = 12) Clusters ............................................. 131

Table 17: Mean Differences in SASB Clusters Across Resilient (n = 53), Overcontroller
(n = 42), and Undercontroller (n = 12) Clusters ............................................. 132

LIST OF FIGURES

Figure 1: Surface 1 of the SASB Model: Combined quadrant & cluster SASB

models .............................................................................................................. 139
Figure 2: Surface 2 of the SASB Model: Combined quadrant & cluster SASB

models .............................................................................................................. 140
Figure 3: Surface 3 of the SASB Model: Combined quadrant & cluster SASB

models .............................................................................................................. 141
Figure 4: Three-factor model of personality ................................................................... 142

Figure 5: Dendrogram of Adaptive (n = 40), Rigid (n = 35), and Dysregulated (n = 7)
Clusters ............................................................................................................ 143

Figure 6: Means of DAPP—BQ Subscales Across Adaptive (n = 40), Rigid (n = 35),
and Dysregulated (n = 7) Clusters ................................................................... 144

Figure 7: Means of DAPP-BQ Subscales Across Resilient (n = 53), Overcontroller
(n = 43), and Undercontroller (n = 12) Clusters ............................................. 145

INTRODUCTION

The primary eating disorders are anorexia nervosa (AN), bulimia nervosa (BN),
and eating disorders not otherwise specified (EDNOS). In AN, there is an intense fear of
weight gain, and a refusal to maintain a normal body weight (i.e., weight less than 85% of
ideal body weight; American Psychiatric Association, 2000). In addition, AN is
characterized by a disturbance in perception of body shape or size and, in women, the
absence of at least three consecutive menstrual cycles. Patients with AN are categorized
into one of two types: 1) Restricting Type; and 2) Binge-Eating/Purging Type. The first
type consists of individuals who have used dieting, fasting, or excessive exercise to lose
weight; binge eating and purging are not consistently used by these individuals. The
second type of AN is made up of individuals who have consistently engaged in binge
eating and/or purging behaviors such as self-induced vomiting, or the misuse of laxatives,
diuretics, or enemas.

BN is characterized by repeated binge eating and inappropriate compensatory
behaviors to prevent weight gain (American Psychiatric Association, 2000). Some of
these behaviors include fasting, excessive exercise, self-induced vomiting, and the misuse
of laxatives, diuretics, or enemas. The binge eating and inappropriate compensatory
behaviors both occur at least twice a week for 3 months. In BN, like AN, there is a
disturbance in the perception of body shape and size, such that these factors unduly
influence self-esteem. Two types of BN have been identified: 1) Purging Type; and 2)
Nonpurging Type. Individuals with purging type have consistently used vomiting and/or

other means including laxatives, diuretics, or enemas to compensate for binge eating. In

contrast, the nonpurging type of BN consists of individuals who have not consistently
vomited or used laxatives, diuretics, or enemas, but who have used other inappropriate
compensatory behaviors such as fasting or excessive exercise instead.

The broad category of EDNOS includes eating disturbances that do not meet
formal criteria for AN or BN, but that nevertheless represent serious eating pathology.
For example, women who meet all criteria for BN except that they do not engage in binge
eating and/or inappropriate compensatory behaviors at the frequency required by the
DSM-IV, fall into this category. Other eating disturbances that are part of this category
include ones that resemble AN. In these cases, women meet all criteria for this disorder,
except they may have had regular menstrual cycles, or they may have lost a significant
amount of weight but their weight is still classified in the normal range.

Eating disorders, like other psychiatric disorders, present a conundrum for both
the researcher and clinician, as these disorders have generally been shown to result from
a combination of genetic and environmental influences (Klump, Miller, Keel, McGue, &
Iacono, 2001; Klump, Wonderlich, Lehoux, Lilenfeld, & Bulik, 2002). Thus, sorting out
their complex etiology is difficult, but critical for a more informed understanding of
effective treatment strategies for women with eating disorders. Knowledge of the factors
that lie behind the overt eating disorder symptoms will allow the clinician to hone in on
areas most amenable to change and subsequently devise ways to effect such change.
Two specific etiological factors of eating pathology that appear highly relevant to

treatment efforts are personality and family relationships.

Personality

The utility of examining personality in women with eating disorders has been
supported by research demonstrating that personality significantly predicts the
development of disordered eating (Leon, Fulkerson, Perry, & Cudeck, 1993; Leon,
Fulkerson, Perry, Keel, & Klump, 1999). AN and BN have been found to have strikingly
similar personality profiles. For example, both AN and BN women exhibit high levels of
negative emotionality (Casper, Hedeker, & McClough, 1992; Pryor & Wiederman,
1996), stress reactivity (Casper et al., 1992; Pryor & Wiederrnan, 1996), and harm
avoidance (Brewerton, Hand, & Bishop, 1993; Bulik, Sullivan, Fear, & Pickering, 2000;
Fassino etal., 2002; Fassino et al., 2003; Kleifield, Sunday, Hurt, & Halmi, 1994; Klump
et al., 2000). The personality traits obsessiveness and perfectionism have also been
associated with both eating disorder diagnoses (Vitousek & Manke, 1994). These
findings suggest that women with AN and BN chronically experience a range of negative
emotions such as anxiety, anger, and resentment and are generally inhibited, cautious,
and averse to novel experiences.

Despite the many personality traits shared by women with these diagnoses, a few
differences on key constructs including novelty seeking and impulsivity have been
observed. Some women with BN evidence high levels of novelty seeking (Brewerton et
al., 1993; Kleifield etal., 1994) and impulsivity (Swift & Wonderlich, 1988), whereas
women with AN are generally low on novelty seeking (Casper et al., 1992; Fassino et al.,
2002; Kleifield et al., 1994) and are instead inhibited (Casper et al., 1992). Nonetheless,
in general, the similarities in personality traits in women with AN and BN seem to

outweigh the differences.

Studies examining subtypes of AN and BN women provide additional support for
overlapping personality traits across AN and BN. For example, AN women who engage
in bulimic behaviors (i.e., binge-eating/purging subtype; BAN) have been found to
resemble BN women more than AN women in their personality styles (Klump et al.,
2000), particularly in their levels of novelty seeking (Casper et al., 1992; Fassino, Abbate
Daga, Piero, Leombruni, & Giacomo Rovera, 2001; Kleifield et al., 1994). These
findings also call into question previously observed differences on the trait novelty
seeking between women with AN and BN. Such differences may be limited to a
particular subtype of AN, rather than all women with AN. As evidence of this, women
with AN who do not engage in bulimic behaviors (i.e., restricting type; RAN) have been
shown to have lower levels of novelty seeking than BAN, BN, and control women
(Kleifield et al., 1994).

One explanation for the broad overlap in personality traits across AN and BN
women is that both diagnoses share a large number of psychological and behavioral
symptoms related to food and body image. In fact, often the only diagnostic difference
between BAN and BN concerns the medical symptom low weight, in which a diagnosis
of BAN is given if a woman has a body weight less than 85% of her ideal body weight.
If her weight is above 85% and she has the same symptoms (i.e., binge eating, purging,
food restriction, self-esteem influenced by body weight/shape), then she is given a
diagnosis of BN rather than AN. This subtle but important diagnostic distinction is
highlighted by the fact that a significant number of patients fully cross-over from one
diagnosis (e.g., BAN) to another (e.g., BN) during their illness, with often the only

symptom change being weight loss or gain relative to the 85% criterion (Tozzi et al.,

2005; Vitousek & Manke, 1994). Not surprisingly then, research suggests that
approximately 50% of women with AN develop BN symptoms at some point during their
illness (Garfinkel, Modlofsky, & Garner, 1980), and approximately 25% to 30% of
patients with BN report a history of AN (Kaye, Klump, Frank, & Strober, 2000; Tozzi et
al., 2005).

Given the heterogeneity within eating disorder diagnoses and frequent diagnostic
cross-overs, Westen and Harnden-Fischer_(2001) argue that attempts made to define pure
personality profiles for AN and BN in traditional personality studies have inevitably
failed. These researchers suggest that personality profiles may be heterogeneous rather
than homogeneous in women with the same eating disorder diagnosis, and that specific
personality clusters may cut across both AN and BN diagnoses. In short, Westen and
Harnden-Fischer have proposed that several personality clusters may be found in women
with AN and BN, and that AN women may share a range of personality traits with BN
women.

This hypothesis has serious implications for the current DSM-IV diagnostic
system. If personality clusters are shown to link women from different diagnoses, and if
considerable heterogeneity in personality clusters exist within eating disorder diagnoses,
then current Axis I diagnoses include heterogeneous patients and may be deficient in
providing consistent predictions of the etiology, course, and treatment outcome for
individual eating disorders. Personality clusters, instead, may offer more accurate
information with regard to these clinical predictions. Recent research has partially
confirmed this idea by using cluster analytic techniques that have found personality

clusters to be better predictors of level of functioning (i.e., psychological, social,

occupational functioning, and history of hospitalization) than eating disorder diagnoses

(Westen & Harnden-Fischer, 2001).

Personality Clusters

Unlike traditional studies of personality in women with eating disorders that
examine personality traits separately across diagnoses, studies using a cluster analytic
method group patients with all eating disorders into clusters based on specific
conglomerations of personality traits. This technique offers a way to measure whether
personality is indeed heterogeneous within AN and BN categories and whether
personality clusters cut across diagnoses. If heterogeneity exists, the clustering method
should yield more than one distinct personality cluster for women with AN and more than
one cluster for women with BN. If personality clusters cut across diagnoses, there should
be overlap in cluster membership between AN and EN women. In general, cluster
analytic studies of personality in eating disordered women have found three consistent
personality clusters that meet both of the above conditions and that appear to be clinically
relevant in terms of predicting treatment length and outcome, level of functioning, and
previous history of sexual abuse (Goldner, Srikameswaran, Schroeder, Livesley, &
Birmingham, 1999; Strober, 1983; Thompson-Brenner & Westen, 2005; Westen &
Harnden—Fischer, 2001).

Strober (1983) conducted the first study examining personality clusters in women
with eating disorders by using a relatively circumscribed approach that included young
women with AN who participated in a treatment program. Minnesota Multiphasic

Personality Inventory (MMPI; Hathaway & McKinley, 1943) profiles of 130 women with

AN were cluster analyzed, and 82% of these patients fit into one of three personality
clusters. The first group consisted of 38% of women who demonstrated minimal levels
of psychopathology. This cluster only evidenced elevated levels of need for approval and
control but not other pathology such as affective or pervasive characterological
disturbances. The second group was made up of 28% of women who had high scores on
neuroticism and interpersonal difficulties. Also, this group showed more disturbances in
rigid control of impulses than the first cluster. The third group consisted of 15% of
women who showed severe levels of depression, impulsivity, and overall serious
psychopathology.

These findings are intriguing in that three distinct clusters were delineated in
women with AN only, thereby supporting the notion of heterogeneity in personality for
eating disordered women within the same diagnostic category. In addition, these
personality clusters were shown to have clinical value by predicting treatment outcome;
the first cluster of patients showed the fewest eating disordered symptoms at three-month
follow-up, while the presence of these symptoms increased linearly across the second and
third clusters. Overall, this early attempt at personality clustering was significant because
it suggested that heterogeneous personality clusters in AN could reliably predict
differences in treatment outcome. Nevertheless, findings were limited by the inclusion of
only AN patients in the study, which prevented a finer analysis of various eating
disordered profiles.

Extending the cluster method to a group of women with varied eating pathology,
Goldner et al. (1999) used the Dimensional Assessment of Personality Pathology — Basic

Questionnaire (DAPP—BQ; Livesley, Jackson, & Schroeder, 1992) to measure personality

groupings. The DAPP-BQ is a self-report instrument that measures 18 factor-derived
dimensions of personality disorder and has been used in several studies with eating
disordered patients (Leonard, Steiger, & Kao, 2003; Steiger, Jabalpurwala, Champagne,
& Stotland, 1997; Steiger, Stotland, Ghadirian, & Whitehead, 1995; Steiger, Stotland,
Trottier, & Ghadirian, 1996). Goldner et a1. (1999) administered the DAPP-BQ to 136
women with eating disorders of which 18 (13%) were diagnosed with AN restricting
type, 19 (14.7%) with AN binge-eating/purging type, 84 (62%) with BN, and 15 (11%)
with eating disorder not otherwise specified.

To analyze DAPP-BQ scores of women with eating disorders, Goldner et a1.
(1999) first conducted an unweighted least squares factor analysis for the 18 personality
dimensions. This analysis revealed five factors accounting for 73.1% of the variation of
DAPP-BQ scores. These factors were: Neuroticism, Psychopathy, Compulsivity,
Interpersonal Difficulties, and Behavioral Disturbance. For Neuroticism, the DAPP-BQ
scales Anxiousness, Submissiveness, Social Avoidance, Identity Problems, Affective
Lability, Insecure Attachment, Oppositionality, and Cognitive Distortion loaded on this
factor. Women scoring high on these scales showed a general unhappiness regarding
themselves and their life, and they experienced elevated levels of anxiety. On the factor
Psychopathy, there were significant loadings for the scales Rejection, Callousness, and
Narcissism. Women scoring high on these scales demonstrated anger, an absence of
interest in others’ concerns, and a need for attention from others. For Compulsivity, the
scales Oppositionality and Compulsivity loaded on this factor, with high scores indicating
compulsive behavior. On the fourth factor, Interpersonal Difficulties, there were loadings

for the scales Social Avoidance, Intimacy Problems, and Restricted Expression. High

scorers had difficulty expressing their feelings and forming close personal relationships.
Finally, the DAPP—BQ scales Stimulus Seeking, Self-Harming Behaviors, Cognitive
Distortion, and Conduct Problems loaded on the Behavioral Disturbance factor. Women
who scored high on these scales showed impulsivity and a lack of social conformity.

Following this factor analysis, Goldner et al. (1999) applied cluster analysis using
Ward’s method (Ward, 1963) to the factors and obtained a three cluster solution that fit
that data better than a two, four, or five cluster solution. These investigators classified
100% of patients into one of three personality clusters that appear to be remarkably
similar in content to those found by Strober (1983), mentioned above. The first cluster
was defined as the “Mild” group (renamed Adaptive in the current study), consisting of
32.4% of patients (breakdown of AN and BN distribution not given), and was made up of
women who had low levels of personality pathology. However, these women displayed
an increased need for approval, elevated anxiety, and difficulties with separation and loss
compared to a general population sample. The second cluster was identified as the
“Rigid” group, consisting of 49.3% of the entire sample and 73% of AN patients and
42% of BN patients. Women in this cluster were characterized by high levels of
interpersonal difficulties and rigid control of impulses compared to the other two groups.
The final cluster was labeled as the “Severe” group (renamed Dysregulated in the current
study) which comprised 18.4% of the sample (AN and BN breakdown again not given).
These women showed the highest overall level of personality pathology including
elevated scores on neuroticism, impulsivity, and acting-out behaviors. This personality
profile is similar to those of the DSM-IV Axis II Cluster B disorders (American

Psychiatric Association, 2000), particularly borderline personality disorder. This finding

is notable given that a significant minority of eating disorder patients, particularly BN
patients, have comorbid Cluster B disorders which tend to predict poorer treatment
outcome (Herzog, Keller, Sacks, Yeh, & Lavori, 1992).

Overall, Goldner et a1. (1999) expanded on the work of Strober (1983) by
examining personality clusters for a range of eating pathology, with the resulting clusters
resembling those previously found. One limitation of this study was that outcome
variables were not assessed, thus making the clinical utility of these clusters unclear.
Nevertheless, Goldner et a1. (1999) point out that the clusters classify women with eating
disorders according to their level of personality pathology, with the clusters increasing in
severity from the “Mild” to “Moderate” to “Severe” clusters. This classification system
may prove useful for tailoring specific treatments for each cluster. For example,
members of the severe cluster may require and benefit from longer-term and intensive
treatments such as psychodynamic therapy or dialectical behavior therapy (Kemberg,
1995; McCabe & Marcus, 2002).

Westen and Harnden-Fischer (2001) were interested in further investigating
personality clusters and their ability to predict level of functioning by using informants
other than the patients themselves. Psychiatrists and psychologists were used as
informants in this study due to their putative expert knowledge about their patients, and
because the researchers felt that patients’ knowledge of their own difficulties and
psychological processes may be limited. Clinicians rated their AN or BN patients using
the SWAP-200 (Westen & Shedler, 1999a; Westen & Shedler, 1999b), a Q-sort measure
that allows for the assessment of a wide range of personality traits and the generation of

specific personality clusters for a given disorder. In terms of lifetime diagnoses, 17

10

patients had a history of restricting type AN, 49 patients had a history of BN, and 35
patients had a history of both AN and BN (ANBN; lifetime diagnoses were not available
for 2 patients).

Based on their SWAP-200 scores, approximately two-thirds of the patients (n =
67) were clustered into one of three groups labeled hi gh-functioning/perfectionistic,
constricted/overcontrolled, and emotionally dysregulated/undercontrolled. These three
clusters again appear to be generally consistent with ones found by Strober (1983) and
Goldner et al. (1999). The first group (i.e., cluster 1) consisted of 26% of eating
disordered women who evidenced the least personality pathology of the three groups;
they demonstrated a number of healthy characteristics, including empathy and
conscientiousness. However, these women were also characterized by feelings of anxiety
and guilt and were prone to perfectionism. In the second group (i.e., cluster 2; 22% of
the women), higher levels of personality pathology and emotional and interpersonal
constriction were present. The final cluster (i.e., cluster 3) consisted of 10% of patients
who had strong, rapidly shifting emotions and impulsive behaviors. They showed
tendencies toward anger, suicidal thoughts and behaviors, and unstable interpersonal
relationships. Interestingly, lifetime eating disorder diagnoses generally cut across these
three clusters. Overall, women with AN tended to aggregate in clusters 1 (18.5% of AN
sample) and 2 (34.8%), while BN and ANBN women were found in all 3 clusters (59.3%,
21.7%, 50%, respectively for BN; 22.2%, 43.5%, 50%, respectively for ANBN).

Similar to Strober (1983), Westen and Harnden-Fischer (2001) also found these
personality clusters to have important clinical utility. Specifically, the three groupings

significantly predicted eating disorder symptoms (i.e., current weight, binge frequency,

11

purge frequency) above and beyond the variance accounted for by DSM-IV eating
disorder diagnoses.

In addition, personality clusters significantly predicted level of psychosocial (i.e.,
interpersonal relationships, occupational functioning) and overall clinical functioning.
DSM-IV Global Assessment of Functioning Scale (GAF) scores were used to measure
psychosocial and overall functioning, whereas history of hospitalization was used to
assess overall clinical functioning. Correlational analyses revealed that GAF scores were
positively correlated with cluster 1 scores and negatively correlated with cluster 3 scores;
cluster 2 scores were not significantly correlated with GAF. This suggests that cluster 1
patients function relatively well in social, occupational, and psychological areas, whereas
cluster 3 patients evidence fairly severe impairments in these areas. Correlational
analyses for history of hospitalization revealed a similar finding, namely that history of
hospitalization (rated yes/no) was negatively associated with cluster 1 scores and
positively associated with cluster 3 scores; cluster 2 scores were not significantly
correlated with hospitalization. Despite the finding that GAF scores were not
significantly correlated with cluster 2 scores, patients in cluster 2 nevertheless had mean
GAF scores that fell between those from cluster 1 and 3. This suggests a trend for
increasing severity of personality pathology and functional impairment across clusters 1-
3. Irnportantly, DSM-IV diagnoses were not significantly related to levels of functioning
in any of the analyses of GAF scores or history of hospitalization.

Another unique finding of this study was that personality clusters predicted sexual
abuse history, a variable shown previously to be related to eating disorders development

(Wonderlich, Brewerton, Jocic, Dansky, & Abbott, 1997). In clusters 1 through 3, the

12

percentage of women reporting a history of sexual abuse was 5.3%, 37.5%, and 83.3%,
respectively. As with results from level of functioning, eating disorder diagnoses failed
to significantly predict sexual abuse.

Based on all of these findings, Westen and Harnden-Fischer (2001) argue that
personality is a critical component to consider in conjunction with traditional DSM-IV
eating disorder diagnoses. These diagnoses alone failed to capture the heterogeneity in
personality that exists for both AN and BN, such that individuals with markedly different
personality profiles are included in the same eating disorder category. Eating disorder
diagnoses were also unsuccessful in accounting for heterogeneity in important clinical
variables including level of functioning and sexual abuse. These investigators suggest
that future research with personality clusters should explore connections between other
etiological factors and personality clusters.

Recently, Thompson-Brenner and Westen (2005) attempted to replicate and
expand on findings with the personality clusters studied by Westen & Harnden-Fischer
(2001). These researchers tested for the presence of the three clusters in 145 patients
with bulimic symptomatology by having clinicians rate their patients with a questionnaire
rather than the SWAP-200. This clinician-report questionnaire included three paragraphs
that were created by combining items with the highest factor scores from each of the
three personality groupings generated from the SWAP-200. Results showed that 84% of
the patients fit into one of the three personality groupings, thereby providing additional
support for the three personality groupings in women with eating pathology. Moreover,
findings replicated previous results indicating that personality clusters were associated

with level of functioning and history of sexual abuse (Westen & Harnden-Fischer, 2001).

13

The investigators also extended their previous work by showing that clusters predicted
treatment length (i.e., number of weeks) and outcome (i.e., level of eating disorder
symptoms and GAF scores at treatment termination). Treatment length was negatively
associated with cluster 1 (high-functioning/perfectionistic) and positively associated with
cluster 3 (emotionally dysregulated/undercontrolled) scores, while treatment outcome
was positively associated with cluster 1 and negatively associated with cluster 3 scores.
Cluster 1 women showed the fastest recovery (i.e., recovering approximately after 51
weeks of treatment), while cluster 2 women (constricted/overcontrolled) recovered
approximately five months later (i.e., after 73 weeks of treatment), and cluster 3 women
recovered after an additional five months (i.e., after 92 weeks of treatment). The
percentage of patients who recovered while in treatment in clusters 1 through 3 was 62%,
50%, and 43%, respectively.

Overall, findings from all of the studies reviewed above converge in suggesting
the presence of three personality clusters in women with eating disorders that are similar
across studies and informants. The first cluster of patients appears to function relatively
well overall, despite some difficulties with anxiety and need for approval. The second
cluster has been found to have more psychological and behavioral problems than the first
cluster, particularly difficulties with interpersonal problems, rigid control of impulses,
and restricted affect. The final cluster appears to have the most severe levels of
personality pathology that is characterized by impulsivity, unstable interpersonal
relationships, and high levels of neuroticism. The cluster labels applied to these groups

by Goldner et a1. (1999) seem particularly apt in highlighting the degree of personality

14

pathology present in each cluster, with the pathology ranging from mild to moderate to
severe.

Regardless of the strength of these studies in finding consistent clusters, they are
limited by four factors. First, studies have not included a large proportion of women with
subclinical eating pathology or EDNOS even though subclinical cases of eating disorders
have been shown to be much more prevalent than clinical cases (Heatherton, Nichols,
Mahamedi, & Keel, 1995; Mintz & Betz, 1988). For example, prevalence rates for
clinical cases of AN and EN are .5%-3%, respectively (American Psychiatric
Association, 2000; Hsu, 1996), whereas the prevalence of subclinical eating pathology in
college students ranges from 11% to 27% (Heatherton et al., 1995; Mintz & Betz, 1988).
Furthermore, the majority of women with eating disturbances who seek treatment have
this type of subclinical eating pathology or EDNOS (Andersen, Bowers, & Watson,
2001). These findings suggest that a substantial number of women with eating pathology
of clinical significance have not yet been characterized.

Second, across many of the studies (Strober, 1983; Thompson-Brenner & Westen,
2005; Westen & Harnden-Fischer, 2001), a significant percentage of patients (16%-35%)
did not fall into any of the three clusters. Unfortunately, the personality and eating-
related characteristics that may have set these women apart were not examined.
Investigating the personality and eating characteristics of this group may shed light on
specific personality traits that are only relevant to a small, but significant, group of
women with eating disorders. For example, a woman with an eating disorder may have
only evidenced difficulties with perfectionism, and because she did not show other traits

necessary to be included in a cluster, she remained unclassified.

15

Third, the clinical utility of the clusters has been studied less extensively, with the
notable exception of the Westen and colleagues studies (Thompson-Brenner & Westen,
2005; Westen & Harnden-Fischer, 2001). However, these studies used fairly crude
indicators of psychosocial functioning. One of these measures, the GAF, is limited in
that it is a single-item scale that has been shown to have significant limitations in
reliability and validity (Bacon, Collins, & Plake, 2002). Finally, little work has examined
how factors that have been shown previously to be related to eating disorders influence
the three personality clusters. With the exception of sexual abuse, no other factors have
been investigated despite the fact that other factors likely also contribute to the clusters.
One such factor may be family relationships given their theoretical and empirical

associations with both eating disorders and personality.

Family Relationships, Eati_ng Disordersind Personalitl

Very little empirical work has explored associations between personality and
family relationships in women with eating disorders, despite the fact that family factors
have been shown to contribute to the development and maintenance of these disorders
(Strober & Humphrey, 1987). Indeed, most psychological theories, including
psychodynamic and cognitive-behavioral (Beck, 1995; Westen, 1998) posit a role for the
family in the development and continuity of personality and psychopathology, and thus it
is probable that family factors may be associated with the personality clusters observed in
eating disordered patients. Specifically, typologies of family relationships may

correspond with distinct personality clusters.

16

Theorists have proposed several types of family relationships that may lead to
eating pathology. Bruch (1973) believed that AN results from a family system marked
by parental overinvolvement and enmeshment, in which the girl with AN is unable to
successfully separate from and establish an individual identity. Minuchin, Rosman, and
Baker (1978) also argued that AN develops in families that display enmeshment, rigidity,
overprotectiveness, conflict avoidance, and poor conflict resolution. Regarding BN, it
has been hypothesized that these women’s families tend to be chaotic, hostile, and
lacking in nurturance (Humphrey, 1986a).

Despite the abundance and richness of these theories, relatively few empirical
studies have examined family relationships in eating disordered women. Most of these
studies have used the Parental Bonding Instrument (PBI; Parker, Tupling, & Brown,
1979) or the Structural Analysis of Social Behavior (SASB; Benjamin, 1974) to measure
disturbed family relationships in women with AN or BN. In general, work with the PBI
has suggested that women with AN and women with BN recall less maternal and paternal
care than control women. For example, Pole, Waller, Stewart, & AParkin-Feigenbaum
(1988) found women with BN remembered their mothers as less caring compared to
control women. Women with BN also remembered their fathers as less caring, although
this finding was only a statistical trend. In another study, Steiger, Van der Feen,
Goldstein, and Leichner (1989) found that women with AN and women with BN both
remembered their fathers as less caring than controls. Nevertheless, no differences
among groups were found for maternal care. Palmer, Oppenheimer, and Marshall (1988)
found that AN and BN women remembered their mothers as less caring than controls, but

that only BN women remembered their fathers as less caring. Finally, in a study of only

17

AN women, it was shown that they recalled less maternal and paternal care than control
women (Bulik etal., 2000).

Overall, the studies using the PBI yielded consistent findings suggesting that both
AN and BN women recall their parents as less caring and nurturing. Notably, none of
these studies found that women with AN and BN remembered their parents as
overprotective, despite previous theories that had suggested this factor would be involved
in eating disorders (Bruch, 1973; Minuchin et al., 1978). Finally, similar to the
personality cluster studies that cut across eating disorders, the PBI studies indicated that
deficits in perceived parental care cut across AN and BN rather than being specific to one
diagnostic type.

Family relationships of AN and BN women have also been measured with the
SASB, which is a circumplex model of interpersonal relationships and their intrapsychic
representations. The SASB is a flexible measure in that it can be used to assess any given
relationship of interest. For example, it is possible to measure the subject’s perception of
any significant other, including mothers, fathers, step-parents, or other caregivers. The
wording of the measure is easily adapted according to the relationship being rated. For
each relationship rated, the SASB requires that both aspects of the relationship be
measured, namely 1) how the subject views the specified significant other, and 2) how
the subject views him or herself in relation to that other person. The SASB provides a
richer analysis of complex family relationships than the PBI by assessing a greater
number of family relationship dimensions, clusters, and combinations of relationships.
This measure also assesses intrapsychic dimensions, which provide data on whether

subjects perceive themselves in a way that is consistent with their perceptions of how

18

they were treated by their parents. For example, a daughter may feel self-destructive
following sexual abuse by a parent (Humphrey & Benjamin, 1986).

In one of the first studies to use the SASB, Humphrey (1986b) found that BN but
not AN women experienced greater disturbances in parental nurturance and empathy than
control women. BN women’s disturbances in nurturance and empathy were found more
frequently in relation to their fathers than their mothers. On other relationship variables,
both AN and BN women reported experiencing higher levels of parental blaming and
sulking, attacking and withdrawing, neglecting and walling off, and lower levels of
parental helping and trusting than control women (Humphrey, 1986b). Therefore, AN
and BN women perceived relationships with parents to be marked by hostility coupled
with no support. In terms of intrapsychic dimensions, or subjects’ experiences of
themselves, findings matched those of perceived parental relationships. Specifically, AN
and EN women reported experiencing disturbances in self-nourishing and cherishing, and
self-accepting and exploring, compared to control women. Moreover, eating disordered
women were more self-oppressing, self-rejecting, self-neglecting, and less self-protecting
than controls. These findings again are at odds with theories that advance parental
hostility as a characteristic found in BN, but not AN women. Instead, findings with the
SASB point to increased levels of parental hostility and lack of parental support in both
AN and BN families, further suggesting that certain troubled parental relationships cut
across AN and BN. The only exception to this was with the factor parental nurturance
and empathy, where deficits were specific to BN.

Two recent studies (Friedman, Wilfley, Welch, & Kunce, 1997; Wonderlich,

Klein, & Council, 1996) employing the SASB with BN women have partially replicated

l9

Humphrey’s (1986b) results. Both of these studies used a short form of the SASB and
chose to only assess dimensions measuring hostile and disengaged farrrily relationships
and the corresponding intrapsychic component, namely whether subjects also perceived
themselves in a hostile manner. Similar to the findings of Humphrey (1986b), BN
women perceived their family as more hostile than control women and also directed more
hostility towards themselves than did controls.

In summary, research investigating family relationships in eating disordered
patients with the PBI revealed that AN and BN women recalled their mothers and fathers
as less caring than controls. Studies using the SASB found that BN women perceived
greater deficits in parental nurturance and empathy than control women, and this finding
was more robust for women’s relationships with their fathers. Additionally, both AN and
BN women perceived more hostility and less support in their parental relationships than
controls. Finally, eating disordered women viewed themselves with the same hostility
that they experienced from their parents.

Across these studies, a common theme emerged, namely that more similarities
than differences in disturbed family relationships were found between AN and EN
women. This offers evidence that family relationships, like personality clusters, may cut
across diagnostic boundaries and be related to heterogeneity within, rather than across,
diagnostic subtypes. These findings also raise the question of whether specific family
relationships may be typologies of personality clusters.

Unfortunately, only one study has examined this possibility. In the Strober (1983)
study described above, family stability was rated based on data obtained from a

structured interview with AN patients and their parents. Ratings were limited to a one-

20

item measure that had a 4—point scale ranging from “close and emotionally satisfying
intrafarnilial relations” to “extreme family disharmony and marital discord.” Women in
the first cluster (i.e., minimal personality pathology) reported family relationships
described as relatively stable and emotionally fulfilling. By contrast, women in the
second cluster (i.e., high neuroticism and social avoidance) had families that were less
stable with more family disharmony than the first and third clusters. The third cluster
(i.e., severe personality pathology) had families with the most unstable and
disharmonious relationships and greater marital discord than the other two clusters.

These findings are significant in suggesting that family functioning is related to
personality clusters in women with eating disorders; indeed, the degree of troubled family
relationships seemed to parallel the increase in personality pathology across clusters.
Nonetheless, only one dimension of family relationships was examined in this study, and
this dimension did not include the factors mentioned above that were previously shown to
be associated with eating disorders (e. g., parental care, nurturance, hostility).

Previous research on women with both personality disorders and eating disorders
provide insight into family relationship typologies that might be found to be related to
each personality cluster, especially the Dysregulated cluster. For example, Wonderlich
and Swift (1990) compared eating disordered women with no personality disorders, those
with borderline personality disorder (BPD), and those with other personality disorders
and found that BPD women viewed their parents as significantly more attacking and
withdrawn than the other two groups of women. Similarly, Johnson, Tobin, and Enright

(1989) compared eating disordered women with BPD to those without BPD and found

21

that BPD women perceived their families as less expressive, more controlling, and more
hostile.

In general, these studies found the most disturbed family relationships in women
with BPD, somewhat disturbed family relationships in women with other personality
disorders, and no clear signs of disturbed family relationships in women with no
personality disorders. Other research investigating women diagnosed with BPD but not
eating disorders has shown these women to have experienced greater parental neglect
(Gunderson, Kerr, & Englund, 1980), maternal overinvolvement, paternal
underinvolvement (Soloff & Millward, 1983), less parental care, and more
overprotectiveness in general (Goldberg, Mann, Wise, & Segall, 1985) compared to
patients with a range of psychiatric disorders and control patients. These findings suggest
that women with BPD alone experience marked family disturbances. Taken a step
further, the findings also suggest that BPD may significantly account for the more severe
family deficits noted in women with comorbid eating disorders than the presence of the
eating disorders alone. Given the fact that women in the eating disordered Dysregulated
personality cluster resemble women with BPD, it may be that women in this cluster have
a comorbid diagnosis of BPD and perceive their families as problematic in areas such as
care, nurturance and empathy, and hostility. Extrapolating from the studies described
above, women in the Rigid cluster who primarily resemble women with personality
disorders other than BPD will likely evidence disturbed family relationships, but to a
lesser degree than women in the Dysregulated cluster. Finally, women in the Adaptive
cluster would be predicted to show the least personality pathology and accordingly the

fewest, if any, disturbed family relationships.

22

Specific AimL and Hypotheses
Overall Purpose of Current Study

The present study sought to extend previous research on personality clusters in
eating disorders by examining these clusters in undergraduate women with a range of
subclinical and clinical eating pathology. If indeed the same personality clusters are
observed in women with subclinical eating pathology, and they predict level of
functioning, this would suggest that these clusters are universal in all eating disordered
patients and may be more robust indicators of the severity and phenomenology of eating
disorders than the DSM-IV. To this point, if subclinical women fall into the same
clusters as clinical women, the DSM-IV categories of AN and BN that are conceptualized
as representing more severe forms of eating pathology than EDNOS would not appear to
accurately reflect this distinction in severity.

Two groups of women with largely subclinical eating pathology were selected to
resemble the pathology present in AN and BN. The first group consisted of women who
restrict food but who do not binge and purge (i.e., R group), while the second group was
made up of women who binge and purge (i.e., B/P group) and who may also restrict their
food intake. These groups roughly correspond to the diagnoses of Restricting Type AN
(RAN), and BN or the Binge-Eating/Purging Type of AN, respectively. A third group of
women without eating pathology was included as a control group (i.e., CON group). This
group was selected to not include women with elevations on any eating disordered
symptoms, as most studies in eating disorders research have commonly examined control
women who evidenced no eating pathology (Casper et al., 1992; Fassino et al., 2002;

Kleifield et al., 1994; Steiger et al., 1997). Examining pure groups allows for analysis of

23

whether levels of specific variables (e. g., personality pathology, family relationships,
psychosocial functioning) are associated with the presence or absence of eating
pathology. Similar to Goldner et a1. (1999), the DAPP-BQ was used in the current study
to assess personality because it provides a more comprehensive analysis of personality
than other measures (e.g., SWAP-200) by allowing for factors as well as clusters to be
generated from its scores.

The current study also extended previous research by assessing the clinical and
psychosocial relevance of personality clusters and by comprehensively examining family
factors as typologies across clusters. Psychosocial and overall level of functioning was
assessed with the Social Adjustment Scale Self-Report (SAS-SR; Weissman & Bothwell,
1976) and self-reported history of treatment. The SAS-SR appears to be a more reliable
and valid measure of functioning than the GAP used in previous cluster studies (Westen
& Harnden-Fischer, 2001; Thompson-Brenner & Westen, 2005). In addition, the SAS-
SR assesses functioning across six areas with multiple items measuring each area,
whereas the GAP consists of only a single-item global scale. A self-reported history of
treatment (e.g., psychotherapy, medication, and hospitalization) was also used to measure
level of functioning. Perkins, Klump, Iacono, and McGue (2005) have shown that
treatment-seeking AN women exhibit higher levels of personality disturbances than non
treatment-seeking AN women, suggesting that history of treatment differentiates levels of
functioning. Furthermore, evidence for this measure has been demonstrated by studies
showing an association between women with a positive history of hospitalization and a

higher level of eating disordered symptoms compared to women never hospitalized, who

24

showed fewer symptoms (Milos, Spindler, Buddeberg, & Ruggiero, 2004; White &
Litovitz, 1998).

Finally, the present study sought to delineate relationships between personality
clusters and family functioning in order to advance knowledge of the types of family
factors important to explore in treatment. Associations between family functioning and
personality clusters were examined using the PBI and SASB. The decision to include
both measures was based on the fact that both have been used extensively by eating
disorder researchers in the past and each contributes unique information about family
relationships. In addition, the SASB assesses more complex family relationships and
intrapsychic dimensions not measured with the PBI. Thus, using both measures in the

current study allowed for a more comprehensive analysis of family relationships.

Specific Aims:

The current study consisted of the following specific aims:

Specific Aim #1: The first aim of the study was to replicate the three personality
clusters found previously by Goldner et a1. (1999) across the disordered eating groups (R
and B/P) using the DAPP-BQ and confirmatory factor analysis and cluster analysis.

I Hypothesis #1 a; Confirmatory factor analysis will yield the same five
initial factors of Goldner et a1. (1999) (Neuroticism, Psychopathy,
Compulsivity, Interpersonal Difficulties, and Behavioral Disturbance)
across R and B/P women.

' Hypothesis #lb: Cluster analysis of these factors will reveal three

distinct personality profiles derived from the R and B/P groups

25

corresponding to the “Mild,” “Rigid,” and “Severe” clusters found by
Goldner et al. (1999).

Specific Aim #2: The second aim of the current study was to assess whether R
and B/P women fall into more than one personality cluster and whether clusters cut
across the R and B/P groups.

I Hypothesis #2: Personality will be heterogeneous in the two
disordered eating groups such that R and B/P women will fall into
more than one cluster. It is expected that the majority of R women
will fall into the Adaptive and Rigid personality clusters, whereas B/P
women will fall into the Adaptive, Rigid, and Dysregulated clusters.

Specific Aim #3: The third aim was to determine if personality clusters formed
from the disordered eating groups significantly predict women’s level of clinical and
psychosocial functioning better than the original disordered eating groups.

' Hypothesis #3: Personality clusters will significantly predict SAS-SR
scores and treatment history. The initial eating categories are expected
to predict less of the variance in these functioning variables than the
personality clusters.

Sugecific Aim #4: The final aim of the present study was to use the PBI and SASB
to examine farrrily relationships across the three personality clusters of women with
eating disturbances as well as control women with no eating pathology.

I Hypothesis #4: The level of severity across personality clusters will
parallel the level of disturbed family relationships. In short, control

women will show the least family disturbances, and the degree of

26

disturbances will increase across the Adaptive, Rigid, and
Dysregulated clusters. Members of the Adaptive personality cluster
will be characterized by less pathological family relationships that
resemble relationships of the control group. The Dysregulated cluster
will consist of women with the most family disturbances, including an
absence of parental nurturance, empathy, and support, along with
increased hostility, as measured by the SASB. These women will also
recall their mothers and fathers as less caring, as measured by the PBI.
In terms of differential parental relationships, it is expected that
Dysregulated women will view their fathers as lacking in nurturance
and empathy significantly more than their mothers. Overall, it is
predicted that the Dysregulated cluster will only differ from the Rigid
cluster in the degree of family disturbances, with the latter cluster
experiencing fewer disturbances than the Dysregulated cluster but
more disturbances than the Adaptive cluster. Finally, it is
hypothesized that members of the Rigid and Dysregulated clusters will
endorse intrapsychic disturbances matching existing parental

disturbances.

Exploratory Aim: Given the present study’s hypothesis that three distinct
personality clusters of women with eating disturbances will emerge, a corollary to this is
to examine the number and types of personality clusters present in the control group of

women who evidence no eating pathology. Previous research points to three personality

27

types commonly found across studies of large samples of individuals who may or may
not have been diagnosed with psychiatric disorders (Asendorpf, Borkenau, Ostendorf, &
van Aken, 2001; Block & Block, 1980; Caspi & Silva, 1995; Rammstedt, Riemann,
Angleitner, & Borkenau, 2004; Robins, John, Caspi, Moffitt, & Stouthamer-Loeber,
1996). These three types have been previously labeled Resilients, Overcontrollers, and
Undercontrollers (Robins et al., 1996), and these labels are predicated on the theory of
two components of personality, ego-control and ego-resiliency, developed by Block and
Block (1980). Ego-control refers to a person’s ability to modulate his or her impulses,
emotions, and desires. Overcontrollers are individuals who exhibit increased ego control
characterized by rigid control over their emotions and impulses, and general inhibition.
By contrast, Undercontrollers are people who express their impulses and emotions freely,
tending to act spontaneously and often impulsively. Unlike Overcontrollers and
Undercontrollers, the Resilient personality type represents individuals who are capable of
modulating their emotions and impulses appropriately in given situations, regardless of
whether they typically tend to over- or undercontrol their emotions and impulses. Thus,
this group is able to adapt to various situational demands and respond effectively.
Previous research has not examined these three personality types in a sample of
non-clinical and eating disordered women. Therefore, it is unclear whether the same
personality types found in studies of eating disordered women will also emerge in a non—
clinical sample. Westen and Harnden-Fischer (2001) drew comparisons between
personality clusters obtained in their study and the Resilient, Overcontroller, and
Undercontroller types. These researchers noted that their constricted/overcontrolled

cluster resembled Overcontrollers and that their emotionally

28

dysregulated/undercontrolled cluster resembled Undercontrollers. Regarding ego-
resiliency, Westen & Harnden-Fischer (2001) argue that the hi gh-
functioning/perfectionistic cluster resembles Resilients, but it was also mentioned that the
high-functioning/perfectionistic cluster contained women who were more distressed than
the typical individual in the Resilient group. It is likely that this same pattern is true with
the other clusters as well. That is, both the constricted/overcontrolled and emotionally
dysregulated/undercontrolled clusters probably represent women with higher or lower
levels of ego control, respectively, than Overcontrollers and Undercontrollers, due to the
severity of eating pathology in the clusters. This issue will be examined directly in the
present study by determining if personality clusters found in eating disordered women
and control women mirror each other along the dimensions of ego-control and ego-
resiliency, but only differ on these constructs by degree of severity.

The control group in the current study will be cluster analyzed, and it is predicted
that three personality clusters approximating Resilients, Overcontrollers, and
Undercontrollers will be found. These clusters will then be compared across level of
functioning (i.e., SAS-SR and treatment history scores) and family relationship variables
(i.e., PBI and SASB scores). The following predictions are posited. First, it is predicted
that the Resilient cluster in the current study will exhibit healthy levels of functioning and
no personality pathology. Second, it is hypothesized that the Overcontrollers and
Undercontrollers will display deficits in their level of functioning and that the former
group will also show elevated levels of interpersonal difficulties and rigid control of
impulses. The Undercontrollers are expected to report the most severe levels of

personality pathology, characterized by impulsivity, unstable interpersonal relationships,

29

and high levels of neuroticism. Finally, because personality pathology is associated with
greater family disturbances (Wonderlich & Swift, 1990), both the Undercontrollers and
Overcontrollers are expected to perceive difficulties in their familial relationships,
whereas the Resilients will endorse stable and satisfying relationships within their

families.

30

METHOD

Participants

Seven hundred ninety-five female undergraduate students at Michigan State
University (MSU) were recruited from the Department of Psychology Subject Pool.
Participants provided informed consent and completed measures on-line through the
secure Subject Pool website. Students participating in the Subject Pool are able to access
this website by first entering a specific access ID number followed by their User ID and
password. Once on the website, students can select experiments they would like to
participate in. No identifying participant information is linked with the data, as
participants’ User IDs for on-line experiments are not matched with questionnaire
responses; only identification numbers assigned by the website are matched with data.
Participants received one course credit for every half-hour devoted to completing the
experiment. Approximate completion time for the current study was 1.5 hours.

After questionnaires were completed, participants were assigned to one of the
three study groups mentioned previously (i.e., R, B/P, CON) based on the severity and
type of eating pathology endorsed (see below). Of the original 795 women that
participated in the study, 20 (2.5%) were included in the R Group, 62 (7.8%) met criteria
for the B/P group, and 109 (13.7%) fell in the CON group. The remaining 604 (76.0%)
women were not examined in analyses because they did not meet inclusion criteria for

any of the three groups.1

 

' However, these women were clustered and compared to the Adaptive. Rigid, and Dysregulated clusters;
results are presented on p. 70.

31

Of the 191 women included in analyses, participant age ranged from 18 to 25
years, with a mean age of 19.19 (SD = 1.32). Regarding year in school, 48.7% of women
were freshmen, 28.3% were sophomores, 13.1% were juniors, and 9.9% were seniors.
The majority of women were Caucasian (86.4%), while the remainder of the sample was
African-American (6.3%), Asian (4.2%), Hispanic (1.6%), or “Other” (1.6%). To assess
socioeconomic status (SES), the average annual combined income of participants’
parents was exarrrined. Most of the participants reported being in the highest SES bracket
(37.7%; over $100,000), followed by 31.9% ($60,000-$100,000), 17.3% ($40,000-

$60,000), 9.9% ($20,000-$40,000), and 3.1% (under $20,000) in other SES categories.

Measures
Demoggrthic Information

A general demographic questionnaire was used to collect information regarding
age, ethnicity, socioeconomic status, and self-reported height and weight. Height and
weight values were used to calculate Body Mass Index (BMI) in the current study
(Weight in pounds / (Height in inches) X (Height in inches)) X 703.
Disordered Eating Measures and Classifications
Measures

Eating Disorder Inventory-2 (EDI-2): The EDI-2 (Garner, 1991) is a 91-item self-
report measure designed to assess psychological and behavioral traits associated with AN
and BN. The EDI-2 consists of eleven subscales including Drive for Thinness, Bulimia,
Body Dissatisfaction, Ineffectiveness, Perfectionism, Interpersonal Distrust, Interoceptive

Awareness, Maturity Fears, Asceticism, Impulse Regulation, and Social Insecurity.

32

The first three of these subscales measure core disordered eating attitudes and
behaviors, whereas the last eight assess personality traits believed to be associated with
eating pathology. Given that the EDI-2 was used in the current study primarily for
classifying women according to disordered eating behaviors, only the Drive for Thinness
(i.e., excessive preoccupation with dieting, weight, and weight gain), Bulimia (i.e.,
thoughts about binge eating and the tendency to engage in bingeing behaviors) and Body
Dissatisfaction (i.e., dissatisfaction with the shape and size of one’s body) subscales were
used in analyses. Only three items (5, 28, 46) from the Bulimia subscale were used for
classification, as these are the only items that ask about actual bingeing behaviors; the
remaining items on this subscale focus on thoughts about bingeing or other behaviors that
do not necessarily indicate bingeing. Respondents rate each EDI-2 item on a six-point
scale, ranging from “always” to “never.” Higher scores indicate more disturbances in
disordered eating attitudes and behaviors.

The EDI-2 shows good internal consistency within eating disordered samples with
alphas ranging from .83 to .93 for the original eight subscales (Garner & Olmsted, 1984),
and .83 to .92 for the three subscales included in this study (Garner, 1991). Internal
consistency is also adequate within community samples (alphas = .79 to .92; Raciti &
Norcross, 1987), albeit slightly lower than in clinical populations. One-week (.67-.95),
three-week (65-97), and one-year (.41-.75) test-retest reliabilities have been found to be
nioderate-to-high in undergraduate college and community samples (Welch, 1988; Wear
& Pratz, 1987; Crowther, Lilly, Crawford, Shepherd, & Oliver, 1990). The validity of
the EDI-2 has been supported by studies indicating that items discriminate between

eating disorder and non-eating disorder subjects (Garner, 1991).

33

Bulimia Test-Revised (BULIT-R): The BULIT-R (Thelen, Farmer, Wonderlich, &
Smith, 1991) is a 36-item self-report measure designed to assess BN, as defined by DSM-
IV criteria. Thus, the BULIT-R measures binge eating, compensatory behaviors, and
weight and shape concerns. Respondents rate each item on a 5-point scale, with higher
scores indicating greater eating disturbances. A total score on the BULIT-R is generated
by summing 28 of the 36 items. The eight items not included in the total score assess
compensatory behaviors (e. g., laxatives, diuretics, fasting, exercising) that do not occur
as frequently as vomiting. Psychometric studies of the BULIT-R have demonstrated it
has good internal consistency within BN and control populations, with alphas ranging
from .92 to .98 (Brelsford, Hummel, & Banios, 1992; Thelen, Mintz, & Vander Wal,
1996). The validity of the BULlT-R has been supported by studies indicating that items
discrirrrinate between BN and control subjects (Thelen et al., 1991; Thelen et al., 1996).

Three-Factor Eating Questionnaire (TFEQ): The TFEQ (Stunkard & Messick,
1985) is a 51-item self-report measure used to assess dietary restraint, disinhibition, and
hunger. The TFEQ consists of three subscales including Cognitive Restraint of Eating
Behavior, Disinhibition of Cognitive Restraint, and Susceptibility to Hunger. Given that
one of the goals of the current study was to assess restricting eating pathology in the
absence of bingeing and purging (i.e., to define the R group), only the Cognitive Restraint
of Eating Behavior subscale was examined in analyses. This subscale consists of 21
items measuring cognitive concerns with food and dieting as well as behavioral methods
used to restrain food intake. Higher scores on the TFEQ Restraint scale represent greater
levels of dietary and cognitive restraint. Internal consistency for this subscale within a

combined group of dieters and nondieters has been shown to be good, with an alpha of

34

.93 (Stunkard & Messick, 1985). Within only a group of dieters, internal consistency has
been demonstrated as adequate (alpha =.79; Stunkard & Messick, 1985). Finally, among
non-obese subjects, coefficient alpha has been found to be .91 (Allison, Kalinsky, &
Gonnan, 1992). Validity of the Restraint subscale of the TFEQ has been supported in
studies showing discrimination in scores of dieters and nondieters (Stunkard & Messick,
1985).
Classifications

Participants were assigned to one of three groups based on their responses on the
EDI-2, BULIT-R, and TFEQ. Tertiles were used as cut-off points for many of the
subscales, as tertiles have been recommended for assigning women to pathological eating
categories based on continuous measures (Placanica, Faunce, & Soames Job, 2002). The
first group (R) consisted of women with elevated levels of symptomatology specific to
dietary restriction without bingeing and purging. To be included in this category,
individuals displayed high levels of dietary restraint on the TFEQ (i.e., top tertile) and no
binge eating and compensatory behaviors. The absence of binge eating and
compensatory behaviors was defined as reporting the absence of: eating impulsively until
feeling stuffed, rapid intake of a large amount of food, laxative or suppository use, self-
induced vomiting, and diuretic use (i.e., BULIT-R questions 6, 8, 9, 15, 27,31, 34, and
36); and the absence of uncontrollable hinges and eating large amounts of food privately
(i.e., EDI-2 questions 5, 28, and 46). Women categorized in the R group were also
required to exhibit high scores on the Body Dissatisfaction and Drive for Thinness scales
of the EDI-2 (i.e., top tertile) since these elevations are common in women with all forms

of eating pathology.

35

Based on these criteria, 20 women (2.5% of original sample) were assigned to the
R group. This number is slightly higher than prevalence rates for threshold AN (.5%-1%;
American Psychiatric Association, 2000), but because the R group contains women with
subthreshold eating pathology, the percentage of R women is consistent with what would
be expected (Klump et al., 2001). It should also be noted that because the R group
includes women who exhibit high levels of dietary restriction, body dissatisfaction, and
drive for thinness, the group is characterized by eating pathology that is above and
beyond that exhibited by women who are typical dieters. As evidence of this, women in
the R group showed significantly higher levels of drive for thinness and body
dissatisfaction than a group of typical dieters (Liebman, Cameron, Carson, Brown, &
Meyer, 2001).

The second group of women with eating pathology included individuals who
engage in bingeing and purging (B/P). These women received a score greater than 60 on
the BULIT-R and endorsed items indicating they engage in both bingeing and
compensatory behaviors. Specifically, they endorsed eating impulsively until feeling
stuffed, the rapid intake of a large amount of food, and the use of at least one
compensatory behavior (e.g., dieting, fasting, laxatives, diuretics, vomiting, or vigorous
exercise) in response to binge eating (i.e., BULIT-R questions 5, 8, 9, 18, and 34). B/P
women also showed high levels of Body Dissatisfaction and Drive for Thinness (i.e., top
tertile in both), and they were permitted to have any dietary restraint score.

Based on these inclusion criteria, 62 women (7.8% of original sample) were
assigned to this group. This number is higher than prevalence rates for threshold BN

(1%-3%; American Psychiatric Association, 2000), but again, the B/P group includes

36

women with subthreshold BN and the prevalence is similar to that in other studies of
subthreshold BN (e.g., Wade, Bulik, & Kendler, 2001). Furthermore, the percentage of
women in the B/P group is consistent with higher prevalence rates for BN in college
samples (Thelen et al., 1991). Confirrrring the validity of the B/P group, the mean
BULIT-R score of this group was 95.45 (SD = 13.23) which is highly similar to the mean
score obtained for women with subthreshold BN (i.e., 98) in previous research (Smith &
Thelen, 2000).

The third group included women who were not assigned to either of the
pathological eating groups and who did not exhibit any general disordered eating (CON
women). These women obtained scores lower than 50 on the BULIT-R, as research has
shown that women with no eating pathology have BULIT-R scores that fall between 0
and 50 (Smith & Thelen, 2000). CON women also scored in the bottom tertiles of the
Restraint subscale of the TFEQ and the Body Dissatisfaction and Drive for Thinness
subscales of the EDI-2. Moreover, CON women did not report any binge eating or
compensatory behaviors. Based on these criteria, 109 women (13.7% of original sample)
were assigned to the CON group, and this percentage is like that in other research
examining similar types of control groups (e.g., Mintz & Betz, 1988).
PersonalityClusters

Dimensional Assesament of Personality Pathology - Basic Questionaajre (DAPP-

BQ)

The DAPP-BQ (Livesley, Jackson, & Schroeder, 1992) is a 290-item self-report
questionnaire designed to assess 18 factor-derived dimensions of personality disorder.

These dimensions are listed and described in Table 1. Respondents rate each item on a 5-

37

point scale ranging from “very unlike me” to “very like me.” Higher scores on items
indicate greater levels of personality disturbance. The DAPP-BQ has demonstrated good
internal consistency, with alphas for the 18 dimensions ranging from .83 to .94 (Livesley
et al., 1992). Furthermore, validity of the DAPP-BQ has been supported by studies
finding that the factor structure of the measure is consistent across clinical and non-
clinical populations (Livesley et al., 1992). Finally, several higher order factors derived
from the DAPP-BQ have been shown to differentiate patients with personality disorders
from individuals in the general population (Livesley et al., 1992).

Parental Relationships

Parental Bonding Instrument (PBI)

The PBI (Parker, Tupling, & Brown, 1979) is a 25-item self-report instrument
designed to assess aspects of the parent—child relationship as remembered by the
respondent during the first 16 years of his/her life. The PBI consists of two subscales
measuring Parental Care and Overprotection. The Parental Care subscale is made up of
12 items measuring empathy, warmth, and affection. Higher scores on this subscale
indicate that subjects have experienced and received greater levels of care from their
parents. The Overprotection subscale is made up of 13 items assessing parental
intrusiveness, overcontrol, and excessive contact. Higher scores on this subscale suggest
increased parental overprotection. Respondents rate each item on a 4-point scale ranging
from “very like my parents” to “very unlike my parents.” Participants in the current
study completed the PBI twice, once for each parent.

Three-week test—retest reliabilities have been shown to be adequate in a non-

clinical sample for the Care (.76) and Overprotection (.63) subscales (Parker et al., 1979),

38

and nine-week test-retest reliabilities have been found to be high in depressed patients
(87-92; Parker, 1983). Validity of the PBI has been demonstrated by studies showing
differences in PBI scores between clinical samples with known disturbed family
functioning and non—clinical samples (Parker, 1983).

Structural Analysis of Social Behavior (SASB)

The SASB (Benjamin, 1974) is a circumplex model measuring interpersonal
relationships and their intrapsychic representations (see Figures 1, 2, 3). Benjamin
(1996) notes that the SASB is based on constructs delineated by Freud (1896), Murray
(1938), and Sullivan (1953), and also builds upon previous circumplex models developed
by Leary (1957) and Schaefer (1965). The SASB consists of two central axes, namely,
affiliation and interdependence. The horizontal affiliation axis, in general, ranges from
love on the right side of the model to hate on the left side. The poles of this axis can be
further described as representing attached/loving and attacking/hostile relationships,
respectively. The vertical interdependence axis ranges from freedom on the upper part of
the model to control on the lower part. In terms of relationships, the poles of this axis
represent differentiated and enmeshed relationships, respectively. Around each of three
circumplexes of the model are eight clusters that characterize the specific pattern of any
given interpersonal interaction, and as such, each cluster depends on the particular
combination of affiliation and interdependence measured in the interpersonal interaction.
For example, maternal overprotectiveness would be classified as highly enmeshed and
slightly affiliative, and thus cluster 5 (“Watching and Controlling”) on the Other
circumplex (see below for a description of the 3 circumplexes) would best describe this

interaction.

39

Both the horizontal and vertical axes are contained in three circumplexes, which
together, make up the full SASB model. Each circumplex indicates a different attentional
focus, with the top surface representing Focus on Other (see Figure 1), the rrriddle surface
representing Focus on Self (see Figure 2), and the lower surface representing an
lntrapsychic focus (see Figure 3). The Focus on Other circumplex assesses transitive
action directed at the subject from another person, including such actions as rejecting or
comforting. For example, the item “He lets me speak freely, and warmly tries to
understand me even if we disagree” refers to the subject’s relationship with a male
significant other and reflects action directed from the significant other to the subject. The
SASB can be used to rate any relationship important to the subject and thus the wording
of each item can be modified to reflect that specified relationship. In the current study,
subjects rated relationships with their mother and father.

The Focus on Self dimension measures the subject’s intransitive reaction to
another person, such as separating from others or approaching and enjoying them. For
example, the item “I let him speak freely, and warmly try to understand him even if we
disagree” captures the subject’s reaction to a male significant other. The third circumplex
characterizes the psychodynarrric notion of introjection, a theoretical concept positing that
a person intemalizes relationships with significant others, thus treating herself as others
have once treated her. The example item “I very tenderly and lovingly appreciate and
value myself” assesses one aspect of how the subject views herself, which is theorized to
be similar to how she was treated by significant others, such as parents.

There are several forms of the SASB, including a short, medium, and long form.

The SASB Medium Form was used in the current study because the author of the SASB

40

highly recommends using either the Medium or Long form in research (Benjamin, 2000),
in part due to the fact that it is not possible to calculate a measure of internal consistency
from the SASB Short Form. Thus, although the Short Form has been used consistently in
previous studies of eating disorder patients (Friedman et al., 1997; Wonderlich et al.,
1996), it does not appear to be as psychometrically sound as the Medium Form. It was
decided to use the Medium Form as opposed to the Long Form because the former
requires less time to complete (45 minutes versus 2 hours) but yields comparative
information. In addition, the Medium Form SASB has shown good internal consistency
across family relationships measured in the three circumplexes (.64-.92; Benjamin,
2000). The validity of the Medium Form has also been demonstrated by its ability to
discriminate between clinical and non-clinical samples (Benjamin, 2000).

The SASB Medium Form consists of 144 self-report items, of which 16 items
measure the Intrapsychic circumplex, and the remaining items measure the Other and
Self circumplexes. In the current study, participants rated 32 items describing their father
(i.e., (1) Father Focuses on Daughter and (2) Father Reacts to Daughter), 32 items
describing themselves in this relationship (i.e., (l) Daughter Focuses on Father and (2)
Daughter Reacts to Father), 32 items describing their mother (i.e., (1) Mother Focuses on
Daughter and (2) Mother Reacts to Daughter), and 32 items describing themselves in this
relationship (i.e., (1) Daughter Focuses on Mother and (2) Daughter Reacts to Mother).

Respondents rate each item on a scale from 0 (“never, not at all”) to 100 (“always,
perfectly”), based on how well the item describes the given relationship. The SASB is
scored by using software which yields eight cluster scores (corresponding to each cluster

in the respective circumplex) for each relationship measured. Lower scores for Clusters

41

1-4 indicate greater interpersonal disturbances with the rated significant other, whereas
higher scores for Cluster 5-8 indicate more pathological interpersonal interactions with
the rated other. Given that the SASB was used in the current study to examine family
dimensions previously found to be associated with eating pathology (e. g., parental
hostility and lack of nurturance and empathy), only Clusters 2, 3, 4, 6, 7, and 8 were
examined in analyses.

Level of Functionigz

Social Adjustment Scale Self-Report (SAS-SR)

The SAS-SR (Weissman & Bothwell, 1976) is a 54-item self-report instrument
measuring six areas of functioning: Work (as a student, homemaker, or worker outside
the home), Social and Leisure Activities, Relationships with Extended Family,
Marital/Partner Roles, Parental Roles, and Family Unit Roles. For each area, questions
on the SAS-SR assess the subject’s performance at given tasks (e. g., attending work,
going out socially with others, managing finances), degree of friction with others (e. g.,
arguments with co—workers, friends, and family), interpersonal behaviors (e.g., reticence,
dependency), and feelings and satisfaction (e. g., interest in work, loneliness, boredom).
Subjects rate each item on a 5-point scale, with higher scores indicating poorer social
adjustment. In addition to the six scales, 3 total score scale is obtained by summing all
items.

In the current study, the subscale “Work as a student” was analyzed and not
“Work as a homemaker” or “Work as a worker outside the home” because all of the
participants were students, whereas the other categories were not applicable to every

participant. The Parental Roles subscale was also not analyzed, as only 1 participant

42

reported having children. Finally, the Family Unit Roles subscale was not analyzed
because only 9 participants endorsed items relevant to this area.

The SAS-SR has been shown to generally have good internal consistency across
all seven scales (Cogley & Keel, 2003). In addition, the validity of this instrument has
been demonstrated by its ability to discriminate between BN and nonclinical women
(Johnson & Bemdt, 1983), and it has been used extensively with eating disordered
patients (Cogley & Keel, 2003; Herzog, Keller, Lavori, & Ott, 1987; Herzog, Norman,
Rigotti, & Pepose, 1986; Norman & Herzog, 1986; Rorty, Yager, Buckwalter, &
Rossotto, 1999).

History of Treatment

A question was included on the demographic questionnaire asking whether
subjects have received or are currently receiving treatment for any psychological disorder
(yes/no). In addition, participants indicated treatment(s) received from a list including
outpatient psychotherapy, outpatient psychiatric medication, hospitalization, and “other”
treatment. If women had received more than one of these treatments, they could respond
accordingly, and each response was considered separately in analyses of specific
treatment types (i.e., participants were double-counted if they received two different
types of treatment). Participants also specified the disorder and/or problems treated.

The final category, “other treatment,” was included in treatment analyses
(yes/no), but was not examined separately in analyses of treatment type because it was
deemed too broad. For this category, participants were not provided an opportunity to
specify the type of “other” treatment, and thus the exact treatment they received was

unknown. In addition, only 12 women (6.3%) endorsed “other”, compared to 47 women

43

(24.6%) who endorsed any form of treatment, 39 (20.4%) who received psychotherapy,
29 (15.2%) who took medication, and 9 (4.7%) who were hospitalized. With the
exception of hospitalization, which is a specific type of treatment, all of the other
treatment categories were endorsed at higher rates than the “other” treatment. Therefore,
it was thought that the first three treatment categories captured the most critical

information regarding treatment history.

Starting Analysis

InteflConsistencies, Variable mnsformations, and Effect Sizes

Internal consistencies were first computed for eating, personality, family
relationship, and level of functioning variables. Variables with positively skewed
distributions were transformed using a standard log transformation method. Seventeen
variables were transformed, thirteen of which were from the SASB: Cluster 7 Introject
(Self-Rejecting & Destroying); Cluster 6 Mother Focuses on Daughter (Belittling &
Blaming); Cluster 6 Daughter Focuses on Mother, Daughter Focuses on Father (Belittling
& Blaming); Cluster 7 Mother Focuses on Daughter, Father Focuses on Daughter
(Attacking & Rejecting); Cluster 7 Mother Reacts to Daughter, Father Reacts to Daughter
(Protesting & Recoiling); Cluster 7 Daughter Focuses on Mother, Daughter Focuses on
Father (Attacking & Rejecting); Cluster 7 Daughter Reacts to Mother, Daughter Reacts to
Father (Protesting & Recoiling); Cluster 8 Mother Focuses on Daughter (Ignoring &
Neglecting). The remaining four variables that were transformed included the DAPP-BQ
subscale Self-harm, and the SAS-SR subscales Student Work, Relationships with

Extended Family, and Marital/Partner Roles.

Across analyses, effect sizes are reported where appropriate in addition to
significance levels due to small sample sizes across groups.

Body Mass Index Across Groups

Body Mass Index (BMI) was assessed by comparing R and B/P groups with one
another and the control group (CON) using one-way ANOVAs and Tukey’s post hoc t-
tests. Although BMI is not a DSM-IV symptom of eating disorders, it is a significant
symptom-related characteristic often considered in studies of women with eating

disorders, and therefore it was analyzed in the current study.

Specific Aima

SJ)ecific Aim #1: Replication of Personality Clusters

To assess the presence of the five personality factors found by Goldner et al.
(1999), DAPP-BQ scores of women in the R and B/P groups were first subjected to a
confirmatory factor analysis (CFA) using the AMOS 5 statistical package. Next,
principal components factor analyses (PCA) were run to determine the factor structure of
the DAPP-BQ scores for the current study. Following this, CFA was again used to assess
this factor structure rendered from PCA. Hierarchical cluster analyses were then
conducted employing the squared Euclidean distance measure and Ward’s method
(Ward, 1963) using the SPSS 12.0 statistical package. Prior to clustering, variables used
in each analysis were transformed to z-scores. Two, three, and four cluster solutions
were tested to determine the best fit to the data, as it was believed that one of these
solutions would capture the most meaningful differences on personality dimensions

across clusters. Thus, the number of clusters examined did not exceed four because it

45

was thought that solutions with clusters greater than four would consist of at least two
clusters that would not significantly differ from each other on observed personality
dimensions. In addition, previous research has indicated that four and five cluster
solutions in women with eating disorders have not represented personality data as well as
a three cluster solution (Goldner et al., 1999). Clusters from the final solution were then
compared with each other and with the control group on DAPP-BQ subscales and BMI
using one-way AN OVAs and Tukey’s post hoc t-tests.

Sflific Aim #2: Persofinality Heterogeneity of R and B/P Grog

Chi-square tests were conducted to determine whether personality was
heterogeneous for both disordered eating groups and whether clusters cut across
diagnoses. These analyses evaluated whether the proportion of R and B/P women
significantly differed across clusters.

Smcific Aim #3: Level of Functioning

One-way ANOVAs and Tukey’s post hoc t—tests were used to compare clusters
across the SAS-SR social adjustment subscales, while chi-square tests were used to
compare clusters across history of treatment variables. Multiple hierarchical multiple
regressions and logistic regressions were then conducted to examine the ability of
personality clusters to predict SAS-SR social adjustment scores and history of treatment
variables above and beyond original eating disordered group classifications. Each
hierarchical multiple regression included one of the five social adjustment subscales of
the SAS-SR as the dependent variable, and the R and B/P groups were entered into the
first block, and the personality clusters (i.e., Adaptive, Rigid, Dysregulated) were

dummy-coded (Adaptive cluster as reference group) and entered into the second block.

46

The rationale for this specified order was due to the notion that eating disorder groups
should theoretically account for the largest amount of variance in level of functioning,
followed by personality clusters. Moreover, previous research utilized this same entry
method (Westen & Harnden-Fischer, 2001). In the logistic regressions, a similar entry
method was utilized, with the exception that the dependent variable for each regression
was one of the four history of treatment variables (i.e., general history of treatment
(yes/no), psychotherapy, psychiatric medication, or hospitalization).

Specific Aim #4: Family Relationships

Family relationships and intrapsychic dimensions were examined across
Adaptive, Rigid, and Dysregulated clusters and the CON group using one-way ANOVAs
and Tukey’s post hoc t-tests. PBI scores were first analyzed, followed by SASB scores.

Empatory Aim: Analysis of Control Group

The CON group was cluster analyzed by first computing squared Euclidean
distances between individuals’ DAPP-BQ scores in the CON group and DAPP-BQ group
means for each of the three clusters formed from the R and B/P groups. This clustering
approach was chosen over others (e. g., hierarchical clustering using Ward’s method)
since previous research has suggested highly similar personality clusters in women with
eating disorders and CON women (see Introduction above) (Asendorpf et al., 2001;
Block & Block, 1980; Caspi & Silva, 1995; Rammstedt et al., 2004; Robins et al., 1996;
Westen & Harnden-Fischer, 2001). Thus, CON clusters were formed based on the three
eating clusters to determine if, in fact, CON women would fall into the clusters that were
developed for the women with eating pathology. Squared Euclidean distance equations

were used—one for each of the clusters in the eating pathology groups. Each equation

47

contained group means for the respective cluster subtracted from CON individuals’
DAPP—BQ scores across the 14 subscales. For example, part of the equation for an
“Adaptive” cluster (see description of this cluster in Results below) was:

(1 = (CON group individuals’ DAPP-BQ scores on Subrnissiveness - Adaptive

cluster group mean on Submissiveness)2 + (CON group individuals’ DAPP—BQ

scores on Cognitive Distortion - Adaptive cluster group mean on Cognitive

Distortion)2 + (each remaining CON DAPP-BQ subscale score — each remaining

Adaptive cluster DAPP-BQ subscale score)2

Based on the equations, each woman in the CON group received one value
approximating their distance in DAPP-BQ scores to members of each of the eating
pathology clusters. These values were then used to classify each CON woman into one
of the final clusters by examining the smallest distance value across the three clusters.
For example, if a CON woman’s smallest distance score across clusters was found for the
“Adaptive” cluster, she was assigned to this cluster because her score was the closest in
distance to other members of this cluster. Following classification, CON clusters were
compared on DAPP-BQ subscales, SAS-SR social adjustment variables, and family
relationship (i.e., PBI and SASB) variables using one-way ANOVAs and Tukey’s post
hoc t-tests. Finally, chi-square analyses were conducted to compare clusters across

history of treatment variables.

48

RESULTS

Internal Consistencies

Table 2 presents internal consistencies for the eating, personality, family
relationship, and level of functioning variables. Overall, high internal consistencies were
found for these variables, indicating that items within each measure were highly
correlated with one another. These internal consistency values are similar to ones
reported previously for community-based samples (Allison et al., 1992; Benjamin, 2000;
Cogley & Keel, 2003; Livesley et al., 1992; Parker et al., 1979; Raciti & Norcross, 1987;
Thelen et al., 1996).
Body Mass Index Across Groups

Mean differences in BMI were examined across R, B/P, and CON groups. Two
outliers were found in the R group (BMI = 31.30, 39.00), and these two scores were
excluded from this analysis. BMI for all three groups fell in the normal range (18.5-
249), indicating that no group as a whole was underweight or overweight. Results from
the ANOVA suggested significant group differences in BMI (F = 15.28, df = 2, 182, p =
< .001). Results indicated that women in the B/P group (n = 61, M = 23.70, SD = 3.62)
showed significantly higher BMI than the CON group (n = 107, M = 20.97, SD = 2.77; p
= .000; d = .88). In addition, R women (n = 17, M = 22.16, SD = 2.77) showed a trend
toward higher BMI than CON women (d = .43), although results were not statistically
significant (p = .30). This unexpected result suggests the possibility that the R group
represents women who have been unsuccessful in their extreme dieting efforts. In other

words, although R women were deemed to have significant eating pathology based on

49

their reported high levels of restraint, drive for thinness, and body dissatisfaction, these
women have not achieved expected low weight. Consistent with this finding, previous
research has indicated that adolescent girls endorsing high levels of dietary restraint were
not successful in their dieting efforts (Stice, Presnell, Shaw, & Rohde, 2005). The low
BMI of the CON group suggests that these women may report no eating pathology
because they are naturally thin and satisfied with their body weight. Overall, despite the
finding that R women showed higher BMI than CON women, the R group nevertheless
evidenced high levels of dietary restraint, body dissatisfaction, and drive for thinness,

indicating that this group represents women with significant eating pathology.

Specific Aim #1: Replicflon of PersonalitLClusters

Personalityflctor Structure

Correlations between the 18 DAPP-BQ subscales were calculated, and these are
presented in Table 3. DAPP-BQ scores of women in the R and B/P groups were then
subjected to a CFA to test hypothesis 1a which predicted that CFA would yield the same
five personality factors (i.e., Neuroticism, Psychopathy, Compulsivity, Interpersonal
Difficulties, and Behavioral Disturbance) previously found by Goldner et a1. (1999).
This CFA indicated that the model could not be identified due to one of the latent factors
having only two indicators (see description below). Therefore, hypothesis 1a of the
current study was not supported.

Upon closer inspection of the 5-factor structure found by Goldner et a1. (1999),
two critical problems were detected. First, one of the putative personality factors (i.e.,

Compulsivity) had only two indicators (i.e., Compulsivity and Oppositionality), and the

50

indicator Oppositionality significantly cross-loaded on another factor (i.e., Neuroticism).
Second, there were multiple other DAPP-BQ subscales that loaded significantly across
factors. For example, the subscale Social Avoidance loaded on the factors Neuroticism
and Interpersonal Difficulties, while the subscale Cognitive Distortion loaded on the
factors Neuroticism and Behavioral Disturbance. Due to these substantial limitations of
the original factor structure proposed by Goldner et a1. (1999), it was decided to examine
personality scores in the current study through a series of PCA’s utilizing direct oblimin
rotation, a method used in previous DAPP-BQ research (Livesley, Jang, & Vernon, 1998;
Pukrop, Gentil, Steinbring, & Steinmeyer, 2001; van Kampen, 2002). PCA’s were
followed by a final CFA to assess the factor structure thought best to represent the
personality data. Determination of PCA factors for each analysis was based on
examination of the scree plot and eigenvalues greater than 1.

The first PCA included the 18 personality subscales of the DAPP-BQ, yielding 4
personality factors that accounted for 71.3 % of the total variance. The eigenvalues were
6.59, 3.00, 2.05, and 1.20. The four-factor solution and factor loadings are presented in
Table 4. Importantly, only two subscales, Oppositionality and Compulsivity, loaded on
the fourth factor, and one of these subscales, Oppositionality, cross-loaded highly on
factor 1. Because factor 4 appeared to be weak relative to the other three factors, a
second PCA was run forcing a three-factor solution. This PCA included the 18 DAPP-
BQ subscales, yielding 3 factors that accounted for 64.6% of the total variance. The
eigenvalues were 6.59, 3.00, and 2.05. Table 5 presents this three-factor solution and
factor loadings. Several difficulties with this solution were also found. Specifically, four

subscales, Oppositionality, Restricted Expression, Insecure Attachment, and Rejection,

51

significantly cross-loaded on multiple factors. Because of significant cross—loadings, it
was thought that these subscales did not represent distinct constructs, unlike the other 14
subscales. In order to obtain a more parsimonious model in which each subscale only
significantly loaded on one factor, these subscales were excluded from further analyses.
Therefore, another PCA was run forcing a three-factor solution excluding these four
subscales.

Three factors were found that accounted for 65.8% of the total variance. The
eigenvalues were 5.55, 2.18, and 1.49. Table 6 presents the three-factor solution and
factor loadings. Although this solution included fewer DAPP-BQ subscales than the
previous model, it accounted for slightly more of the total variance. In addition, no
subscales significantly loaded across personality factors in this model. Correlations
between the 14 subscales revealed no significant correlations above 0.75. Furthermore,
correction for attenuation due to lack of reliability of subscales was calculated for all
subscale inter-correlations above 0.60, and the highest resulting value was 0.81. As this
final value did not exceed 0.90, the 14 subscales appeared to measure distinct constructs,
and therefore, no subscales were combined. Finally, a PCA forcing a two-factor solution
was run with the 14 DAPP-BQ subscales. The two resulting factors accounted for only
55.2% of the total variance. Therefore, the 3-factor solution that utilized the 14 DAPP-
BQ subscales (Table 6) appeared to best represent the data, as it accounted for the most
total variance (65.8%) and contained no significant subscale cross-loadings.

Based on the content of the 14 subscales and the factor loadings, the three factors
were labeled Neuroticism, Psychopathy, and Intimacy Problems. The Neuroticism factor

overall appears to resemble the same factor found by Goldner et a1. (1999). Although the

52

names of the Psychopathy and Intimacy Problems factors found in the current study
generally match the Psychopathy and Interpersonal Difficulties factors found by Goldner
et a1. (1999), the content of the factor loadings is different across studies. For example,
the present study found that the subscales Callousness, Conduct Problems, and Stimulus
Seeking loaded on the Psychopathy factor. However, Goldner et a1. (1999) found that
Conduct Problems and Stimulus Seeking loaded on a different factor, Behavioral
Disturbance (not found in the current study). In addition, Callousness, Narcissism, and
Rejection loaded on Psychopathy in Goldner et al. (1999).

The final 3-factor model was next subjected to CFA to assess the fit of the model.
When this model was initially run, modification indices suggested that the subscale
Affective Instability should negatively load on the Intimacy Problems factor and be
allowed to cross-load on the Neuroticism factor. This modification was incorporated, and
the model was re-run. Figure 4 presents this model. Overall, fit indices indicated that the
model was a poor fit to the data (x2 = 170.33, df= 73, p = .000; RMSEA = .13; CFI = .83;
GFI = .77). This finding may be due to the tenuousness of the higher-order factor
structure of the DAPP-BQ. Indeed, the author of the DAPP-BQ has indicated that he has
not conducted CFAs on the putative hi gher-order structure of the DAPP-BQ, nor is he
aware of any other researchers who have done so (W. J. Livesley, personal
communication, October 31, 2005). Therefore, perhaps a meaningful higher-order factor
structure cannot be gleaned from the DAPP-BQ. As such, it is likely that the 14 DAPP-
BQ subscales found to measure distinct personality constructs in the current study offer

more reliable information regarding personality pathology than the hi gher-order factors.

53

Thus, the 14 DAPP-BQ subscales, rather than the hi gher-order factors, were used for all
cluster analyses.

Cluster Arglyses of Disordered Eating Groups

Two, three, and four cluster solutions were tested to determine the best fit to the
data. Both the two and four cluster solutions generated clusters that did not appear to
best account for personality scores. For example, the two-cluster solution divided women
into one cluster who were high on most of the DAPP—BQ dimensions, and the other who
were low on these dimensions. Thus, it was decided that a greater number of clusters
might allow finer gradations in personality dimensions. Results from the four-cluster
solution verified this notion, indicating that three of the clusters appeared to significantly
differentiate women according to Adaptive, Rigid, and Dysregulated levels of DAPP-BQ
dimensions. However, two of the four clusters each appeared to represent moderate
pathology and did not seem to differ from each other in meaningful ways on personality
dimensions. For example, both of these clusters showed moderate levels of personality
pathology on most dimensions compared to the Adaptive and Dysregulated clusters.

Results from the three-cluster solution indicated this model was the best fit to the
data, as each cluster represented a distinct class of women according to the level and type
of personality pathology (i.e., Adaptive, Rigid, and Dysregulated), thereby supporting
hypothesis lb. Forty women (48.8%) fell in the Adaptive cluster, 35 (42.7%) fell in the
Rigid cluster, and 7 (8.5%) fell in the Dysregulated cluster. Figure 5 presents the
dendrogram from the three-cluster analysis, illustrating the three distinct clusters.

The clusters were next formally compared with each other and the CON group on

the 14 DAPP-BQ subscales, and results are presented in Table 7 and Figure 6. In

54

general, the three clusters differed significantly on their level of personality pathology
across Adaptive, Rigid, and Dysregulated clusters and were consistent with clusters
found in previous studies (Goldner et al., 1999; Strober, 1983; Thompson-Brenner &
Westen, 2005; Westen & Harnden-Fischer, 2001) with regard to the level of personality
pathology and the percentage of women in respective clusters. Therefore, these three
clusters (i.e., Adaptive, Rigid, Dysregulated) were used in the remainder of analyses.

Women in the Dysregulated cluster showed the highest elevations on all DAPP-
BQ subscales, with the exception of Intimacy Problems. This cluster represents women
with overall serious psychopathology, and specifically, unstable interpersonal
relationships, high neuroticism, emotional dysregulation, suicidal ideation, derealization,
disorganized thought processes, addictive behaviors, and hypervigilance. Members of
the Rigid cluster showed significant elevations on most variables compared to the
Adaptive cluster and CON group, but lower levels of personality pathology than the
Dysregulated cluster on all variables except Intimacy Problems. Notably, the Rigid
cluster demonstrated higher Intimacy Problems than all groups, indicating that these
women have particular difficulties with sexual relationships and loving, intimate
relationships. These women also showed the lowest levels of Conduct Problems,
suggesting that they are overcontrolled. Although not statistically significant, women in
the Rigid cluster also showed the lowest levels of Stimulus Seeking compared to the
Adaptive and Dysregulated clusters (d = .45, .66, respectively), again pointing to a
tendency for these women to be more cautious and overcontrolled. Thus, overall this
cluster is distinguished by moderate levels of personality pathology, deficits in

interpersonal relationships, and tendencies toward overcontrol.

55

Women in the Adaptive cluster were characterized by higher levels of
Submissiveness, Affective Instability, Anxiety, Conduct Problems, and Narcissism
relative to the CON group, but lower levels of these variables (except Conduct Problems)
compared to the Rigid and Dysregulated clusters. Therefore, women in the Adaptive
Cluster display the least pathology of the three clusters, but nevertheless show increased
anxiety, need for approval, and acting-out behaviors.

Disordered Eating Symptoms Across Clusters of Eatipg Pathology

Women in the Adaptive, Rigid, and Dysregulated clusters were compared on
disordered eating symptoms to investigate potential differences across clusters. Table 8
presents mean differences on EDI-2 Drive for Thinness and Body Dissatisfaction,
BULIT-R Total Score, TFEQ Restraint, and BMI. In cases where differences among
clusters on these variables were not statistically significant (likely because of small
sample sizes), medium-to-large effect sizes were typically present. In general, the
Dysregulated cluster displayed the highest levels on all variables except BMI compared
to the Rigid and Adaptive clusters, whereas the Rigid cluster showed the next highest
levels, followed by the Adaptive cluster. The clearest difference among clusters could be
seen on the BULIT-R Total Score, suggesting that binge eating and compensatory
behaviors occur most frequently and/or at a greater intensity in women with Dysregulated
personality pathology, and decreases across the clusters. No differences in BMI were
observed across clusters.

Specific Aim #2: Personalifleterogeneity of R and B/P Groups
In order to evaluate hypothesis 2 that predicted personality to be heterogeneous

across R and B/P groups, the proportion of R and B/P women in each of the three clusters

56

was examined. Despite a non-significant chi—square result ()8 = 2.99, df = 2, p = .23),
hypothesis 2 was supported. As hypothesized, R (n = 20) and B/P (n = 62) women each
fell into more than one personality cluster. More specifically, R women fell into the
Adaptive (12/40, 30.0%) and Rigid (8/35, 22.9%) clusters; no R women fell in the
Dysregulated Cluster. By contrast, B/P women fell into the Adaptive (28/40, 70%),
Rigid (27/35, 77.1%), and Dysregulated (7/7, 100%) clusters. Both of these findings
were consistent with predictions. Moreover, it was thought that clusters would generally
cut across the R and B/P groups, and this was confirmed by findings indicating that each
cluster, with the exception of the Dysregulated cluster, contained R and B/P women.
Specific Aim #3: Level of Functioning

Cormparisons Across Clusters

 

The Adaptive, Rigid, and Dysregulated clusters were examined in analyses of the
SAS-SR subscales Student Work, Social and Leisure Activities, Relationships with
Extended Family, Marital/Partner Roles, and Total Score. Table 9 presents mean
differences in these level of functioning variables for the Adaptive, Rigid, and
Dysregulated clusters and overall CON group. Overall, predicted differences between
clusters and the CON group were found, whereby the Dysregulated cluster evidenced the
greatest difficulties in psychosocial functioning across all variables, followed by the
Rigid and Adaptive clusters and the control group. Thus, the Rigid and Adaptive
Clusters showed moderate and mild disruptions, respectively, in their level of functioning
compared to the CON group. The only variable that did not reach statistical significance
across groups was Marital/Partner Roles. This is likely due to small sample sizes in each

group because this area was not applicable to all women in the study. However, medium-

57

to-large effect sizes found for this variable suggest that reliable differences were present
across clusters and the CON group.

Level of functioning was also assessed by the history of treatment measure, which
asked women if they had ever received psychological treatment (yes/no) and if yes, what
type of treatment: outpatient psychotherapy (yes/no), outpatient psychiatric medication
(yes/no), and hospitalization (yes/no). Overall, only one variable, outpatient psychiatric
medication, was found to be significantly different across clusters (x2 = 6.82, df = 2, p =
.03). Across Adaptive (n = 40), Rigid (n = 32), and Dysregulated (n = 7) clusters, the
percentage of women who received psychiatric medication increased (4/40, 10%; 10/32,
31.2%; 3/7, 42.9%, respectively), which is consistent with previous research that found
an association between increasing levels of personality pathology and treatment-seeking
(Perkins et al., 2005). Despite non-significant findings on the other three history of
treatment variables, similar patterns of treatment-seeking were found. For the variable
assessing history of any treatment, 30% (12/40) of Adaptive women responded
positively, compared to 42.9% (15/35) of Rigid women and 57.1% (4/7) of Dysregulated
women. Regarding psychotherapy, 23.1% (9/39) of Adaptive women reported a history
of this treatment, relative to 41.9% (13/31) of Rigid women and 57.1% (4/7) of
Dysregulated women. Hospitalization was reported by 5% (2/40) of Adaptive women,
9.4% (3/32) of Rigid women, and 14.3% (1/7) of Dysregulated women. Interestingly,
across the three clusters, 16.1% (5/31) of women indicated that they sought treatment for
an eating disorder, 19.4% (6/31) sought treatment for anxiety, 16.1% (5/31) sought
treatment for depression, 16.1% (5/31) sought treatment for comorbid anxiety and

depression, and 10 women did not report the disorder for which they sought treatment.

58

Predictive Relationships between Original Disordered Eating Groups, Personality
Clusters. and Functioning
I To test hypothesis 3, which predicted that personality clusters formed from the

disordered eating groups would significantly predict women’s level of functioning better
than the original disordered eating groups, five hierarchical multiple regressions and four
logistic regressions were performed. For the multiple regressions, personality clusters
significantly predicted four of the SAS-SR subscales, namely, Student Work, Social and
Leisure Activities, Relationships with Extended Family, and Total Score (see Table 10).
Across the four regressions, results indicated that the original R and B/P groups did not
account for significant proportions of variance in the prediction of any psychosocial
functioning variables. By contrast, personality clusters accounted for significant variance
in the prediction of all four SAS-SR variables. Thus, personality clusters, and not
disordered eating groups, were found to significantly predict psychosocial functioning,
supporting hypothesis 3. Neither personality clusters nor disordered eating groups
significantly predicted the SAS-SR subscale, Marital/Partner Roles. This may be due to
the relatively small number of women who indicated that they were married or had a
partner (43/82, 52.4%).

Logistic regression analyses showed that only one history of treatment variable,
psychiatric medication, was significantly predicted by either the original disordered
eating groups or personality clusters. Results from this analysis are presented in Table
11. This finding is consistent with results discussed previously indicating that a history
of psychiatric medication use was the only treatment variable that differed significantly

across clusters. Although the overall model in the first step that included only R and B/P

59

women was significant, the R and B/P groups did not significantly account for variance
in history of psychiatric medication (Wald = 3.09, p = .08). When personality clusters
were included in the second step, the overall model was significant, but the addition of
the personality clusters was only significant at a trend-level (Step 2, p = .06).
Nevertheless, in the final model, only the Rigid cluster was a significant predictor of
psychiatric medication (Wald = 4.26, p = .04). It is likely that the Dysregulated cluster
would also have been a significant predictor of psychiatric medication (Wald = 2.92, p =
.09), but this cluster’s small sample size (n = 7) limited significant findings.

Overall, results from the history of treatment analyses only provided modest
additional support that personality clusters were better predictors of level of functioning
than eating disorder groups. However, because the study sample consisted of women
from the community who reported low levels of treatrnent-seeking in general, it was
difficult to obtain significant findings in these analyses. These findings notwithstanding,
results from analyses of SAS-SR subscales provide strong evidence that personality
clusters significantly predicted women’s psychosocial functioning, whereas disordered
eating groups were not predictive of functioning.

Specific Aim #4: Family Relationships

PBI Scores

Hypothesis 4 was evaluated by first comparing PBI scores across the Adaptive,
Rigid, and Dysregulated clusters, and CON group to determine if the severity of family
disturbances paralleled the severity of personality pathology across clusters. Mean
differences in PBI Maternal Care, Maternal Overprotection, Paternal Care, and Paternal

Overprotection among eating clusters and the CON group are presented in Table 12.

60

Results indicated that the Dysregulated cluster recalled the least Maternal and Paternal
Care compared to the other two clusters and the CON group, supporting hypothesis 4. In
addition, this difference appeared to be even more robust in the Dysregulated cluster for
Paternal Care, whereby women recalled their fathers as less caring (M = 13.33, SD =
11.64) than their mothers (M = 19.29, SD = 14.47). Although this finding was not
statistically significant, a medium effect size was found (d = .45). In general, the results
for Maternal and Paternal Care across the other clusters were consistent with
expectations, such that the Rigid cluster recalled less Parental Care than women in the
Adaptive cluster and CON group. Also, the Adaptive cluster and CON group did not
differ on Maternal Care, whereas a small—to-medium effect size (d = .36) was found for
Paternal Care between these groups, indicating that Adaptive women recalled less
Paternal Care than controls.

Novel findings were detected for Parental Overprotection among clusters and the
CON group. Specifically, women in the Dysregulated cluster recalled their mothers as
more overprotective than Adaptive and Rigid women and CON women. Differences in
Maternal Overprotection were also found across the CON group and Adaptive and Rigid
clusters, suggesting that the severity of personality pathology mirrored the increase in
recall of Overprotection across clusters. Notably, women’s scores on Paternal
Overprotection paralleled these findings with one important exception, namely, that
women in the Dysregulated cluster recalled the least Paternal Overprotection compared to
both other clusters and the CON group. It is possible that this finding reflects

Dysregulated women’s perception that their fathers were underinvolved or absent.

61

SASB Scores

Hypothesis 4 was also tested by analyzing SASB scores across the Adaptive,
Rigid, and Dysregulated clusters, and CON group. Correlations between maternal
variables (e. g., Mother Focuses on Daughter, Mother Reacts to Daughter, Daughter
Focuses on Mother, and Daughter Reacts to Mother) on SASB Clusters 2, 3, and 4 of the
Focus on Other and Focus on Self Circumplexes revealed significant correlations ranging
from 0.6 to 0.9, with many of these correlations above 0.8. Similarly, correlations
between paternal variables (e.g., Father Focuses on Daughter, Father Reacts to Daughter,
Daughter Focuses on Father, and Daughter Reacts to Father) on these three SASB
clusters indicated significant correlations in the same range as maternal variables.
Correction for attenuation due to lack of reliability of subscales was calculated separately
for maternal and paternal variables of Clusters 2, 3, and 4 with inter—correlations above
0.60, and once corrected, the majority of resulting values were .90 or above. Therefore,
all of the variables across these SASB Clusters were thought to be only measuring two
distinct constructs. As a result, variables of Clusters 2, 3, and 4 were combined by
computing two mean variable scores (i.e., one for mothers and one for fathers) to form
two variables that were used in analyses. These variables were labeled Maternal
Empathy and Nurturance and Paternal Empathy and Nurturance.

Mean differences for the Adaptive, Rigid, and Dysregulated clusters and CON
group on the SASB Clusters 6, 7, and 8 and the combined Parental Empathy and
Nurturance Cluster are presented in Table 13. Women in the Dysregulated cluster
perceived their relationships with their mothers as less mutually empathic and nurturing

than women in the Adaptive cluster and CON women. Levels of empathy and nurturance

62

also paralleled levels of personality pathology in the other clusters, such that Rigid
women perceived less empathy and nurturance compared to Adaptive women, and this
latter group perceived less empathy and nurturance than the control group.

In general, Dysregulated women also perceived relationships with their mothers
as the most mutually Belittling, Attacking, and Ignoring than women in the other clusters
and CON women. The levels of Belittling, Attacking, and Ignoring generally decreased
across Rigid, Adaptive, and control women, but Rigid women did not differ from
Adaptive women on these variables. Two exceptions to this pattern were that levels of
Attacking did not differ between Dysregulated and Adaptive clusters, and Attacking was
higher in Adaptive women compared to Rigid women. On variables assessing perceived
reactions between mothers and daughters, Dysregulated women reported the highest
mutual levels of Sulking, Protesting, and Walling Off relative to other clusters and
controls. As before, levels of these variables generally decreased across the other clusters
and controls, except between Rigid and Adaptive women.

Regarding paternal relationships, Dysregulated women perceived the lowest
levels of empathy and nurturance in relationships with their fathers compared to the other
clusters and CON women. However, unlike differences found between Adaptive and
Rigid clusters and control women on Maternal Empathy and Nurturance, perceptions of
Paternal Empathy and Nurturance were similar across the Rigid and Adaptive clusters
and CON women. Nevertheless, Dysregulated women appeared to view their fathers as
less empathic (M = 46.32, SD = 31.97) than their mothers (M = 58.57, SD = 33.54).
Although this finding was not statistically significant, a small-to-medium effect size (d =

.37) was found between these variables. In terms of hostile relationships between fathers

63

and daughters, Dysregulated women perceived the most Belittling, Attacking, and
Ignoring relationships with their fathers compared to other clusters and control women.
Again, the level of these hostile dimensions generally increased alongside the level of
personality pathology, but women in the Adaptive cluster did not differ from those in the
Rigid cluster on these variables. The only other exceptions to this were found on
Belittling, such that Rigid women evidenced lower levels of Belittling than Adaptive
women, and no differences were found between Rigid and CON women. Reactions
between fathers and daughters were shown to be different among clusters and CON
women, with Dysregulated women reporting the highest levels of Protesting and Walling
Off compared to Adaptive and Rigid clusters and controls. The levels of these variables
also mirrored levels of personality pathology, but women in the Adaptive cluster did not
differ from those in the Rigid cluster on these variables. Finally, no differences were
found on the variable Sulking between the three clusters, but all clusters showed higher
levels of Sulking compared to the control group.

Differences in intrapsychic, or introjected dimensions, were found between
Adaptive, Rigid, and Dysregulated clusters and CON women. Specifically, Dysregulated
women showed the lowest Self-Accepting, Self—Loving, and Self-Nourishing relative to
the other clusters and control group. Furthermore, Dysregulated women showed the
highest levels of Self-Indicting, Self-Rejecting, and Daydreaming compared to all groups.
As with previous analyses, a similar pattern was observed across these variables, whereby
levels of personality pathology corresponded with levels of family disturbances.

Overall, findings from the SASB support hypothesis 4, in that CON women

showed the least family disturbances, and the degree of disturbances generally increased

across Adaptive, Rigid, and Dysregulated clusters. Moreover, the Dysregulated cluster
consisted of women with the most family disturbances, including an absence of parental
empathy and nurturance, and increased hostility. Also consistent with hypothesis 4,
Dysregulated women perceived their fathers as lacking in nurturance and empathy more
than their mothers. Adaptive women perceived the least family disturbances of the three
clusters, but nevertheless perceived low levels of parental empathy and nurturance and
increased hostility compared to controls. Rigid women perceived a moderate degree of
family disturbances, characterized by levels of parental empathy and nurturance and
hostility that were lower than women in the Dysregulated cluster. Finally, expected

differences were found on intrapsychic dimensions between clusters and the CON group.

Exploratory Aim: Analysis of Control Group

Cluster Analysis of Control Group

The CON group was analyzed according to the squared Euclidean distance
method described in the statistical analysis section. The three CON clusters were then
compared on the 14 DAPP-BQ subscales to evaluate the exploratory hypothesis that
CON women would fall into three clusters resembling Resilients, Overcontrollers, and
Undercontrollers. Results are presented in Table 14 and Figure 7. Overall, the three
clusters were a good fit to the data, as differences were found on significant personality
variables to differentiate the clusters according to Resilients (49.1% of CON group),
Overcontrollers (39.8% of CON group), and Undercontrollers (11.1% of CON group),
providing support for the exploratory hypothesis. The Undercontrollers showed

elevations on most personality variables relative to the Overcontrollers and Resilients, but

65

the Undercontrollers showed specific elevations relative to these groups on the following
subscales: Self-harm, Identity Problems, Affective Instability, Stimulus Seeking,
Callousness, and Conduct Problems. Therefore, as predicted, the Undercontroller cluster
consisted of women characterized by impulsivity, unstable interpersonal relationships,
and high levels of neuroticism.

The second cluster, the Overcontrollers, generally had scores on the DAPP—BQ
subscales that fell between the Resilients and Undercontrollers. As hypothesized, they
evidenced elevated levels of interpersonal difficulties and rigid control of impulses, as
measured by the Social Avoidance, Intimacy Problems, Stimulus Seeking, and
Compulsivity subscales. Specifically, Overcontrollers showed elevated levels of Social
Avoidance and Intimacy Problems. In addition, they demonstrated low levels of
Stimulus Seeking, characteristic of those higher on ego-control. Finally, although it did
not reach statistical significance, Overcontrollers showed the highest level of
Compulsivity compared to the other two clusters. As support for this finding, a small-to-
medium effect size ((1 = .35) for Compulsivity was found between Overcontrollers and
UIldeI‘Controllers, as well as a small effect size (d = .20) between Overcontrollers and
ReSilients. The final cluster, Resilients, demonstrated the lowest levels on all variables
except Stimulus Seeking and Compulsivity relative to the other clusters. Therefore, as
PrediCted, these women do not display any specific symptoms of personality pathology
and Seem to be psychologically well-adjusted.

Level of Functioning

Across women in the Resilient, Overcontroller, and Undercontroller clusters,

mean differences on the SAS-SR subscales were analyzed, and results are presented in

66

Table 15. Consistent with predictions and with previous research (Costa, Herbst,
McCrae, Samuels, & Ozer, 2002), Overcontrollers and Undercontrollers demonstrated
deficits in their psychosocial functioning in the range of areas measured by the SAS-SR.
As was found with the Adaptive, Rigid, and Dysregulated clusters, the Marital/Partner
Roles subscale did not show statistically significant mean differences across Resilients,
Overcontrollers, and Undercontrollers. Nevertheless, small-to—large effect sizes were
observed between clusters on this variable. Also, although not statistically significant,
results showed that Undercontrollers evidenced greater impairments in psychosocial
functioning than Overcontrollers on all SAS-SR variables. Small-to-medium effect sizes
were found between these two clusters on level of functioning variables, with the largest
effect size (d = .63) on the SAS-SR Total Score.

Resilients, Overcontrollers, and Undercontrollers were not shown to differ
significantly on any of the history of treatment variables, although patterns consistent
with expectations were found on three of the four variables. For example, only 9.4%
(5/53) of Resilients reported a general history of treatment, compared to 16.3% (7/43) of
Overcontrollers and 33.3% (4/12) of Undercontrollers. Furthermore, only 6.1% (3/49) of
Resilients had received psychotherapy, whereas 17.9% (7/39) of Overcontrollers and
27.3% (3/11) of Undercontrollers reported a history of this treatment. Finally, 8.2%
(4/49) of Resilients reported a history of psychiatric medication, relative to 15% (6/40) of
Overcontrollers and 16.7% (2/12) of Undercontrollers. History of hospitalization did not
show similar patterns as the other treatment variables, as only three women in total
indicated a prior history of hospitalization. This is not surprising given that women in

these clusters were members of the CON group and from a community sample. Finally,

67

across the three personality types, nearly all women who sought treatment (14/ 16; 87.5%)
indicated that they were treated for anxiety or depression (2 women did not report their
disorder treated).

Overall, Resilients, Overcontrollers, and Undercontrollers demonstrated
increasing rates of treatment-seekin g across clusters, suggesting that Undercontrollers
may experience the highest levels of personality pathology of the three clusters. This
result also fits with the previously observed finding in the current study that
Undercontrollers showed the most psychosocial deficits.

Family Relationships

The exploratory hypothesis that Overcontrollers and Undercontrollers would
display greater levels of family difficulties compared to Resilients was tested by
examining PBI and SASB scores across these clusters. Mean differences in PBI variables
are presented in Table 16. Undercontrollers recalled the least Maternal and Paternal Care
relative to the other two clusters, and Overcontrollers recalled less Parental Care than
Resilients, consistent with predictions. Similar to results found with the Dysregulated
eating cluster, Undercontrollers recalled their fathers as less caring than their mothers (t =
2.22, p = .05; d = .81). Regarding Parental Overprotection, Undercontrollers recalled
their mothers as more overprotective than Overcontrollers and Resilients, and
Overcontrollers recalled their mothers are more overprotective than Resilients. For
Paternal Overprotection, Undercontrollers recalled their fathers as more overprotective
than Overcontrollers; however, Overcontrollers and Resilients did not display differences
in their recall of Paternal Overprotection, and therefore, both of these groups are

considered to evidence no difficulties on this family dimension.

68

SASB scores were also analyzed, and mean differences on Clusters 6, 7, and 8
and the combined Parental Empathy and Nurturance Cluster are presented in Table 17.
On variables assessing daughters’ perceptions of their relationships with their mothers, it
was consistently shown that Resilients displayed the least family disturbances,
Overcontrollers demonstrated moderate difficulties, and Undercontrollers showed the
most family disturbances. As evidence of this, Undercontrollers perceived their
relationships with their mothers as the least mutually empathic and nurturing compared to
other clusters. Moreover, Undercontrollers perceived the most elevated levels of
Belittling, Attacking, and Ignoring with their mothers, and their reactions to their mothers
were marked by the greatest levels of Sulking, Protesting, and Walling Off compared to
the other clusters. On all of these family dimensions, Overcontrollers reported lower
levels of disturbances than Undercontrollers, but higher levels than Resilients. Similar
results were found with daughters’ perceptions of their relationships with their fathers.
Again, Undercontrollers perceived the least mutual empathy and nurturance and the most
Belittling, Attacking, Ignoring, Sulking, Protesting, and Walling Off with their fathers.
Overcontrollers reported fewer of these disturbances, and Resilients reported the fewest
disturbances.

Finally, regarding intrapsychic dimensions, the same pattern was observed across
Resilients, Overcontrollers, and Undercontrollers, whereby the latter cluster reported the
most deficits in Self-Accepting, Self-Loving, Self-Nourishing, Self-Indicting, Self-
Rejecting, and Daydreaming. Overall, results from the SASB across Resilients,
Overcontrollers, and Undercontrollers confirm the exploratory hypothesis that the latter

two groups would perceive family disturbances relative to Resilients. Importantly,

69

although not predicted, differential family pathology was found between Overcontrollers
and Undercontrollers, such that Undercontrollers consistently showed more deficits than

Overcontrollers.2

 

2 The 604 women excluded from other analyses were cluster analyzed using the same method employed
with the CON group, and almost identical results were found across both groups of women (data not
shown). Specifically, the 604 women were clustered into Resilients, Overcontrollers, and Undercontrollers,
and these clusters differed predictably on level of functioning and family relationship variables. The 604
women were also compared to the Adaptive, Rigid, and Dysregulated clusters, and in general, similar
results were found between these groups on personality, level of functioning, and family variables as were
found between the CON group and the three clusters (data not shown).

70

DISCUSSION

Findings from the current study represent a valuable contribution to the literature
on eating disorders, personality clusters, and family relationships, as this was the first
study to examine these constructs in a sample of women with subclinical eating
pathology. Although three personality clusters have been documented in the eating
disorders literature (Goldner et al., 1999; Strober, 1983; Thompson-Brenner & Westen,
2005; Westen & Harnden-Fischer, 2001), all previous studies investigated women with
clinical eating disorders. Thus, the generalizability of these personality clusters to
women with subclinical eating pathology or Eating Disorders Not Otherwise Specified
has been unknown. Given that the majority of women with eating disturbances who seek
treatment tend to have subclinical eating pathology (Andersen et al., 2001), findings from
the present study provide crucial information regarding the validity of the personality
clusters for these women. The fact that the three personality clusters were linked to
meaningful external correlates such as psychosocial functioning and family relationships
suggests that an evaluation of personality pathology should be the sine qua non in the
assessment and treatment of women with eating pathology.

This study also improved upon previous research by including stronger measures
of psychosocial and clinical functioning and by investigating family relationships of each
of the personality clusters. In addition, this study was unique by virtue of its exploratory
aim, which consisted of an analysis of personality clusters among women without eating
pathology and their corresponding levels of functioning and family relationships. This

made it possible to show that control women fell into similar clusters as eating disordered

71

women based on the personality dimensions of ego-control and ego-resiliency, thereby
underscoring the need to examine personality clusters in groups of individuals with other
forms of psychopathology. Overall, findings from the current study have the potential to
inform the eating disorders literature by suggesting that personality clusters are universal
in all eating disordered patients and are more robust indicators of the severity and
phenomenology of eating disorders than the DSM-IV eating disorder diagnoses.
Moreover, findings point to specific family relationships of each cluster that should be
targeted in treatment. Finally, findings from this study suggest that personality clusters
may be used to diagnose personality pathology, and as such, represent a viable alternative
to current Axis H diagnoses.
Personality Clusters in Women with Subclinical Eating Pathology

The first goal of this study was to replicate the three personality clusters found
among eating disordered women in previous research (Goldner et al., 1999; Strober,
1983; Thompson-Brenner & Westen, 2005; Westen & Hamden-Fischer, 2001) by using
the DAPP-BQ with women exhibiting a variety of eating pathology. It was anticipated
that confirmatory factor analysis would yield the same five-factor personality structure
found by Goldner et al. (1999) in their study of women with a range of eating disorders,
which also employed the DAPP-BQ. However, support for this factor structure was not
found, and clustering of the R and B/P women was done by using the 14 DAPP-BQ
subscales.

The inability of the current study to render a five—factor structure akin to the one
found in the previous study of eating disordered women (Goldner et al., 1999) can be

explained by carefully assessing this prior study and others that have utilized the DAPP-

72

BQ. In most studies of the DAPP—BQ that have used either a clinical sample consisting
of personality disordered patients or a non-clinical sample, a four-factor structure has
been found (Bagge & Trull, 2003; Livesley et al., 1998; Pukrop et al., 2001; van
Kampen, 2002). Although this four-factor structure has been replicated across studies, no
confirmatory factor analyses have been reported, and the resulting factor structure
consistently described appears to be problematic. Interestingly, Livesley et al. (1998)
suggest that the Compulsivity factor may not be a robust higher-order factor and that a
three-factor solution instead may adequately represent personality pathology.

Despite the expectation that the three-factor solution would provide an accurate
higher-order model of personality pathology in the current study, the 14 DAPP-BQ
subscales proved to yield a three-cluster solution that best fit the data and closely
approximated clusters found in previous studies. Thus, the first aim of the study to
replicate three personality clusters was accomplished, but by using individual subscales
of the DAPP-BQ, and not the hi gher-order factors. The three clusters were labeled
Adaptive, Rigid, and Dysregulated, based on the level and type of personality pathology
present, with the majority of women falling in the Adaptive cluster followed by the Rigid
and Dysregulated clusters. Women in the Dysregulated cluster were characterized by
serious psychopathology, including high levels of emotional dysregulation, neuroticism,
unstable interpersonal relationships, suicidal ideation, derealization, disorganized thought
processes, addictive behaviors, and hypervigilance. This cluster appears to be similar to
the Severe cluster found by Goldner et al. (1999), the cluster of women with serious
pathology found by Strober (1983), and the emotionally dysregulated/undercontrolled

cluster found by Westen and Harnden-Fischer (2001).

73

Women in the Rigid cluster generally showed lower levels of personality
pathology than the Dysregulated cluster, but higher elevations compared to the Adaptive
cluster. However, Rigid women showed the highest level of Intimacy Problems
compared to the other clusters, indicating that these women have particular difficulties
with sexual relationships and interpersonal relationships in general. These women also
reported low levels of Stimulus Seeking, which suggests they may be prone to
overcontrol. This cluster also bears a strong resemblance to a cluster of eating disordered
women described in other studies as Rigid (Goldner et al., 1999), or characterized by
interpersonal difficulties and control of impulses (Strober, 1983), or labeled as
constricted/overcontrolled (Westen and Harnden-Fischer, 2001).

The last personality cluster, Adaptive, was represented by women who displayed
the least personality pathology of the three clusters, but nevertheless showed increased
anxiety, need for approval, and acting-out behaviors compared to the control group. As
with the other two clusters, the Adaptive cluster is consistent with other studies that have
found a group of eating disordered women with mild, but still notable personality
pathology (Goldner et al., 1999; Strober, 1983; Westen and Harnden-Fischer, 2001).

In terms of eating disordered symptoms across clusters, the Dysregulated cluster
displayed the highest levels of Drive for Thinness, Body Dissatisfaction, Restraint, and
overall frequency and intensity of bingeing and compensatory behaviors compared to the
Rigid and Adaptive clusters. The Rigid cluster showed the next highest levels of these
symptoms, followed by the Adaptive cluster. The most striking difference in
symptomatology across clusters was found on binge eating and the use of compensatory

behaviors, suggesting an association between these types of behaviors and greater levels

74

of personality pathology. Indeed, Strober (1983) found a similar pattern in the Severe
cluster in his study of women with AN, such that Severe women demonstrated the
highest levels of binge eating and compensatory behaviors compared to the other clusters.
These findings strongly suggest that binge eating and compensatory behaviors are
markers of more serious pathology, especially given that Strober (1983) found these
results in a sample of women exclusively with AN. It is also interesting that binge eating
and compensatory behaviors are most strongly associated with women who have trouble
with impulse control and in regulating affect (i.e., the Dysregulated cluster), as these
disordered eating behaviors appear to be overt symptoms of this particular personality
pathology (Westen & Harnden-Fischer, 2001).
Heterogeneity of Personam in Women with Subclinical Eating Pathology

The second goal of the current study was to determine if personality was
heterogeneous in the original R and WP groups and whether clusters cut across these
diagnostic groups. Findings indicated support for both of these hypotheses, whereby
members of the R and B/P groups fell into more than one personality cluster, and each of
these clusters, with the exception of the Dysregulated cluster, contained both R and B/P
women. These findings parallel previous research which found that women with clinical
diagnoses of AN, BN and AN/BN demonstrated heterogeneous personality profiles and
that the three profiles cut across eating disorder diagnoses (Westen & Harnden-Fischer,
2001). Furthermore, Westen and Harnden—Fischer (2001) found that AN women fell into
only two of the clusters and not the Dysregulated cluster, which also matches current
findings, in that the R group could be grouped only into Adaptive or Rigid personality

pathology clusters. This finding provides validity for the accuracy of the R group, as this

75

group included women who manifested symptoms similar to Restricting Type AN, just as
the previous study’s AN diagnosis only included those women with Restricting Type and
not Binge-Eating/Purging Type. Nevertheless, the R group had higher BMIs than the
CON group, suggesting that the former group was unable to achieve the low weight
characteristic of women with Restricting Type AN. However, it is notable that women in
the R group were of normal weight yet had very high levels of body dissatisfaction and
drive for thinness. This mimics women with AN at the beginning of the disorder when
they typically are of normal weight (American Psychiatric Association, 2000), which
makes their body preoccupation and desire to lose weight all the more perplexing.
Therefore, while the R group did not exactly mirror women with Restricting Type AN,
they did share common characteristics such as cognitive dietary restraint, body
dissatisfaction, and drive for thinness.
Level of Functioningin Women with Subclinical Eating Pathology
Comparisons Across Clusters

The third aim of the study was to compare levels of psychosocial and clinical
functioning across clusters and evaluate whether original eating disorder classifications
(i.e., R and B/P groups) or personality clusters better predicted functioning in women
with eating disturbances. Dysregulated women showed the most deficits in psychosocial
functioning across areas including their performance in school, social and leisure
activities, relationships with extended family, their marital or partner role, and overall
functioning. Thus, these women appear to have particular difficulties in completing and
taking interest in their schoolwork, tend to be socially isolated and not engaged in

recreational activities, maintain distant or problematic relationships with relatives, and

76

experience relationship difficulties with partners. Across clusters, Rigid women reported
moderate levels of these psychosocial disturbances, indicating that they have difficulties
across the same psychosocial domains as Dysregulated women, but to a lesser degree.
Thus, Rigid women generally have trouble adapting to and coping with life demands such
as academic pressure, and they experience interpersonal difficulties. Adaptive women
reported minimal problems, but nonetheless reported elevations compared to the CON
group. Therefore, Adaptive women appear to experience little, but some difficulty in
adapting to life demands.

Regarding treatment history, women with Dysregulated personality pathology
sought treatment of all types (psychotherapy, psychiatric medication, hospitalization) at
higher rates than Rigid or Adaptive women, and the rate of treatment-seeking decreased
across clusters, again suggesting that women with higher levels of personality pathology
experience greater distress that likely prompts them to pursue treatment.

Comparisons of Clusters versus R and B/P Groupings

In general, personality clusters were better predictors of history of treatment and
psychosocial functioning than the original R and B/P groupings. Limited support was
found for the clusters as opposed to the R and B/P groups to predict treatment history,
namely psychiatric medication. However, findings may have been limited by the small
number of women who reported a history of treatment. This low rate of treatment-
seeking is consistent with research showing that only one-third of women with AN and
only 6% of women with BN from the community ever receive mental health treatment
(Hock & van Hoeken, 2003). Given that women exhibited subthreshold pathology and

were drawn from the community in the current study, it is remarkable that even modest

77

support was found for the clusters’ ability to predict treatment history. This suggests
even more strongly that personality pathology, rather than eating pathology, may be the
impetus for some cases of treatment-seeking, especially in women with greater levels of
personality pathology. As such, it was found in the current study that history of
psychiatric medication was significantly predicted by Rigid personality pathology and
also predicted at a trend—level by Dysregulated personality pathology.

When R and B/P groups and the three personality clusters were compared to each
other in their ability to predict psychosocial functioning variables, meaningful results
emerged. Personality clusters, and not the R and B/P groups, accounted for significant
variance in the prediction of the several psychosocial areas. Similar findings were
reported in prior research that examined personality clusters in clinical cases of eating
disorders (Thompson-Brenner & Westen, 2005; Westen & Harnden-Fischer, 2001). This
converging evidence attests to the validity of these personality clusters in women with a
broad range of eating pathology. Furthermore, it indicates that the DSM-IV Axis I eating
disorder diagnoses may be deficient in yielding consistent predictions of clinical
variables such as level of functioning for individual eating disorders.

_F_amily RLhrtionships in Women with Subclinical Eating Pathology

The fourth overarching goal of the study was to investigate family relationships
across personality clusters to test whether the severity of clusters would mirror the level
of family disturbances. Overall, findings provided evidence for this hypothesis. Women
in the Dysregulated cluster perceived serious disturbances in their family functioning
across a number of domains. For example, they reported the least parental empathy,

warmth, and affection, and the most parental hostility and maternal overprotection

78

compared to other groups. Thus, these women appear to experience considerable family
conflict marked by parents who do not provide adequate care and by mothers who tend to
be intrusive and overinvolved in their daughters’ lives. Regarding paternal relationships,
Dysregulated women viewed their fathers as even more emotionally absent and
underinvolved in their lives than their mothers. As support for this, previous research
with BN women has shown that their fathers were perceived as distant (Sights &
Richards, 1984). Finally, women with Dysregulated personality pathology showed the
lowest levels of self-care, but the highest levels of self-attacking relative to other groups.
This suggests that Dysregulated women treat themselves in a manner similar to how they
have been treated by their parents.

These findings also suggest that Dysregulated women may have a comorbid
diagnosis of Borderline Personality Disorder (BPD). Consistent with previous research
on women with both eating disorders and BPD (Johnson et al., 1989; Wonderlich &
Swift, 1990), current findings showed that Dysregulated women perceived the highest
levels of parental hostility coupled with no family support compared to other groups.
Findings also fit with previous research on women with BPD and not eating disorders,
which indicated these women experienced the most maternal overinvolvement, paternal
underinvolvement (Soloff & Millward, 1983), and the least parental care compared to
patients with a range of psychiatric disorders and control patients (Goldberg et al., 1985).

Women with Rigid personality pathology generally perceived moderate family
disturbances across several dimensions compared to other groups. That is, they reported
notable deficits in parental empathy, warmth, and affection, but increased parental

hostility and overprotection. Therefore, like Dysregulated women, Rigid women

79

experience family conflict with both parents and inadequate parental care, but Rigid
women do not experience such extreme levels of these disturbances as Dysregulated
women. Unlike Dysregulated women, Rigid women do not perceive their fathers as less
involved or less caring than their mothers. This suggests the possibility that level of
paternal involvement and care may play a role in the etiology and/or maintenance of
eating and personality pathology, such that less paternal involvement and care is
associated with more eating and personality disturbances. Finally, women with Rigid
pathology perceived moderate impairment in self-care and increased self-attacking,
suggesting that these women treat themselves similarly to how they have been treated by
their parents.

Adaptive women showed the fewest and least severe family disturbances across
clusters, but still mild elevations compared to CON women. For example, they perceived
slight deficits in parental empathy, warmth, and affection, and increased parental hostility
and overprotection. These findings suggest that Adaptive women may experience less
frequent or less intense family disturbances, but nevertheless view their parents as at least
occasionally intrusive and not empathic. Adaptive women also demonstrated a similar
pattern as the other clusters, in that they tended to provide less self-care and more self-
attacking, suggesting that they treat themselves similarly to how their parents have
treated them.

The finding that levels of Parental Overprotection differed across groups fits well
with theories that posit parental overprotection as a component in the formation of AN
(Bruch, 1973; Minuchin et al., 1978). However, the finding in the current study is not

specific to AN, but rather to a group of women classified by level of personality

80

pathology, which included women with symptoms of AN and BN. Previous studies of
family relationships in women with eating disorders have not detected differences in
Parental Overprotection, but possibly because these studies did not examine personality
clusters of the eating disordered women; they only examined ED diagnoses (Bulik et al.,
2000; Palmer et al., 1988; Pole et al., 1988; Steiger et al., 1989). Thus, these findings add
to accumulating evidence that examining eating disordered women by levels of
personality pathology may be more useful than only considering diagnoses, particularly
when looking at family relationships.

Apalysis of Control Group

The exploratory piece of this study involved cluster analyzing the CON group to
replicate the three personality clusters (i.e., Resilients, Overcontrollers, and
Undercontrollers) that have been found in large samples of both children and adults.
These three clusters were developed, and they mirrored personality clusters found in the
eating disordered women along the lines of personality features, psychosocial
functioning, history of treatment, and family relationships. Specifically, Undercontrollers
most resembled women in the Dysregulated cluster, Overcontrollers were similar to the
Rigid women, and Resilients resembled Adaptive women.

These findings allow for stronger statements to be made concerning the centrality
of personality pathology in predicting clinically relevant correlates for eating disordered
patients and for possibly informing models of other forms of psychopathology as well.
Even though the eating disorder clusters exhibited higher levels of personality pathology
than the CON clusters, the key personality dimensions of ego-control and ego-resiliency

(Block & Block, 1980) appear to differentiate clusters across both sets of women. The

81

almost nearly identical results found with the CON clusters on level of functioning and
family disturbances provides more support that personality, and not eating
symptomatology, accounts for the significance of the external correlates in the current
study. These findings are also intriguing in suggesting that three personality clusters are
not specific to women with eating disorders and may be found in individuals with other
forms of psychopathology such as depression or anxiety, and that these clusters may have
clinical utility for these diagnoses. As support for this, nearly all women in the CON
group who reported a history of treatment indicated that they sought treatment for either
depression or anxiety. Finally, since these clusters have been found in a range of
populations, the clusters may predict psychosocial functioning and other relevant

variables in individuals with no formal psychiatric diagnoses.

Conclusions and Implications

Overall, this study demonstrated that three personality clusters are present in all
women with eating disorders regardless of severity of the eating disorder or treatment
status. These clusters also predicted important clinical variables such as psychosocial
functioning and were associated with specific family disturbances. These findings are
highly germane for the classification and treatment of women with eating pathology.
Importantly, the original eating disordered groups (i.e., R, B/P), unlike personality
clusters, did not predict psychosocial functioning or history of treatment. As both
previous research (Thompson-Brenner & Westen, 2005; Westen & Harnden-Fischer,
2001) and the current study have shown that personality clusters and not eating disorder

diagnoses predict clinical functioning, personality clusters appear to be robust entities

82

that have more clinical utility than Axis I eating disorder diagnoses. In addition, because
subclinical women fell into the same clusters as clinical women, this suggests that the
DSM-IV categories of AN and BN that putatively represent more severe forms of eating
pathology than EDNOS do not accurately reflect that distinction in severity. Findings
also revealed that personality was heterogeneous in the original eating disordered groups
and that clusters cut across these groups, suggesting that Axis I eating disorder diagnoses
include heterogeneous patients and may not offer reliable predictions of the etiology,
course, and treatment outcome for individual eating disorders. Therefore, one of the most
significant contributions of this study is that it suggests that the three personality clusters
are stronger indicators of the phenomenology and severity of eating disorders than the
traditional Axis I eating disorder diagnoses, even in women with subthreshold eating
pathology.

This study also found personality clusters in a non-clinical sample of women,
providing compelling evidence that clusters are not related to treatment-seeking in a
specific way. This is important because treatment-seeking women with eating pathology
have been shown to have higher levels of eating pathology, social impairment, and
personality disturbances than community samples (Fairbum, Welch, Norman, O’Connor,
& Doll, 1996). Thus, it is possible that previous research examining personality clusters
in women with clinical eating disorders has been confined to this subgroup. As
previously noted, due to the fact that such a small percentage of women with AN and BN
from the community ever seek treatment (Hock & van Hoeken, 2003), it is critical to
investigate the large group of women who do not seek treatment in order to understand

the extent to which personality clusters are relevant for all eating disorders.

83

'l
_ .
. .

 

The heterogeneity of personality characteristics in women with eating disorders
suggests that it is vital to identify the specific constellation of personality pathology
instead of only considering the Axis I eating disorder diagnosis. The direct assessment of
levels of personality pathology should be at the heart of sound clinical assessment of
eating disorders since they robustly predict psychosocial functioning and treatment
outcomes. The three personality clusters also offer an alternative approach to personality
diagnosis that may be more accurate than current Axis H diagnoses. The current Axis H
personality disorder (PD) diagnostic system has several limitations including high levels
of comorbidity between PD diagnoses and diagnostic criteria that have been difficult to
operationalize. Thus, it is possible that more unequivocal criteria could be derived from
personality clusters. Future research should examine if these clusters are in fact better
predictors of clinical variables than Axis II diagnoses. More specifically, research should
analyze women with eating disorders by Axis II personality disorder diagnoses and by
the three personality clusters to determine which diagnostic system better predicts
variables such as treatment length and outcome, and psychosocial functioning.

The assessment of the personality clusters identified in this study at the outset of
treatment may be useful. A practical rating instrument should be developed for use in
clinical practice and research that can quickly and reliably classify women according to
level of personality pathology. Current personality measures are time-consuming to
administer and contain redundant and superfluous items. A newly designed personality
assessment tool would remedy this problem and allow for the application of the

personality clusters in clinical practice.

84

Until such a tool is developed, the most parsimonious approach would be for the
clinician to assess the patient’s ego-control to determine if she is more prone to
undercontrol or overcontrol. As this study has shown, those women with higher levels of
undercontrol will present with the following personality and behavioral features: unstable
interpersonal relationships, emotional dysregulation, suicidal ideation, derealization,
disorganized thought processes, addictive behaviors, and hypervigilance. In addition,
they will demonstrate severe deficits in psychosocial functioning and likely come from
families with significant disturbances. By contrast, women high on overcontrol will
present with the following features: difficulties in sexual relationships and interpersonal
relationships, and low levels of stimulus seeking. Furthermore, they will evidence
moderate psychosocial difficulties and family disturbances. Finally, those with Adaptive
personality profiles will show few if any psychosocial deficits and family disturbances,
and will likely show increased anxiety, need for approval, and acting-out behaviors.

Following the assessment of personality pathology, the clinician will be able to
tailor treatment most appropriate to the level of personality pathology present. Such
treatment-matching is vital to the potential success of therapy, as certain personality
pathology will require the clinician to use specific therapeutic techniques to facilitate
decreased personality disturbances in the patient. According to previous research, eating
disordered patients who present with Adaptive personality profiles will recover at the
highest rate and in the shortest amount of time compared to Rigid and Dysregulated
women, who show lower rates and longer time to recovery, respectively (Thompson-
Brenner & Westen, 2005). Thus, cognitive—behavioral therapy (CBT) or short-term

psychodynamic therapy would appear to be well-suited for Adaptive patients, given that

85

these treatments are typically time-limited. In the therapy of women with Rigid
personality pathology, research (Thompson-Brenner & Westen, 2005) has demonstrated
the intriguing finding that clinicians identified as psychodynamic in theoretical
orientation tend to use more CBT techniques (e.g., being more directive and didactic) in
treating these women. By contrast, clinicians identified as CBT tend to use more
psychodynamic techniques (e.g., focusing on the therapeutic relationship and the
expression of the patient’s painful feelings) when working with Rigid women.
Interestingly, Thompson-Brenner & Westen (2005) also showed that both CBT and
psychodynamic clinicians working with Dysregulated women tended to use
psychodynamic techniques with these women. These findings suggest that a combination
of psychodynamic and CBT techniques may prove useful in effectively treating women
with Rigid and Dysregulated personality pathology. Nevertheless, future research should
seek to determine the most effective types of treatment for women with eating pathology
in each personality cluster. If in fact psychodynamic and CBT techniques are used
interchangeably in treating women with Rigid personality pathology, a manual should be
developed for this form of therapy and tested. Treatment studies should then compare
this novel treatment to more established therapies to evaluate its effectiveness.

Findings from the current study also shed light on the types of family
relationships to target in treatment and the expected responses that patients from each
cluster will have in working through these difficulties. For example, Dysregulated
women were shown to report the most parental hostility and maternal overprotection, and
the least parental caring and empathy. The nature of these family dynamics would be

critical to explore in individual and/or family treatment, and the greater intensity of

86

family disturbances will call for greater sensitivity and skill on the part of clinician in
working through these complicated issues. The level of severity in the family will
certainly influence the type of transference that develops in the treatment. For example,
if the patient has been verbally or physically abused by the parents and has frequently
withdrawn from them, it is likely that this patient will perceive some form of verbal abuse
from the therapist, and as a result, may shut down in therapy or cancel appointments and
avoid the therapist. An astute clinician would pick up on such a pattern and discuss this
with the patient in order to begin effecting change in this problematic pattern, although
the Dysregulated patient will likely have difficulty accepting and trusting the clinician in
general. Thus, these patterns will likely be frequently revisited in the therapy until
change occurs.

In cases of women with Rigid personality pathology, it is probable that they have
experienced a lack of Parental Care and increased Parental Overprotection, but they may
be better able than Dysregulated women to discuss these patterns. Thus, although the
patient may have difficulty at times expressing his/her self, the transference will probably
be marked by themes of approach and withdrawal on the part of the patient. The Rigid
patient may be more involved in the treatment than the Dysregulated patient and more
amenable to transference interpretations. Finally, Adaptive women will presumably
report the fewest family disturbances and will generally be able to trust the clinician and
express their feelings. These patients are the most likely to work through familial
conflicts in therapy in part due to the less severe nature of the conflicts and the absence of

more problematic personality pathology.

87

Limitations

Several limitations of the current study should be noted. First, although 795
women were originally assessed as part of this study, only 82 women met criteria for
inclusion into either the R or B/P group. The former group only consisted of 20 women,
whereas the latter group was made up of 62 women. In addition, once groups were
clustered, only 7 women were included in the Dysregulated cluster, 35 women in the
Rigid cluster, and 40 women in the Adaptive cluster. Thus, small sample sizes in these
groups likely limited power to detect significant differences across groups in some
analyses. Nevertheless, effect sizes were generally medium-to-large, indicating that
results were robust.

Second, this study was limited in that self-report measures rather than structured
interviews were used for classification of women into the R and B/P groups. Interviews
have been shown to be more reliable than questionnaires, especially for assessment of
binge eating (Wolk, Loeb, & Walsh, 2005). As a result of using self-report
questionnaires, some women in each of the groups may have been incorrectly classified.
Nevertheless, as the goal of the study was to include women with a range of eating
pathology and to primarily analyze personality clusters formed from these groups,
incorrect classifications into the original eating disordered groups were not likely to
detract from findings. Furthermore, even if individuals who did not actually have
significant eating pathology were incorrectly included in the original eating disordered
groups, fewer significant differences would have been found in this study. Regardless,
future studies should confirm personality clusters in women with subclinical eating

pathology using structured interview assessments.

88

Third, although the R group was constructed to approximate women with
Restricting Type AN, this latter diagnosis was not able to be directly examined in the
present study due to a significant minority of R women who exhibited low BMI. Thus, it
is unclear whether findings from the R group would also apply to women with subclinical
Restricting Type AN who have low BMI. Finally, the current study did not examine
eating disorder diagnoses per se. Future research should examine EDNOS to confirm the
presence of similar personality clusters and associated level of functioning and family
relationships in women with these formal subthreshold eating disorders. Moreover,
future research should investigate EDNOS alongside clinical eating disorders in order to
verify that subclinical and clinical eating disorders yield similar clusters based on

personality pathology, rather than severity of the eating disorder.

89

References

Allison, D. B., Kalinsky, L. B., & Gorman, B. S. (1992). A comparison of the
psychometric pr0perties of three measures of dietary restraint. Psychological
Assessment, 4, 391-398.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental
disorders (4th Text Revision ed.). Washington, DC: Author.

Andersen, A. E., Bowers, W. A., & Watson, T. (2001). A slimming program for eating
disorders not otherwise specified: Reconceptualizing a confusing, residual
diagnostic category. Psychiatric Clinics of North America, 24, 271-280.

Asendorpf, J. B., Borkenau, P., Ostendorf, F., & van Aken, M. A. G. (2001). Carving
personality description at its joints: Confirmation of three replicable personality
prototypes for both children and adults. European Journal of Personality, 15,
169-198.

Bacon, S. F., Collins, M. J ., & Plake, E. V. (2002). Does the Global Assessment of
Functioning assess functioning? Journal of Mental Health Counseling, 24, 202-
212.

Bagge, C. L., & Trull, T. J. (2003). DAPP-BQ: Factor structure and relations to
personality disorder symptoms in a non-clinical sample. Journal of Personality
Disorders, 17, 19-32.

Beck, J. S. (1995). Cognitive therapy: Basics and beyond. New York: Guilford Press.

Benjamin, L. S. (1974). Structural analysis of social behavior. Psychological Review, 81,
392-425.

Benjamin, L. S. (1996). An interpersonal theory of personality disorders. In J. F. Clarkin
& M. F. Lenzenweger (Eds), Major theories of personality disorder (pp. 141-
220). New York: Guilford Press.

Benjamin, L. S. (2000). SASB Intrex User’s Manual for short, medium and long form
questionnaires. Salt Lake City, UT: University of Utah.

Block, J. H., & Block, J. (1980). The role of ego-control and ego-resiliency in the
organization of behavior. In W. A. Collins (Ed.), Minnesota symposium on child
psychology (Vol. 13, pp. 39-101). Hillsdale, NJ: Erlbaum.

Brelsford, T. N., Hummel, R. M., & Barrios, B. A. (1992). The Bulimia Test--Revised: A
psychometric investigation. Psychological Assessment, 4, 399-401.

90

Brewerton, T. D., Hand, L. D., & Bishop, E. R., Jr. (1993). The Tridimensional
Personality Questionnaire in eating disorder patients. International Journal of
Eating Disorders, 14, 213-218.

Bruch, H. (1973). Eating Disorders: Obesity, anorexia nervosa, and the person within.
New York: Basic Books.

Bulik, C. M., Sullivan, P. F., Fear, J. L., & Pickering, A. (2000). Outcome of anorexia
nervosa: Eating attitudes, personality, and parental bonding. International Journal
of Eating Disorders, 28, 139-147.

Casper, R. C., Hedeker, D., & McClough, J. F. (1992). Personality dimensions in eating
disorders and their relevance for subtyping. Journal of the American Academy of
Child and Adolescent Psychiatry, 31 , 830-840.

Caspi, A., & Silva, P. A. (1995). Temperamental qualities at age three predict personality
traits in young adulthood: Longitudinal evidence from a birth cohort. Child
Development, 66, 486-498.

Cogley, C. B., & Keel, P. K. (2003). Requiring remission of undue influence of weight
and shape on self-evaluation in the definition of recovery for bulimia nervosa.
International Journal of Eating Disorders, 34, 200-210.

Costa, P. T., Jr., Herbst, J. H., McCrae, R. R., Samuels, J., & Ozer, D. J. (2002). The
replicability and utility of three personality types. European Journal of
Personality Special Issue, 16, S73-S87.

Crowther, J. H., Lilly, R. 8., Crawford, P. A., Shepherd, K. L., & Oliver, L. L. (1990).
The stability of the Eating Disorder Inventory. Paper presented at the Annual
Meeting of the American Psychological Association, Boston, MA.

Fairbum, C. G., & Beglin, S. J. (1994). Assessment of eating disorders: Interview or self-
report questionnaire? International Journal of Eating Disorders, 16, 363-370.

Fairbum, C. G., Welch, S. L., Norman, P. A., O’Connor, M. E., & Doll, H. A. (1996).
Bias and bulimia nervosa: How typical are clinic cases? The American Journal of
Psychiatry, 153, 386~391.

Fassino, S., Amianto, F., Daga, G. A., Leombruni, P., Garzaro, L., Levi, M., et al. (2003).
Bulinric family dynamics: Role of parents' personality--a controlled study with

the Temperament and Character Inventory. Comprehensive Psychiatry, 44, 70-77.

Fassino, S., Daga, G. A., Piero, A., Leombruni, P., & Rovera, G. G. (2001). Anger and
personality in eating disorders. Journal of Psychosomatic Research, 51, 757-764.

Fassino, S., Svrakic, D., Abbate-Daga, G., Leombruni, P., Amianto, F., Stanic, S., et a1.

91

(2002). Anorectic family dynamics: Temperament and character data.
Comprehensive Psychiatry, 43, 114-120.

Freud, S. (1896). Heredity and the etiology of the neuroses. In J. Riviere (T rans.),
Sigmund Freud collected papers (Vol.1 ). New York: Basic Books.

Friedman, M. A., Wilfley, D. E., Welch, R. R., & Kunce, J. T. (1997). Self-directed
hostility and family functioning in normal-weight bulirrrics and overweight binge
eaters. Addictive Behaviors, 22, 367-375.

Garfinkel, P. E., Modlofsky, H., & Garner, D. M. (1980). The heterogeneity of anorexia
nervosa: Bulimia as a distinct subgroup. Archives of General Psychiatry, 37,
1036-1040.

Garner, D. M. (1991). Eating Disorder Inventory-2 Professional Manual. Odessa, FL:
Psychological Assessment Resources.

Garner, D. M., & Olmsted, M. P. (1984). The Eating Disorder Inventory Manual.
Odessa, FL: Psychological Assessment Resources.

Goldberg, R. L., Mann, L. 8., Wise, T. N., & Segall, E. A. (1985). Parental qualities as
perceived by borderline personality disorders. Hillside Journal of Clinical
Psychiatry, 7, 134-140.

Goldner, E. M., Srikameswaran, S., Schroeder, M. L., Livesley, W. J., & Birmingham, C.
L. (1999). Dimensional assessment of personality pathology in patients with
eating disorders. Psychiatry Research, 85, 151-159.

Gunderson, J. G., Kerr, J., & Englund, D. W. (1980). The families of borderlines: A
comparative study. Archives of General Psychiatry, 3 7, 27-33.

Hathaway, S. R., & McKinley, J. C. (1943). The Minnesota multiphasic personality
inventory (Rev. ed., 2nd printing). Minneapolis, MN: University of Minnesota
Press.

Heatherton, T. F., Nichols, P., Mahamedi, F., & Keel, P. (1995). Body weight, dieting,
and eating disorder symptoms among college students, 1982 to 1992. American
Joumal of Psychiatry, 152, 1623-1629.

Herzog, D. 8., Keller, M. B., Lavori, P. W., & Ott, I. L. (1987). Social impairment in
bulimia. International Journal of Eating Disorders, 6, 741-747.

Herzog, D. B., Keller, M. B., Sacks, N. R., Yeh, C. J., & Lavori, P. W. (1992).
Psychiatric comorbidity in treatment-seeking anorexics and bulimics. Journal of
the American Academy of Child and Adolescent Psychiatry, 31 , 810-818.

92

Herzog, D. B., Norman, D. K., Rigotti, N. A., & Pepose, M. (1986). Frequency of
bulimic behaviors and associated social maladjustment in female graduate
students. Journal of Psychiatric Research, 20, 355-361.

Hoek, H. W., & van Hoeken, D. (2003). Review of the prevalence and incidence of
eating disorders. International Journal of Eating Disorders, 34, 383-396.

Hsu, L. K. G. (1996). Epidemiology of the eating disorders. Psychiatric Clinics of North
America, 19, 681-700.

Humphrey, L. L. (1986a). Family dynamics in bulimia. Annals of Adolescent Psychiatry:
Developmental and Clinical Studies, I 3, 315-322.

Humphrey, L. L. (1986b). Structural analysis of parent-child relationships in eating
disorders. Journal of Abnormal Psychology, 95, 395-402.

Humphrey, L. L., & Benjamin, L. S. (1986). Using structural analysis of social behavior
to assess critical but elusive family processes: A new solution to an old problem.
American Psychologist, 41, 979-989.

Johnson, C., & Bemdt, D. J. (1983). Preliminary investigation of bulimia and life
adjustment. American Journal of Psychiatry, 140, 774-777.

Johnson, C., Tobin, D., & Enright, A. (1989). Prevalence and clinical characteristics of
borderline patients in an eating-disordered population. Journal of Clinical
Psychiatry, 50, 9-15.

Kaye, W. H., Klump, K. L., Frank, G. K. W., & Strober, M. (2000). Anorexia and
bulimia nervosa. Annual Review of Medicine, 51, 299-313.

Kemberg, O. F. (1995). Technical approach to eating disorders in patients with borderline
personality organization. Annual of Psychoanalysis, 23, 33-48.

Kleifield, E. 1., Sunday, 8., Hurt, S., & Halmi, K. A. (1994). The Tridimensional
Personality Questionnaire: An exploration of personality traits in eating disorders.
Journal of Psychiatric Research, 28, 413-423.

Klump, K. L., Bulik, C. M., Pollice, C., Halrrri, K. A., Fichter, M. M., Berrettini, W. H.,
et al. (2000). Temperament and character in women with anorexia nervosa.
Journal of Nervous and Mental Disease, 188, 559-567.

Klump, K. L., Miller, K. B., Keel, P. K., McGue, M., & Iacono, W. G. (2001). Genetic
and environmental influences on anorexia nervosa syndromes in a population-

based twin sample. Psychological Medicine,31, 737-740.

Klump, K. L., Wonderlich, S., Lehoux, P., Lilenfeld, L. R., & Bulik, C. M. (2002). Does

93

environment matter? A review of nonshared environment and eating disorders.
International Journal of Eating Disorders, 31, 118-135.

Leary, T. (1957). Interpersonal diagnosis of personality: A functional theory and
methodology for personalin evaluation. New York: Ronald Press.

Leon, G. R., Fulkerson, J. A., Perry, C. L., & Cudeck, R. ( 1993). Personality and
behavioral vulnerabilities associated with risk status for eating disorders in
adolescent girls. Journal of Abnormal Psychology, 102, 438-444.

Leon, G. R., Fulkerson, J. A., Perry, C. L., Keel, P. K., & Klump, K. L. (1999). Three to
four year prospective evaluation of personality and behavioral risk factors for
later disordered eating in adolescent girls and boys. Journal of Youth and
Adolescence, 28, 181-196.

Leonard, 8., Steiger, H., & Kao, A. (2003). Childhood and adulthood abuse in bulimic
and nonbulimic women: Prevalences and psychological correlates. International
Journal of Eating Disorders, 33, 397-405.

Liebman, M., Cameron, B. A., Carson, D. K., Brown, D. M., & Meyer, S. S. (2001).
Dietary fat reduction behaviors in college students: Relationship to dieting status,
gender, and key psychosocial variables. Appetite, 36, 51-56.

Livesley, W. J ., Jackson, D. N., & Schroeder, M. L. (1992). Factorial structure of traits
delineating personality disorders in clinical and general population samples.
Journal of Abnormal Psychology, 101, 432-440.

Livesley, W. J ., Jang, K. L., & Vernon, P. A. (1998). Phenotypic and genetic structure of
traits delineating personality disorder. Archives of General Psychiatry, 55,
941-948.

McCabe, E. B., & Marcus, M. D. (2002). Question: Is dialectical behavior therapy useful
in the management of anorexia nervosa? Eating Disorders: The Journal of

Treatment and Prevention, 10, 335-337.

Milos, G., Spindler, A., Buddeberg, C., & Ruggiero, G. (2004). Eating disorders
symptomatology and inpatient treatment experience in eating-disordered subjects.
International Journal of Eating Disorders, 35, 161-168.

Mintz, L. B., & Betz, N. E. (1988). Prevalence and correlates of eating disordered
behaviors among undergraduate women. Journal of Counseling Psychology, 35 ,
463-471.

Minuchin, S., Rosman, B. L., & Baker, L. (1978). Psychosomatic families: Anorexia
nervosa in context. Cambridge, MA: Harvard U Press.

94

Murray, H. A. (1938). Explorations in personality. New York: Oxford Press.

Norman, D. K., & Herzog, D. B. (1986). A 3-year outcome study of normal-weight
bulimia: Assessment of psychosocial functioning and eating attitudes. Psychiatry
Research, 19, 199-205.

Palmer, R. L., Oppenheimer, R., & Marshall, P. D. (1988). Eating-disordered patients
remember their parents: A study using the parental-bonding instrument.
International Journal of Eating Disorders, 7, 101-106.

Parker, G. (1983). Parental Overprotection: A risk factor in psychosocial development.
New York: Grune & Stratton.

Parker, G., Tupling, H., & Brown, L. B. (1979). A parental bonding instrument. British
Journal of Medical Psychology, 52, 1-10.

Perkins, P. S., Klump, K. L., Iacono, W. G., & McGue, M. (2005). Personality traits in
women with anorexia nervosa: Evidence for a treatment-seekin g bias?
International Journal of Eating Disorders, 3 7, 32-37.

Placanica, J. L., Faunce, G. J., & Job, R. F. S. (2002). The effect of fasting on attentional
biases for food and body shape/weight words in high and low eating disorder
inventory scorers. International Journal of Eating Disorders, 32, 79-90.

Pole, R., Waller, D. A., Stewart, S. M., & Parkin-Feigenbaum, L. (1988). Parental caring
versus overprotection in bulimia. lntemational Journal of Eating Disorders, 7,
601-606.

Pryor, T., & Wiederman, M. W. (1996). Measurement of nonclinical personality
characteristics of women with anorexia nervosa or bulimia nervosa. Journal of
Personality Assessment, 67, 414-421.

Pukrop, R., Gentil, 1., Steinbring, I., & Steinmeyer, E. (2001). Factorial structure of the
German version of the Dimensional Assessment of Personality Pathology-Basic

Questionnaire in clinical and nonclinical samples. Journal of Personality
Disorders, 15, 450-456.

Raciti, M. C., & Norcross, J. C. (1987). The BAT and EDI: Screening, interrelationships,
and psychometrics. lntemational Journal of Eating Disorders, 6, 579-586.

Rammstedt, B., Riemann, R., Angleitner, A., & Borkenau, P. (2004). Resilients,
overcontrollers, and undercontrollers: The replicability of the three personality
prototypes across informants. European Journal of Personality, 18, 1-14.

Robins, R. W., John, O. P., Caspi, A., Moffitt, T. E., & Stouthamer-Loeber, M. (1996).
Resilient, overcontrolled, and undercontrolled boys: Three replicable personalin

95

types. Journal of Personality and Social Psychology, 70, 157-171.

Rorty, M., Yager, J., Buckwalter, J. G., & Rossotto, E. (1999). Social support, social
adjustment, and recovery status in bulimia nervosa. lntemational Journal of
Eating Disorders, 26, 1-12.

Schaefer, E. S. (1965). A configurational analysis of children's reports of parent behavior.
Journal of Consulting Psychology, 29, 552-557.

Sights, J. R., & Richards, H. C. (1984). Parents of bulimic women. lntemational Journal
of Eating Disorders, 3, 3-13.

Smith, M. C., & Thelen, M. H. (2000). Bulimia Test-Revised (BULIT-R). Retrieved July
20, 2003.

Soloff, P. H., & Millward, J. W. (1983). Developmental histories of borderline patients.
Comprehensive Psychiatry, 24, 547—588.

Steiger, H., Jabalpurwala, S., Champagne, J., & Stotland, S. (1997). A controlled study of
trait narcissism in anorexia and bulimia nervosa. lntemational Journal of Eating
Disorders, 22, 173-178.

Steiger, H., Stotland, S., Ghadirian, A. M., & Whitehead, V. (1995). Controlled study of
eating concerns and psychopathological traits in relatives of eating-disordered

probands: Do familial traits exist? lntemational Journal of Eating Disorders, 18,
107-1 18.

Steiger, H., Stotland, S., Trottier, J., & Ghadirian, A. M. (1996). Familial eating concerns
and psychopathological traits: Causal implications of transgenerational effects.
International Journal of Eating Disorders, 19, 147-157.

Steiger, H., Van der Feen, J ., Goldstein, C., & Leichner, P. (1989). Defense styles and
parental bonding in eating-disordered women. lntemational Journal of Eating
Disorders, 8, 131-140.

Stice, E., Presnell, K., Shaw, H., & Rohde, P. (2005). Psychological and behavioral risk
factors for obesity onset in adolescent girls: A prospective study. Journal of
Consulting and Clinical Psychology, 73, 195-202.

Strober, M. (1983). An empirically derived typology of anorexia nervosa. In P. L. Darby,
P. E. Garfinkel, D. M. Garner & D. V. Coscina (Eds.), Anorexia nervosa: Recent
developments in research (pp. 185-196). New York: Alan R. Liss.

Strober, M., & Humphrey, L. L. (1987). Familial contributions to the etiology and course

of anorexia nervosa and bulimia. Journal of Consulting and Clinical Psychology,
55 , 654-659.

96

Stunkard, A. J ., & Messick, S. (1985). The three-factor eating questionnaire to measure
dietary restraint, disinhibition and hunger. Journal of Psychosomatic Research,
29, 71-83.

Sullivan, H. S. (1953). The interpersonal theory of psychiatry. New York: Norton.

Swift, W. J., & Wonderlich, S. A. (1988). Personality factors and diagnosis in eating
disorders: Traits, disorders, and structures. In D. M. Garner & P. E. Garfinkel
(Eds.), Diagnostic issues in anorexia nervosa and bulimia nervosa (pp. 121-165).
Philadelphia: Brunner/Mazel.

Thelen, M. H., Farmer, J., Wonderlich, S., & Smith, M. (1991). A revision of the Bulimia
Test: The BULIT--R. Psychological Assessment, 3, 119-124.

Thelen, M. H., Mintz, L. B., & Vander Wal, J. S. (1996). The Bulimia Test--Revised:
Validation with DSM-IV criteria for bulimia nervosa. Psychological Assessment,
8, 219-221.

Thompson-Brenner, H., & Westen, D. (2005). Personality subtypes in eating disorders:
Validation of a classification in a naturalistic sample. British Journal of
Psychiatry, 186, 516-524.

Tozzi, F., Thornton, L. M., Klump, K. L., Bulik, C. M., Devlin, B., Fichter, M. D.,
Halmi, K. A., Kaplan, A. S., Strober, M., Woodside, D. B., Crow, 8., Mitchell,
M., Rotondo, A., Mauri, M., Cassano, G., Keel, P. K., Plotnicov, K. H., Pollice,
C., Lilenfeld, L. R., Berrettini, W. H., & Kaye, W. H. (2005). Symptom
fluctuation in eating disorders: Correlates of diagnostic crossover. American
Journal of Psychiatry, 162, 732-740.

van Kampen, D. (2002). The DAPP-BQ in the Netherlands: Factor structure and
relationship with basic personality dimensions. Journal of Personality Disorders,
16, 235-254.

Vitousek, K., & Manke, F. (1994). Personality variables and disorders in anorexia
nervosa and bulimia nervosa. Journal of Abnormal Psychology, 103, 137-147.

Wade, T. D., Bulik, C. M., & Kendler, K. S. (2001). Investigation of quality of the
parental relationship as a risk factor for subclinical bulimia nervosa. lntemational
Journal of Eating Disorders, 30, 389-400.

Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal of
the American Statistical Association, 58, 236-244.

Wear, R. W., & Pratz, O. (1987). Test-retest reliability for the Eating Disorder Inventory.
lntemational Journal of Eating Disorders, 6, 767-769.

97

Weissman, M. M., & Bothwell, S. (1976). Assessment of social adjustment by patient
self-report. Archives of General Psychiatry, 33, 11 1 1-1 115.

Welch, G. (1988). Selected multivariate statistical techniques and eating disorders.
Unpublished doctoral dissertation, University of Otago, New Zealand.

Westen, D. (1998). The scientific legacy of Sigmund Freud: Toward a
psychodynamically informed psychological science. Psychological Bulletin, 124,
333-371.

Westen, D., & Harnden-Fischer, J. (2001). Personality profiles in eating disorders:
Rethinking the distinction between axis I and axis H. American Journal of
Psychiatry, 158, 547-562.

Westen, D., & Shedler, J. (1999a). Revising and assessing axis 11, Part I: developing a
clinically and empirically valid assessment method. American Journal of
Psychiatry, 156, 258-272.

Westen, D., & Shedler, J. (1999b). Revising and assessing axis II, Part II: Toward an

empirically based and clinically useful classification of personality disorders.
American Journal of Psychiatry, 156, 273-285.

White, J. H., & Litovitz, G. (1998). A comparison of inpatient and outpatient women
with eating disorders. Archives of Psychiatric Nursing, 12, 181-194.

Wolk, S. L., Loeb, K. L., & Walsh, B. T. (2005). Assessment of patients with anorexia
nervosa: Interview versus self-report. lntemational Joumal of Eating Disorders,
37, 92-99.

Wonderlich, S., Klein, M. H., & Council, J. R. (1996). Relationship of social perceptions
and self-concept in bulimia nervosa. Journal of Consulting and Clinical
Psychology, 64, 1231-1237.

Wonderlich, S. A., Brewerton, T. D., Jocic, Z., Dansky, B. S., & Abbott, D. W. (1997).
Relationship of childhood sexual abuse and eating disorders. Journal of the
American Academy of Child and Adolescent Psychiatry, 36, 1107-1115.

Wonderlich, S. A., & Swift, W. J. (1990). Borderline versus other personality disorders in

the eating disorders: Clinical description. lntemational Journal of Eating
Disorders, 9, 629-638.

98

APPENDICES

99

APPENDIX A

Tables

100

Table l

Descriptions of DAPP-BQ Dimensions

 

 

Dimension Description
Affective Instability Affects are easily aroused and unstable
Anxiousness Trait anxiety
Callousness Interpersonal insensitivity
Cognitive Distortion External world is experienced as unreal; unusual patterns of
speech and illusions occur
Compulsivity Focus on order and precision

Conduct Problems
Identity Problems
Insecure Attachment
Intimacy Problems
Narcissism
Oppositionality
Rejection

Restricted Expression
Self-harm

Social Avoidance
Stimulus Seeking
Subrrrissiveness

Suspiciousness

Antisocial and addictive behaviors; acting-out behaviors
Disturbance in self-concept

Concern with separation and loss

Inhibited sexual expression and inability to form close
relationships

Need to seek attention and receive approval and admiration
Disorganization, indecisiveness, and oppositional behaviors
Hostile and judgmental behavior with extemalization of blame
Reluctance to share feelings and to be open with others
Self-injurious behaviors

Social apprehensiveness

Sensation seeking and recklessness

Dependence on others and suggestibility

Interpersonal mistrust

 

\Note. DAPP-BQ = Dimensional Assessment of Personality Pathology - Basic

Questionnaire.

101

Table 2

Internal Consistencies of the Eating, Personality, Family Relationship, and Level of
Functioning Variables

 

 

Variable N a
EDI-2
Drive for Thinness 191 .97
Body Dissatisfaction 191 .97
BULIT-R Total Score 191 .97
TFEQ Restraint 191 .94
DAPP-BQ
Subrrrissiveness 191 .92
Cognitive Distortion 191 .92
Identity Problems 191 .95
Affective Instability 191 .93
Stimulus Seeking 191 .88
Compulsivity 191 .91
Callousness 191 .82
Intimacy Problems 191 .87
Anxiety 191 .96
Conduct Problems 191 .83
Suspiciousness 191 .93
Social Avoidance 191 .94
Narcissism 190 .87
Self-harm 191 .95
Restricted Expression 191 .90
Oppositionality 191 .89
Rejection 191 .84
Insecure Attachment 191 .94
PBI
Maternal Care 190 .96
Maternal Overprotection 190 .89
Paternal Care 188 .96
Paternal Overprotection 188 .86

 

Note. EDI-2 = Eating Disorder Inventory-2; BULIT-R = Bulimia Test-Revised; TFEQ =
Three-Factor Eating Questionnaire; DAPP—BQ = Dimensional Assessment of Personality
Pathology-Basic Questionnaire; PBI = Parental Bonding Instrument.

102

Table 2 (cont’d).

 

 

Variable N a

SASB
Introject 191 .71
Mother Focuses on Daughter 189 .84
Mother Reacts to Daughter 189 .78
Daughter Focuses on Mother 189 .82
Daughter Reacts to Mother 189 .77
Father Focuses on Daughter 187 .81
Father Reacts to Daughter 187 .79
Daughter Focuses on Father 187 .81
Daughter Reacts to Father 187 .81

SAS-SR
Work as a Student 188 .78
Social and Leisure Activities 187 .76
Relationships with Extended Family 180 .72
Marital/Partner Roles 77 .74
Overall Adjustment 71 .90

 

Note. SASB = Structural Analysis of Social Behavior; SAS-SR = Social Adjustment
Scale-Self-Report.

103

.5. v a...
Egg-=8 N «am "econ—cogs 2:08.: n E

”83365: n HZ ”00:36; 368 u now ”wwuemsofiawem u mam ”mac—pox“ 83:50 u :00 ”.9258 n x=< E2838 u Hem

M80305 hose—:5 u 9: ”5:22:85“? u 50 5852230 n EU "comma-axe 38:52 .I- 5% 5363588 n :80
"web—8m 2:256 M 8m ”5:58.05 ciao-eta n c< $80305 5:52 u E: 25:53.6 93ch0 u woo
”805336-58 u new 652.5; M E> ”23585020 Ewan-30.05am 3:28qu we EoEmwomw< 556585 n Om-mnia .262

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

*Nv. co; - - - - - -- - - - -- - - - - - - 3
S. ... 5 m. 50,— - - - l - - - - - - - - -- - - HeZ
*3. *9”. mm. 5; - -- - - - - - - - - - - - -- mom
*5 . *vm. ism. *3. on: - - - - - - - - -- - -- - - mam
...N-m. 3. m5. 5.- ... L m. co; - -- - -- - - - - - - - - :00
*3. *mm. em. *8. *8. mo: co; - - - - -- - - - - - - xe<
mm. *mm. *3. S .- iv. 1%. oo 54 - - - - -- - -- - - -- Um
I. .8...- icmu EN. 5.- co.- 2. 5.- co; - -- - -- - -- - - -- 5
...NM. ...:- 3. 3m. in am. .am. mm. 8.- o3 -- -- -- -- -- -- -- -- .30
mm. on. *9». L _. *3. *mm. L m *mm. 8. *om. co; - - - - - - -- RU
.Lm. 5.- 3: *mm. mm. 5.- *3. mm.- 50. cm. 5. co; - l -- - - - EM
2. on. 2. cm. ism. 2.- *am. mo. 5. mm.- 5. 3. co; - - - - - :50
on. 5.. E. on: 5. *3. 5.- 0m. co. *3. *mm. 5.- 3.- co; - -- - - 8m
*3. ...:u 3v. Km. .55 mm. in. Re. 2.- *3. *5. 3. ...am 5. co; - - -- we.
*wm. ...ov. MS. *mn. ...mo. 9. .43.. 55. em. *wm. S. ta. 2 5. *5 co; - - am
*5. 5m. 5.. *5. *mn. mm. *3. mm. 9. *3. .33. Re. MN. ON. *5. *3- 54 - #100
*3. *am. a. *3. ...—v. 5.- ...-mm. mm.- co. *Nm. 5. *mm. .Lm. mo: *5. *mm. 4mm. 5; saw
«cm 5 E Z mom mam :oU x=< EM 9: “50 80 Ex :50 8m a< m2 woo new S>

 

 

an n 5 $835 5:56 ate 38323.6

m 033.

 

Table 4

Direct Oblimin Rotated Principal Components Factor Loadings of 18 DAPP-BQ
subscales Across R (n = 20) and B/P (n = 62 ) Groups: F our-Factor Solution

 

 

Factor

Variable 1 2 3 4

Identity Problems .91 .08 .15 -.06
Social Avoidance .84 -. 16 .10 .04
Anxiety .83 -.09 -.04 .19
Cognitive Distortion .80 .27 .05 .03
Submissiveness .75 —.28 -.06 -.03
Oppositionality .68 .22 -.20 -.61
Affective Instability .66 .25 -.33 .17
Suspiciousness .56 .36 -.09 .35
Self-harm .52 .39 .12 . 15
Conduct Problems -.02 .79 .05 -. 13
Callousness .07 .75 -.03 .04
Rejection -. 14 .73 -.33 .23
Stimulus Seeking -.08 .68 .13 -.28
Intimacy Problems .14 .13 .92 .13
Restricted Expression .58 -.01 .69 .06
Insecure Attachment .55 .01 -.67 .08
Narcissism .19 .24 -.57 .14
Compulsivity .14 -.08 .01 .83

 

Note. Loadings greater than or equal to .40 appear in bold. DAPP-BQ = Dimensional
Assessment of Personality Pathology-Basic Questionnaire; R = restrictor group; B/P =
binge-purge group.

105

Table 5

Direct Oblimin Rotated Principal Components Factor Loadings of 18 DAPP-BQ
subscales Across R (n = 20) and B/P (n = 62 ) Groups: Three-Factor Solution

 

 

Factor

Variable l 2 3

Identity Problems .90 -.19 .11
Anxiety .88 .04 -.15
Cognitive Distortion .84 -.05 .25
Social Avoidance .84 -.15 -.16
Affective Instability .73 .38 .16
Suspiciousness .72 .21 .21
Submissiveness .69 -.02 -.25
Self-harm .62 —.04 .31
Oppositionality .47 .03 .44
Compulsivity .43 .20 -.39
Intimacy Problems .24 -.86 .10
Restricted Expression .63 -.69 -.01
Insecure Attachment .54 .66 -.04
Narcissism .24 .62 . l6
Conduct Problems .03 .03 .78
Stimulus Seeking -.10 -.11 .74
Callousness . 17 .14 .69
Rejection .01 .49 .58

 

Note. Loadings greater than or equal to .40 appear in bold. DAPP-BQ = Dimensional
Assessment of Personality Pathology-Basic Questionnaire; R = restrictor group; B/P =
binge-purge group.

106

Table 6

Direct Oblimin Rotated Principal Components Factor Loadings of 14 DAPP-BQ
subscales Across R (n = 20) and B/P (n = 62 ) Groups: Three-Factor Solution

 

 

Factor
Variable Neuroticism Psychopathy Intimacy Problems
Identity Problems .90 .08 .19
Anxiety .87 -.15 -.02
Cognitive Distortion .85 .25 .07
Social Avoidance .83 -.17 . 12
Suspiciousness .74 .23 -.19
Affective Instability .73 .17 -.34
Submissiveness .67 -.21 -.02
Self-harm .65 .35 . 15
Compulsivity .44 -.35 -.22
Conduct Problems .05 .82 -.03
Stimulus Seeking -.10 .76 .04
Callousness .17 .68 -.25
Intimacy Problems .27 .04 .83
Narcissism .20 .17 -.76

 

Note. Loadings greater than or equal to .40 appear in bold. DAPP-BQ = Dimensional
Assessment of Personality Pathology-Basic Questionnaire; R = restrictor group; B/P =
binge-purge group.

107

.5. v a... .8. v a...
.005 Soto H B A2222>0u 23:83 £80.: 0.0 0028/ 550% 32:00 u 200 £03.20 Ewt u DOM £0620 03.980
n mQ< £0520 60.25063 N m>D ”0205—26030 23932055 b=§ofi0m .6 2083003.. 3:22.085 .I- Ominzfl .082

 

:N ......ZOU A GUM
mm; **mQ< A GUM
amfi ......ZOU A m>Q
wm.— ......QOM A m>Q
3mm 3&3 A m>Q _OO.V N55 Amwd—v mus-6N 80.0 hmfio CV63 omdm Aomdv moém mace—no.5 5982

Kb; ......ZOU A QOM
mm; sauna A Dam
NNd ......200 A m>Q

Om; ......QOM A m>Q 9.02205
omd Inn? A m>D 55V 3.9. 6.13 3.3 85.53 00.5 AWN-.3 56¢ GYE 3.5 032:on
om. *ZOU A An?

mwé ......ZOU A GEM
mm; .533 A QOM
5N ......ZOU A m>Q

108

 

S mom A $5
03 in? A $5 8v 8% 625 3.2 2.3 :8 33: 8.2 See 8.; 30523553
0 285258 a $2.? 6580 833329 Ema Eagle. 259$

 

83
u 5 20D 23 203.59 R u 5 30833ng BS» .«wm u 5 33m .8» n 5 09:303.. 9523.‘ 830.33% @505va S 000=0x0§~Q =00:

w 033.

Ac. V Q: .mo. V R."
.005 000:0 u 3 5:02:50 Eavesmv 5:008 000 5020> 95% 35:00 u 200 £0520 Ewt .I. COM £0520 02330
u an? £0520 00533530 n m>Q ”020525020 050m4Awo—oz5m bzwcoflfi mo E08505< 3:055:05 u Ominzn .082

 

:u **ZOU A QOm
gm. 5&3 A COM
mo. *m>Q A GUM ~OOV Omd 65.8 hodm Aomfiv wlfim Sn; 3 QVSM 80.3 awwN 55030.5 how—:9:

$02 **ZOU A m>Q
MNA *vam A m>Q

NOA “wag A m>Q _O.V mm.m ANQSV :dm Ana; 3 Vfimv Ahmfiv “5.2.” 898 owNn 30:30:00

QVA **ZOU A WWQ

ho. DOM A m>Q

~02 "Inna A WWD ~8.V 5N6 A©~.— C hhdv 2‘de mmdc Chum: 2.4% 26.93 ovdv 0336:5500
.. 5: no. hmd 30.3 00.3 $663 mvfiv CNNC vodm Aomd—v wad? wEv—oom man—Sum

Om. *ZOU A m%

cm; ..iZOU A GUM
ma. flag A GUM
NWN ......ZOU A m>D
mmN .10va A m>Q
3A” 5&3 A WFQ ~8.V mmdm AwN.NC omdm Amvfiv mméh 3 fig _m.vm AhNd—v Nmfiv 5:532: o>fl8t<

 

.0 28508.00 a 6:.me .8250 32:58.58 Ema. 03530.. 233$

 

2.0.880 n 0:3

109

.8. v a: .8. v 3.

.3; Soto u n .Gcozagoc 2.3.83 £82: Eu 825/ .msew 65:00 n 200 £03.20 Emu H GUM £0520 03:33
N ma roams—o toga—=35? H m>Q ”83585030 o_mam->wo_ofimm b=mcombm «o Eofimmomm< 356585 N Om-mm<Q $32

 

 

«2 £28 A com

mo; **m% A GUM

:mm **ZOU A m>Q

8a ......nom A $5

$4. 3%? A m5 89v 8.2. A58 38 and 3% £5 2.2 $03 3.3 aussosawsm

we; ......zou A m3.

3. 1%? v mom

:a ......zoo A 85

5: 38m A $6

3. m9. A $5 Sov PM: GE 3mm 83 :3 $9 8.5 $6 2.2 €285 9250

me. ..iZOU A mg

Sm ......zou A mom

omn— **n—Q< A DOM

SN ......zou A 95

gm max A $5

m3 2.9.. A $5 _8.v 8.3 $3: R2.” 8va 8.? $9 3.8 :3 a 8.3. 525%
n mcacfleou a :Edm .835 Bfiameao 35 93%.... 03%;

 

3.253 n 2.3.

110

.8. v at. .8. v a...

.86 “onto u n .Amcoufififi 93:53 282: 8a mo:_m> .9.on 35:8 u ZOO cams—o Emu n mom CBS—o 03:53“
N mQ< cuss—o Baa—$893 u m>Q ”88505050 Baum—-mwo—ofimm bzacoflom ho EoEwmumm< awe—822:5 u Omdmad $82

 

 

mu. ......zou A com

mam ......zou A $5

$4 ......aom A $5

and 1%? A ms 53 3.8 a: v3 83 BE 83 mi 2.3 w: 5.353
:2 ......zou A an?

5w. **ZOU A DOM

Om.N “IL/HOD A m>D

a2 thou A 95

w? inn? A $5 Sov 3.: €an 32. ES: mic and: 8% A38 omen 822032

52 £28 A mom

av; tin/x A GUM

th ..iZOU A m>D

ma. com A $5

m2 3%? A 95 39v ~03” a3: 8.2 5.8 mtmc 8o.— : Sam and 2.2 35662 :68
n 285%.5 m 5:.me .825 83%me Ema 223% 033$

 

33:58 N. 255.

111

.8. v A; .8. v a...
.58”... .8 n z.
.3: .880 u a $8.338 26:33

2:35 2a mos—«5 .889: was:— xuonl I lam .Eaccoamoso wcsmm LoaoaméoLF—II I Ommh. .BmSoMILmoH «EL—Emu MIPSDm

 

"N46952: 5985 macaw" N Ham Loan—o Ewan DOM cams—o o>zmmcal an? .5820 3833.53" m>Q $82

 

 

m: m». am. so. a 8. mm AP. 8 8. 8 GL c 8. 8 Em
mm. mm? v mom
2.. com A 95
mm. LQ< A $5 mm. 8. 828 8.2 9.8.3 and 83¢ 8.: 858m
dlH ME..-
8. :82 A max
3.. 99. A $5
8. 1%? A man 8v :8 $3: 8.8 83$ 8% A883 8.8 28m Bop
”ImEISm.
S. 15¢. A com
8. com A $5
8. an? A $5 SV 88 ewe 9.8 813 8:8 :03 8.8 Sauemsamfi
ram
9.. com A 95
mm. mm? A $5 mm. 8. a; 8.: 5.8 8.2 82.8 8.: ”8:55.
.5“ Erran—
”30m
8 28538 a 83E 833355 Ems. 2,32% 052$

 

23.35 «N u E 3&53388 .33 .«wm. u S 3E2 .8» u 5 $53< .3653“ Esgam LMPEQQ mama S 39:3»wa as»:

w 2an

112

.8. v A... .8. v a...

AVmfimvnr we u 2.. .mm H 2..
.38 Soto H V Amp—28300 Eaucfimv 0:008 08 823/ 95% .8250 n 200
£0820 Emu H DOM £08.20 020960 n. ma £0820 wouflzwemxc u m>Q ”tomomfiomI0—a0m E0583€< 8.2% u Mm-m<m .082

 

 

 

mm. ......ZOU A mg
mm; ......200 A com
9.. me. A com
hm; ......ZOU A m>Q
3. 90m A m>Q bgm voucouxm
8. :5... A $5 83 N3: 63 2.. 2 _ .0 mm. E: a... 88.0 9.. 5.3 852.03%
3“. zoo A L9.
83 ......zou A mom
8. .39.. A com
0.2 $200 A $5
No. DOM A m>Q 323004..
a: Ian... A $5 83 2.8 a: 83 33 SN 83 M3 83 88 2:23 93 38m
mm. zoo A .E<
m2 ......zou A com
3. 2%? A com
3: ....200 A $5
8. com A $5
mm; 1.5... A 95 83 .32 ..E: 9.. 33 mm. .83 mm. 83 3.. is,» “swam
N. 3859.80 m :2.me .2080 0222323 208 2,38,... 22.55
:28... 82

u S 3:200 33 203.59 C n 5 30333.28 350 .«wm H 5 ENE .3». H E 0>LLQ§< 000L0< 0030.33,». Marga. S amuzmngQ :3:

a 030,—.

113

.mo. V 3.... .mo. V R...

.Ao@.mvn~= do H :0 .m H :0 .2 H 2.. .mN H :0 AVwfime .wo— H :0 .mm H ta
.008 000:0 u n 8.530300 082.83 8:008 0.8 8020> 25% 3250 n 200
£0820 Emu n 90m £0820 023200 N mQ< £0820 00023383 .I. m>Q €80M£0m080m E0882o< Eoom n Mm-m<m .082

 

 

mm. .28 A 09..

m..— **ZOU A GUM

8. $89.. A oom

$8 ......zoo A $5

8. oom A $5

80 3&9. A ES 83 8.8 83 m: 83 8;... 83 mom 83 v3 288 3.8
3.. zoo A 09...

8. zoo A com

83 zoo A $5

2.. oom A go 830

E. .82 A $5 8v =80 0:3 9.. .33 mm. .83 80... .83 3.. 025.5352
0 885058 ~ $2.80 Sacco 83:02.5 208 90.83. 0325»

 

8.2.80 8 2.3

114

Table 10

Hierarchical Multiple Regression of R and B/P Groups and Personality Clusters on SAS-
SR Student Work (N = 80), Social and Leisure Activities (N = 82), Relationships with
Extended Family (N = 82 ), and Total Score (N = 82 )

 

 

Variable B R R7- ARZ
Student Work
Step 1 -- .20 .04 .04
R and B/P Groups .20 -- -- --
Step 2 -- .46 .21 .17**
R and B/P Groups .14 -- -- --
Rigid Cluster .39** -— -- --

Dysregulated Cluster .28* -- -- --

Social and Leisure

Step 1 -- .12 .01 .01
R and B/P Groups .12 -- -- --
Step 2 -- .45 .20 .l9**
R and B/P Groups .04 -- -- --
Rigid Cluster .37** -- -- --

Dysregulated Cluster .35** -- -- -_

Relationships with Extended

Family
Step 1 -- .11 .01 .01
R and B/P Groups .11 -- -- --
Step 2 -- .30 .09 .08*
R and BI? Groups .06 -- -- --
Rigid Cluster .21 -- -- --
Dysregulated Cluster 25* -- —- --
Total Score
Step 1 -- .16 .03 .03
R and B/P Groups .16 -- -- --
Step 2 -- .49 .24 21 **
R and B/P Groups .09 -- -- --
Rigid Cluster .39** -- -- __

Dysregulated Cluster .38** -- -- --

 

Note. R = restrictor group; B/P = binge-purge group; SAS-SR = Social Adjustment Scale-
Self-Report. Adaptive cluster was used as reference group and not included as a
predictor.

*p < .05. **p < .01.

115

Table 11

Logistic Regression of R and B/P Groups and Personality Clusters on History of
Psychiatric Medication (N = 79)

 

 

Variable 12 df p Wald df p
Step 1 4.85 1 .03 -- -- --
Block 4.85 l .03 -- —- --
Model 4.85 l .03 -- -- --
R and B/P Groups -- -- -- 3.09 1 .08
Step 2 5.64 2 .06 -- -- --
Block 5.64 2 .06 -- -- --
Model 10.49 3 .02 -- -- --
R and B/P Groups -- -- -- 2.45 1 .12
Rigid Cluster -- -- -- 4.26 1 .04
Dysregulated Cluster -- -- -- 2.92 1 .09

 

Note. R = restrictor group; B/P = binge-purge group.

116

.mo. V has." .mc. V R...

AVwfimVnr .2: u 2.. .o n 2a

.008 Soto H V 3.52030“. 358V 2.00:. 0.3 0033/ .95.»

3.950 n ZOO £0820 2w... u DOM £0820 033200 N MD< £0820 “.80—sweat ..u m>D ”0:08.505 wig—om 3:005 N am .032

 

117

 

om. ZOO v MD<

om. ZOO v DOM

mm; .950 v m>D

om. DOM v m>D

an. MD< v me 5V KEV 93:6 2.0m «A323 9.2 Amadv omdm 362$ mm.mm 080 EEBNM

Vm. ZOO A MD<

mo. “1.200 A DOM

om. MD< A DOM

co; ......ZOO A me

G. DOM A m>D 55008930

8. an? A $5 Sov 8.2 88.8 8.: 83V 58 A300 2.2 84.80 82: 38%:

av. ZOO v DOM

ow. MD< v DOM

o: ...ZOO v m>D

mm. DOM v m>D

co; Mn? v m>D 3V hmé vav 8.0m A363 QNMD Ammdv vfivm 28C 3.3 0:00 36082
B 8:859:00 m 336E .8280 00025000me 23M 030903.. 030ta>

 

.0233
8.2 u 5 zoo as 223.0 o u 5 003302an Be .5 u 5 385 .8» u s 030% 3&3. $8330. :2 s §§§Q :8:

ND 030%

AC. V RI. .mo. V Q...

Avwfimvnr .2: u 2.. .0 n ....

.008 00020 n .0 Amp—28300 220532 8:00.: 000 802.3 95%

30:50 N ZOO £0820 2m: n DOM £0820 03.960 u MD< £0820 0002:3820 u me £58285 wcficom 3:02AM n EM .082

118

 

mm. *ZOO A MD<

 

N0 .28 A no”.

8... zoo v 8.5

8.. oom v 8.5 808.8805

a... 09. v 8.5 5v .888 988.00 mm... .830 S... and 8.2 805 3.: 3832
n 285058 a. 6.2.00 .828 83:00.85 20:. 260%... 0.9%.,

 

20.853 NJ 030,—.

.8. v ...... .8. v a...

.3” u c... A389"! .3 u c.
.008 .0020 H V .Amcouflgo 00.20503 8:008 000 32.3 doom.“ 3.550 n ZOO
£0820 2w: u DOM £0820 032.2200 M MD< £0820 0002:3820 n me £o_>0:0m 300% Do 8820.5. 2020.5m n Dm<m .082

 

 

 

G. ......ZOO A MD<
NV. ZOO A DOM
cm; .1200 A myth—
NND ...DOM A m>D wExbsom %
mm. MD< A me _oo.v ..vad Good 3.8 $8va owdm Awww: Eda «AmwNNV mmdm mix—am 8:00.02
mm. *ZOO A MD<
mo. ...ZOO A DOM
RA ......ZOO A m>D
he. DOM A m>D wag—m 0%
we. MD< A m>D Doov amos $3 3. 85 $2 Dog 2: «83 0o. wEE=0m 35082
mm. ZOO v MD<
on. ......ZOO v DOM
cm. MD< v DOM
no. ZOO v m>D 00:02:: Z 0%
om. MD< v m>D 5V wmd Ania: 2.65 3.00.0 hwmm Amodmv some 80.0: 3.2. 289:”:— 35082
n 885058 m 62.0”. .380 8330883 Ewe. 95.83. OBES
5on:

8.2 u ... zoo as $0.5 2 u 5 $338283 as. .23 u s 30.2 .8» u s 0335.... est... romeo ma 2.. 8820.03 so:

2 030M.

119

40. V 2*... .mo. V R...

.wm n 2.. Amwfimvn—a .mm H :0
.008 .0020 u 00 8:09.300 0:20:83 8:00.: 0.8 802.3 95% .0550 u ZOO
00.3.0 20... u now 00.3.0 2.8.3.. u 09. 02.8.0 33:00.83 u m»: 53200 .288 8 8.2.32 .8338 n 08.8 .032

 

 

Nw. ZOO A MD<

we. ZOO A DOM

No.2 ZOO A m>D

we. DOM A m>D m2=o00M 0%

mm. MD< A m>D mov ..me.m Gee mm. Ammo we; 2: me. 025 me. 38080..— 8:00:02

we. ......ZOO A MD<

oe. ...ZOO A DOM

wa. ZOO A m>D w:e00.a0M 0%

em. DOM A m>D 3V ..woe Aeme om. Aeme mm. 23 ee. «$5 on. w:200:< 85082

ww. ...ZOO A MD<

mm. ...ZOO A DOM

ea. ...ZOO A m>D

0w. DOM A m>D graham 0%

em. MD< A m>D 5V name Anedmv seem demv omdw 020;: Dwdm Cod: 2.0m w:o=:m bbEwaaD

:m ......ZOU A nun—400

mm. ......ZOO A DOM

MN; ......ch A m>Q wag—m

Dw. DOM A mWD 0% 0002:023—

ww. MD< A m>D .8.v mead— Amee mm. Awee 1m.— uAme.v we; gee we. 200.325
0 8:85.058 0 3.2.8: .8250 33:02:: .08.: 2.8.03. o.fifi>

 

3.880 2 038.

120

.8. v 5.... .8. v a...

can n =0 Amwfimvmo .om u a.
.007. 000:0 n V Amp—200330 3:33 0:005 000 0023/ .95—w 35:00 N ZOO
£000.20 Emu u DOM £00020 023on u MD< £8030 o0§.=w2@€ u m>D cogs—Dom 300% Mo max—22 .8325m u mm<m .032

 

 

o... zoo A 80

3. zoo A $5

2.. 80 A $5 050505 a. to

3. ARE A 95 BV 230 83.3 3.2 839 :9. $05 $8 his 0.08 05:55 26%:

S. zoo A 09..

we. ......zoo A oom

mw. zoo A $5 wéfiwz a.

0m. 03. A $5 8v .2 .m :8 E. :3 mm. 3.5 MK. .50 S. 0:080; 35032

9.. zoo A 09..

N0. .28 A mom

:4 ...zoo A $5 3:080

wv. 000 A $5 a. 050885

m0. “:2 A $5 av .23 :3 3. $3 mi .32 a. $5 on 362053

mm. **ZOU A ma

3. zoo A mom

mm. 09.. v mom 0508.5

03 ...zoo A $5 a. 0:283

vm. o9. A $5 89v Amos 83 R. :3 a. .50 a. $5 3.. €9,025
0 285055 m 62.00 ..SSo Baswoio 230 90.52 03%;

 

.5380 Q 0:3

121

.Do. V 53.. .mo. V A...

.Amefivm0 .2: n 2... .0 n =0 .va n =0 Amwfimvnr .mm n :0
.006 00800 n h .Amcocfifio oaocflmv 0:008 000 0023/ .9.on .80—50 H ZOO

033.0 20.: u 000 03020 2,0080 u .09. 03020 02030203 u 005 02550 E80 00 mini 33250 n 00% .032

 

 

_0. ......zoo A AE<

00. “:2 v now

no. ZOO A m>D wins—D 0%

00. oom A 005 8v 000.0 .200: 8:2 020.08 00.3. Davms 00.8 80.08 00.0w 0:50:00 .0828

5. zoo v 005

E. mom v 005 85:35 z a.

00. 0% v 005 cm. .00; .503 00.00 $8.5 2.00 .258 00.00 20.5 910 200050 020200

R. zoo A “:2

2.. ......zoo A mom

am. an? A com

03 ...zoo A 005

cm. DOM A m>D was—505 0% tO

00. 03. A 005 80v 002. 80.80 2.2 80.3 00.00 060.08 00.0m 3.0.80 2.3 05:03 .zEzwao

0m. zoo A 09..

N0. ...zoo A com

0m. 09.. A mom

mu; ......ZOO A m>D w5000—w0Z

on. DOM A m>D 0% wctocwD

N: £9. A 005 .8v pm: :00: 00.0 80.5 020 95.3 min 5.08 00.2 zggwao
0 88.5058 0 63.9.0 .880 03030805 20.5. 20903. 200.55

 

.5380 2 2009

122

00.11.0090...

Amwfimvnr .2: H :0 .0 n 2.. ....m u :0 A319} .00“ u :0
.006 000000 n 00 .30—200300 900533 0:008 0.00 0020> .9.on 3.250 n ZOO
£00020 230 u DOM £00020 33930 n AM90. £00020 6000330030 u m>D co_>0n0m .Eoom 0o 0003050 1002085 .I. Dm<m .082

 

 

9.. zoo A 09.

00. zoo A mom

g; ......ZOU A WWQ

cm; ...DOM A m>D mum—500M 0%

o: .09.. A 005 3v 000.0 .800 00. .50 00.0 .30 00. A05 00. 0000280 08200

9.. zoo A 09..

00. zoo A mom

00. zoo A 005

cm. DOM A m>D wEfioFM 0%

00. 090 A 00E 00.0. :00 .000 00. 030.0 00. 005 00. A05 00. 00.2092 35200

00. .28 A 09..

3.. zoo A no.0 0:00:80 0.

00. 09.. v 005 00v 20.0 00.000 00.00 080.000 8.00 000.000 2.00 80.000 00.00 000.30 0020030

00. zoo A 09..

000 ...zoo A 006 05820

03 000 A 005 .0 0:00:00

00. 090 A 000 00v 000.0 .600 00. .50 03 .25 00. $5 00. €020.30

00. zoo A 09.

3.. zoo A oom 050500 .0

00. zoo A 005 00v 000.0 .300: 00.0. 060.000 00.00 .8008 :00 A000: 00.00 000.30 36300
0. 00800058 0 600000 .9008 020302000 20:. 2,0003. 2000;

 

.35—Boy MD 030D.

123

.00. v 0...... .00. v 0...

Ana—.9000 .02 n =0 .0 n =0 .3” n =0 A0038} dm u a.
.0010. 000000 u 00 .A0:o00_>00 0000:0000 0:02: 0:0 0020> .9.on 35:00 N ZOO

”000030 Ewt u DOM £00020 0200000 u MD< £00020 0000330030 u m>D uo_>0;0m Eoom 00 0:030:40. 30:000.:m u Dm<m .0002

 

 

00. zoo A 09..

00.. zoo A com

:1 ...zoo A 005

mm. DOM A m>D w:_o:000_D 0% 0.00

0.0. 09. A 005 00v 00.0.0 000.000 0.0.0.0 .6300 00.00 .8009 00.00 $0.000 00.00 05003 0.6200

00.. zoo A 09.

00.0 ......zoo A 005

2: *DOM A m>D w:000_w0Z 0%

00. 09.. A 005 00v 02.0 000.000 00.3 .3000 00.00 .0000 00.00 00.000 00.00 0:00:00 0.5200

mm: *ZOO A m>D w:=_o00M

cm; ...DOM A m>D 0% w:0000o:m

00. 09. A 005 00.v .000 .85 0.0. H.50 2.: .05 00. 2.0.0 00. 0.020003

00. zoo A 002

00. zoo A mom

mm; *ZOO A m>D w:000.00M
00. com A 005 .0 000.800.

00. 09.. A 005 00v 000.0 ”0.0.0 0.0. 03.0.0 00. .300 2.. 5.0 3.. 0.030003
0. 0085058 0 300000 .0005 80200205 2000 20003. 2000.5

 

$3303 2 030D.

124

.00. v 0...... .00. v 0...

@019”: .2: n =0 Amwfimyw .2: n =0 .0 n =0 .vm u :0 .Gwfimvnr .om n :0
.050 000:0 u h $028300 900533 0:00.: 000 0023/ .9.on 65.50 H ZOO
£00020 2w: u DOM £00020 033000 N MD< £00020 nous—$000.00 n m>D uo_>0:0m Boom Mo max—0.2 332.00% u mw<m .032

 

 

a. .3200 V mg

00.0 ......zoo v 80

No; "Inna V DOM

00.0 ......zoo v 005 050055

00.0 “.090 v 05 _00.v .0000 060.0: 00.00 20.000 0.10 000.000 00.00 00.0: 0000 0 0:33.030
54 **ZOU V ma

0_ .0 ......zoo v 80

00. 09. v oom

_0.0 .128 v 005

00. 06.0 v 005 0002000

00.0 002 v 05 ~00.v E .00 05.0: 00.00 80.000 :00 00.000 00.00 60.0: 00.00 0. 000082.000

00. zoo A 80

00.0 ...zoo A 005

00. 8.0 A 005 058305 0 to

00. 002 A 95 00v .00.0 060.000 030 “000.000 00.00 000.000 030 00.00 00.00 05003 0.52003

00 zooA0o<

00. zoo A oom

0E .28 A 03 0:09:02

00. 0:00 A 05 0. 0:05:0—

00. 002 A 006 .0v 00:. ”001000000. H000.5 03... 000.000 020 30.000 00.0.0 3620.3
0 285058 .0 301000 .838 03030203 2005>

 

.9583 MM 0308

125

..0. v 0...... .00. v 0...

Aha—.99. .2: u :0 .03—.09 .2: u go .c n 00.. .3 n to $3.9M: .3 u c.

.006 0000.00 N V .30—200380 ESESO 0500:. 0.00 003.3 .03—w 3.00:8 u ZOO

£00020 Emu u DOM £00020 03090—00 u nMD< ”000020 0000330003 u m>D uo_>0:0m 38% Mo £03050 Ragga u Dm<m .0002
Nw. **ZOU A mg
50.: ......ZOO A DOM

om. MD< A DOM
2:0. 1200 A m>D
mm. DOM A m>D wExobfiD
a: *MD< A m>D 50V iofim 033 000. Am: 5.: :3 NY: $3 co: 0% w::00.00M.:0m

woA 0.00200 A ma
3.: ......ZOO A DOM
ow. IMD< A DOM
th ......ZOO A m>D
mo. DOM A m>D $500030
we: **MD< A m>D SQV 00:90 02003 9.9 DNONV 8.2. Gmémv Ede Aomdmv 8.20 0% wcuomofifiom

mo. **ZOU V @9400
m5: 3.200 V GUM
Nb. "Inna V DOM
Nm.N 0:07:00 V m>D

 

_0. 060 v 005 0:650:00
.2 :50. v 05 .00v 000.00 080.30 00.00 00.000 00.00. $0.00 00.00 00.0: 00.00 0. 000.00.820.00
00 0085055 .0 6010000 .0508 030302000. 0.0.0 2,0003 20050,

 

50.0053 2 0300M.

126

..0. v 0...... .00. v 0...

203.000.. .2: n =0 Amfifivm. .R: n =0 .0 n :0 .000 n 20.200130: .00 n :0

.050 000.000 u 00 2005:0300 Ewocflfl 0500.: 0.0.0 002.00» .950» 30:50 N ZOO

£00020 Emu n DOM £00030 0309000 H nMD< "000020 Bum—sweat H m>D uo0>mc0m Boom Mo 0:02—0:00 30300.05 n Dm<m .0002
mm. *ZOO A MD<
om; **ZOU A vam

 

00. 0.002 A 000.

00.0 ......zoo A 005

00. .50.. A 0.5 0.00 00 0:08.002

.0.. :09. A 005 .00.v .0000 020.0: 00.0. 0.0: .000 5.000 00.0 0.0.0.0 00.00 00 00.03.0000
0 00800058 .0 60.00”. .0026 09200203 0.0.0. 2,0000... 0.00.5;

 

.8253 m: 03.09

127

.3. v 52.. .3. v 00*
.000 000000 N n 0:20.030: 0:00:05 0:00.: 000 002.5 .0:0=o.::o0:0>o u ZOOO ”00:02.00:
n mmM ”0023:5082: u ZOOO .00_0::o:00..0 2009320500: .02—0:00:00: 0o 3080000000 3:205:05 .I. OmMm<D .0002

 

00m. *mmM A ZOOO
we: ......ZOOO A ZOOD
bvfi 1.8M A ZOOO 52v $0.8 8WD OWNV €28 mmdm 25.3 :.\.N 000:0..o__0O
-- 0: 9.. mm. 8:.mC mmfiv $0.: 3 50.; 3:63 3.20 b_>_0_:quO

mm. *ZOOO A ZOOD
hm. MMM A ZOOD mo.v 9.0“ God: owov Amm.wv howm 3:63 0.02:0 wEv—00m 02::Em

VMA inmmm A ZOOO
mm: “0.02000 A ZOO:
VON ......mmmm A 2003 ~8.V mmdv 32$ wcfim Amhdv 0.0.30 30va mfimm 5:50:05 030000.30

mvd imMM A ZOOO
0b.: **ZOOO A ZOOO
co...“ ......mmDM A ZOOD 50V 0003 8:: 3 No.30 200.3 :w.mm Awwdv am._m 0.0.0395 25:02

0%.: ......mmM A ZOOO

 

cc: ......ZOOO A ZOOD :oEoED
omé tam—M A ZOOD SQV 8.3 28.8 mm? Swdv No.3“ 22.0 00.3 0>EcwoO
z: **mmMAZOUO

mv: ......mmM A ZOOO 59v mmdm 30.2 ND? 34.8 00:0 80.3 :60” 0005200080300
00 0:00t0QEoO & $2.8m ..0=o.::o0..00:D :0__o.::oo..0>O 0:0=_00M 033...;

 

0.0000005 «N0 H 5 0000000280035 .33 4%». H E 030003000005 5.0. n 3 0.00.00.00.02 000030 0.030300% Gmikvd 5 008.000th 500:

3 033.

128

4o. V RI. .mo. V a...

.37. Soto .I. n .AmcoaaFuv 23:83 288 8a mesa; $230685; n ZOOO ”35:58
u E "EB—05588:: u ZOUD ”2.8520850 Baum—-mwo—ofiwm bficoaom mo EoEmmomm< 3:06.585 n Omdm<fl .082

 

 

we. *mmm A ZOOO
24 ......2000 A 2003
Rd ......mm—m A ZOUD _oo.v 34m 8N9 .m._ AN: 2 ._ Gog 8; ESE—0m
5. .9500 A 2003
mm; ......mmm A 2003 _o.v wfim 8mg 8.0m Swwv 2.90 Avawv 8.90 8&3ng
GOA ..cwmmmm A ZOOO
hm. ......ZOUO A 2003
CM ......mmm A 2003 50V Emom Ammd: 38¢ 83 C ocdm A29 340m oo:m_0_o>< Eoom
2 .— gimm—M A ZOOO
mm; ......ZOUO A ZOUD
cod ......mmm A ZOUD SQV omdm A306: oodm $5.8 $8.0. Amowv cvdm mmocmaofiamsm
G. *mmm A ZOOO
cm. .2000 A ZOUD
we; ......mmd A 2003 Sov 3.3 3.68 $.mm Cod amdm 200 No.3 mEoBoE 83:00
and 3.33 A 2000
ow. .2000 A 2003
mod ......mmm A 2003 50V Bump A92 C 03% $5.8 wosv 22$ aflwm fio§=<
E. 1.8% A 2000
0». mm:— A zoo: 5v 3.0 5.6 8.2 56 8.2 3.8 8.8 8.285 >85...
0 acomtamfioo m Ana—.NE 00:95:0830—5 02.055890 2.058% 032.5,

 

.3553 E 033.

129

.8. v 0...... .8. v 0...

find”: .w u a... .2 u 2.. .8 u 5.031s": .9. u c.
.86 .8000 u 0 02.838 235% 28:. pa 82.3
M5.0.2.853 u 280 55:52 n ma ”20:389.? u 28: Baum..=om-o_8m 5:533 38m u «3% .082

 

 

mm; ......mmm A ZOOO
no. ZOOO A 2003
mm; imam A 2003 Sov moém 8mg mod $3 cw; $3 mm; Boom Eon.
mm. mmm A ZOOO
No. ZOOO A 2003 86M
S. E A 2003 S. REN 089v 3.. .0399 30. “hog av. 8505:3th
Hm. ......Mm—M A ZOOO
cm. 280 A 200: 02:30 822$
$4 .1.me A 2003 59v 2 .2 A23 cw. Goo m0. Gog mm. 5:5 maimcocflom
cw. “Inmmm A 2000
mm. 2000 A 2003 moE>uQ<
on; ......mmm A 2003 Sov an: 809 EN 809 mod $3 mm; 0:53 28 Boom
mm; ......me A 2000
Cm. 2000 A 2002
EA ......mmd A 2002 SQV 0.3.3 89V 30. «Gog $0. 33 mm. x83 EuvBm
u mcgcaanU m Ame—.mvm 00:95:09.0ch 026005890 28.30% 03.0.53

 

0000.35 «N ~ u S 0.30002800ng V50 .«MV u .0» 00300300005 .«Mm. .I. 5 0.00.5002 0.3030 0.030300% $0-3 5 0002000§5 500:

m ~ 033.

130

.5. v 3:." .3. v Q...

$2.33 .2 u 000 .mm u an .9. u 00“

.onmm Soto u h $50333 93:83 2:38 2a 825/

”EB—obcoego n 2000 ”85:68 n mmm NEE—05:00.85. u 2003 3.58265 wcficom .3505— " RE .082

 

 

we; ......ZOUO A 2003 55088930
hm; ......mmm A 2003 _o.v uNNH 020D 3.3 30.8 3.2 ..C_.8 3.3 3585
No. *mma v ZOOO
aw. ......ZOUO v 200:
mm; 1.8% v 2003 ECV 0.3.2 UAwde 3.2 Gmdv ~v.v~ ..Gwdv cvdm Emu EEBE
hm. *mmm A 2000
mv. 2000 A 2003 5588930
84 Immm A 2003 3V N50 ANNE mmwz «308 ONE Sfimv cod 35222
cm. mmm v ZOOO
mm. ZOOO v 2003
ow. *mmm v 2003 3V and 30.3 8.3 «Gog cvdm Smwv mmdm 0.30 35032
n mnemtmeoU & AX: .mvm 02—00209—0623 02.880890 2.236% 033.80,

 

0.08.35 «N0 u 5 03~©00280§b 33 .2.» u 5 033038086 .«Mm. u 5 300.53% 93030 0.3833,». ~$~ 5 0.3.0000wa Sun:

2 2an

131

”82658058 u zooo ”3:232 n ma neo__§=oe8§ u zoo: .8320: 38m .6 2322 3.38% u 3% ....Bz

.8. v 5.... .8. v a...

$2.va $2.33. .3 n :0 .mm H 2,. .mV u an
.86 Soto H V Amp—2830: Ego—35 £52: 05 woo—Er

 

 

mm. mm:— A zooo
E. zooo A zoo: meg? a.
m: ......mmm A zoo: 8v 3w 83 Q. 83 «m. $3 2. @2232 :52“:
S. mom A zooo magnum a.
2.. Em A zoo: 8. 2A 633 8.2 83$ 8.; 838 a: $2.3 €023:
ow. ...mmm A zooo @555 a.
ow. Em A zoo: 3v 6:. G3 cm. 33 8. AR: mm. mega: 3523:
wimp—sow a.
-- ...: on. mm. 3:: 5.: 5.5 was 6qu 8.2 $35 36an

a. mom A zooo
:. zooo A zoo: @833
3: 18¢ A zoo: 8v wow 83 2.: $8 8. :3 S. a. 39:»: secs:

2.. ma v zooo
9.. zooo v zoo: 85552 a.
3. ...mmm v zoo: 8v 3;. 82.: 8.8 and: 3.? $3: 3.; 585m 35%:
V wCOw 59500 R AVA: .an— uo=ObEOBOUCD uo=eucoouo>0 “com—EDM— gnaw—«xi

 

€8.35 «NN u 5 xozoufioosofinb 33 .«NV H 5 g3~o§=oo§>© .«Mm. u 5 3w§mo¥ 823‘ $835 93.6 S nogosofiQ so»:

2 2an

132

.8. v 3.... .8. v a...

Amoufivnr Ana—fir? .: u no .3 u 2.. .mm H 2.
.37. Soto .I. V .AmcoasSov ESE—3v «:38 03 82.3
”flu—355805 u ZOOO "35:68 n ma "EB—ohcoooooco u 200: 6222.3— Eoom go mum—SE goo—5m n 9310: $32

 

 

mm. *8”: A zooo wasfiwoz a.
Q. Em A zoo: 3v 5m 93: mm: awed: 3.2 $3: an $53 352?:
8. mm”: A zooo

mm. zooo A zoo: @6522: a.
a. ...mmm A zoo: 8v 2:. $30 2 am 833 :2 2&5 a .2 to 3:95 3.532
mm. Em A zooo

an. ...zooo A zoo: wéfiwoz a.
8.: ......mmm A zoo: 5v :3 85 8. a8 2. £3 M2. mafia: secs:
2. Em A zooo

G. zooo A zoo: wéeoom a
o: ......mmm A zoo: 5v 3m 83 3. 85 mm. 83 a. @328: 35:33:
9.. ma A zooo

cm. zooo A zoo: 3:85 a.
8.: ......mmm A zoo: 5v vow a3 8. :3 mm. 83 3. €283. 3.023:
on. zooo A zoo: @583: a
2.. ...ma A zoo: 8. mod :5 E. 83 N... $3 R. $9.395 fig:
V mCOmSMHEOU Q AX: .anm uO-Eb—SBQUCD uo=gcoeo>o “com—$0M 032:5,

 

.8288 S 033.

133

Ac. V RI” .mo. V Q...

Ame—.39 Ame—.3"? .: u 5 .mv n 2: .Nm u a.
.37. Soto n V Amp—2633 35$ 9.3:. 0.3 333/
woo—353803 u ZOOO ”85.sz H mm:— HEo=ob=ooouU== n 2003 uo_>~zom Boom mo max—«5‘ 123035 N mm<m .SoZ

 

 

R. mom A zooo @558

S. mom A zoo: 8. com: homo on: 9893 3.2.. «:33 2.x a $25 352on

g. 3.8:. A zooo

E. zooo A zoo: $55 a.

«2 2.3:. A zoo: Sov 3.2 03.3 N2 963 2. .83 a. 33:»: 352mg:
maisom do

-- ..e K. 6%. 039:: $2 54%.sz 3: .823 $1.. $23 352$

mm. ...mmm A zooo

mt zooo A zoo: magi

8.: .....mmm A zoo: 5v 1.3 033.9 5.2 $8.8: 8.2 “ammo .2 s a mango: 362...:

mm. 9.2 v zooo

8. ...zooo v zoo: 855:2 a

8: IE v zoo: av .26 .G_ .8: 3.3 93.8 3% games $3 338: 353$

mm. ...mmm A zooo

G. zooo A zoo: @6525 a to

3.. tom”. A zoo: 59v 5.... saws moan 533 ES and: 3.: mafia ztsswao

u 38.52550 «N Ava—.Nvm 5.3.5.5802: oo=ob=oooo>0 E058“ 033:9

 

53:53 2 £an

134

AC. V 3:." .mo. V R...

Amofimvnr Amofimvmu .2 n :0 .mV n 2.. .Nm u 2a
.36 Soto u n .AmcosESo 235sz 33.: En 325/
”woo—.8285; u ZOOO ”85:82 n ma "EB—obcooooocs n 2003 uo_>a:om Boom mo mam—SE 3.58qu n mm<m $82

 

 

3. .....mmm A zooo
3. zooo A zoo: 3.85.0.5 a
3: :8». A zoo: Sov ”Rd. gains on; ..EGQ 8% .233 3.2 to 9.555 1.6%“:
R. .8”. A zooo
E. zooo A zoo: wéfiwoz a.
9: ......mmm A zoo: Sov KS .6: do 3% 933.8 8.2 3.3: O? mafia: BEBE
R. mm:— A zooo
8. ...zooo A zoo: wazooom a.
9: :8: A zoo: 8v an: .83 8: $3 8. .23 cm. @528: 3323:
3. ma A zooo
%. zooo A zoo: 3:85 a
2 ._ ......mmm A zoo: 8v 6an .85 we. .83 mm. .83 3. $283. 252?:
3. Em A zooo
G. zooo A zoo: wceouom a.
S ._ 1.32 A zoo: 5v aw: 0:5 mm. 52.3 S. .93: 2. 3522A: 3528
8. ma A zooo
R. ...zooo A zoo: mass?
8.. 3mm:— A zoo: _8.v am? .65 3. ago mm. .83 2. a 3:83. was“;
w macaw—«9:00 Q AVA: .anm uo=ob=080fic3 u0=Ob=OP~D>O “com—«wov— OENE>

 

.Cobzoov : 03.;

135

.3. v 3.... .3. v Q...
$2.va $222.8”? .2 u ..u .9. u .... .mm H ....
dim Soto H .o .Amco..m.>oo 9.3533 2.8:. o... moo—Sr

”EB—8.5805 H ZOOO $2.026»... u mud ”EB—obcoofiocs n 2003 uo_>~:um Boom mo max—«.5. EBoEm u mm<m $82

 

 

—N._ **mm—M A ZOOO
o... zooo A zoo: 352.8
3.. ......ma A zoo: .8v .38 605 3.3 93.03 mm: Sue: om... a. fission
Nb. tummy“ V ZOOO
.3. zooo v zoo: masses.
8.. ......mmm v zoo: 83 .....o 5...: 2% gamma 3:. 83.. .3... a. w=.=mu=oz...om
wh. ......mmm V ZOOO
.m. zooo v zoo: 3555
8.. 3%... v zoo: .2? .Rd 82$ 3% .85: 2.2 $3: $3 a. m=.>3...om
8. .45. v zooo
N... ...zooo v zoo: 98.9.:
mm. ......mmm v zoo: .2? .8... 833 SS .313 o. E 63.: m3... ... $383....om
v.0. ......mmmd A ZOOO
E. zooo A zoo: $655.: a. to
t... ......mmm A zoo: 83 .2... ..mw...s 3.... £3.23 3% ..mm.-. mm... 3...“? rain“:
mm. mm... A zooo
Nb. zooo A zoo: $38.32 a
3.. in... A zoo: ....v axon ..3..5 3.3. 95...: man. .52. mm... macaw. unsfiao
h 385.350 R Ave-dz 8.3588023 ...—3:50.90 25:63— 03a5>

 

APEoov N.— 2an

136

137

.8. v at. .8. v a...

$2.3m. $2.8m. .: u a. .Q n 2.. .Nm u 2.

.oflm Soto n w .Amcoumgoo 93:33 282: 0.8 mos—Er

”moo—35.5803 u ZOOO ”85:68 n mmm WEE—obcooboos u ZOUD ”5320mm Boom mo mam—£2 wagosbm u mm<m $82
8: ......ma A zooo

 

ac. ..zooo A zoo: :3 oo 382sz
KN ......mmm A 2003 Sov ”flaw 48de Sdm ndeC wwfim Che: 3w ow wagepfiwn
Nw. ......mmd A ZOOO

co. ......ZOUO A ZOOO wcfigmon
Ned “swam—M A 2003 39V 0343 Awwg MN; p3©¢ 3. five mm. Q w::8.~od..:ow
3 385950 a Ave—.NE 5:93—58ch 8.355890 22:93 032...;

 

.9203 S 2%:

APPENDIX B

Figures

138

uoEEQm Eoom mo nah—«Ex RES—Em u Mm<m .853 2:
go comment 53, 32.:va .533 «o 3653.5 05 .3 32 EwFEoU A d .nmgmgghmozu Eok M23 V5» .EEVoE 93$.
53:3: fa: 3%: 3% .88 dais: .m J E ...< 88% 2E-_§52 {8: 5.55 92a .8 853:: 88m ....32

QZ—u—x—OmPZOU Q 9259—23 .mé

          

QZ—sm—POMm Q UZEDPKDZ .9; 62:24}:— Q 02—AEAm—m .04

8.52:5 .21—St ... .>_ 3.5:: 2.75: ._=

GEM—BM

OZ—flU<O~—AE< Q 675,0»— .mé Q GZEU<E< RA
33:33 >3:th augSPEW .— >533: 9:32. 8.9:: .=

GZ—GZ<Pm¢m—QZD Q GZE—mh< .NA GZ—hUm—x—UHZ Q 025—02019—

 

UZ—EHUKOh Q alum—E Tu ZED ZC mach

 

.288. 5% .333 a .333 8:585 H.88,“ mm<m 2.. we _ 09c...» 4 2&2

139

geafiom Boom o6 mam—SE RESP—um n mm<m .853 05 we soumiom £5 Bianca .55
mo bioicD 05 .3 83 EwFEoU A d awkwagfiamam Eek mrs BS» €53: 98$. S§§§E “.135
«was 54% .88 ...Eifim .m A E ...< 888 25-352 f8: 5:5 3% 5 magma: 58m s32

.05:ng Q clam—EH9 .mun

         

 

0254—: Q GZ—hmDmh .Q.N UZFfl—MDUw Q UZ—MADw 6-N
=58: .23—3.5 ...: .2..an ~38: .=_
“HZ—doom:—
UZEDm—ZZOO >AADh>OH .m-n Q UZ—hmmhomm .h.N
biz—3:: .m=.:o_.¢ :7:— ._ €85.35 «:35: 31.9. .=
DZ—UZ<.—.w—D
Gamay—mug Q 02501—099 .N-N Q Paglia—<3 Q."
fax—Hm ZO wDUOh

02—8355 Q Uémmwm< .~.N

 

.288. 5% .330 a .535: 3.558 ”is: 55 o... a. N 855.9. .N ~35

.SEEom 3690; we Ema—SE Bacabm n mm<m .853 05
no coamqum :23 32.53% .55 m0 365th 2: .3 83 EmFEoU A d .nwxmgotum3Q ESE used .33 .553»: ESE
.8535“: “.135 Hwhé aim. doom .=§.Eom .m i— .3 ...< sonoom o_E-_§:aE £95 55.: mm<m Sm men—mi: 82n— 632

02259:.me Q UZEOF—ZOE ran—mm m4” .92—manic

Q UZ—FU—QZ—fidflm .3

          

UZ—UZEH Q UZBm—MDOZQJMm .v-m

 

Ozgnxmhwmn Q
:3 295:3 Jug—:2 .>_ :3 maul—ago ._= UZFUH—Hu—fidnam
.hé
Glam—mama Q UéOAéAHm .m.m
:3. zine .Eooo< .— :3 some: ._ _
Olav—Sanka Q UZ—Em—UU< Haw—mum .Num RAM—m ho Uz—hbm—w—Um—Z
Q QZ—E<H¢§<G .anm
AH D HZ<FZOhm .Tm
h m m C wDUOh GEM—HOE—

 

.282: 3% .333 a “Ego 8:558 H.982 5% 05 is m anm .m as»:

141

Figure 4. Three-factor model of personality

W???”

idpfin anxfin cogfin soafin afifin susfin subfin shafinlg comfin

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_88 .83 -q9 73 .70 .77 .57 .66

.33
Q

  
  

  
 

Intimacy

Ps cho ath
Problems y p y

.43 fl35 -58 .49 .79

 

 

 

 

inpfin narfin confin ssefin calfin

Note. Idpfin = identity problems; anxfm = anxiety; cogfin = cognitive distortion; soafin = social
avoidance; afifin = affective instability; susfin = suspiciousness; subfin: submissiveness; shafinlg
= self-harm; comfin = compulsivity; inpfin = intimacy problems; narfin = narcissism; confin =
conduct problems; ssefin = stimulus seeking; calfin = callousness.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

142

Figure 5. Dendrogram of Adaptive (n = 40), Rigid (n = 35), and Dysregulated (n = 7)
Clusters.

 

 

 

Adaptive

 

 

 

 

Dysregulated

 

 

Wffi“%%fl%fl

 

 

LE]?

 

Rigid

 

 

 

l

 

llllllllllll:|l|:lll|:lllll_J

 

143

“mu-119s

Insisspm N

compimv

macs

 

98553850 33932055” baa—Sauna we acofimmomm< RaoEEEa n Omdm<Q .202

 

 

 

EfibcoouovaD ...... 8:09.885 I I I “=3:qu J
saga 353
s S
m” I m I do .
m mm m Wm m Mmmmmmmwm m.
s s s
u flm. . q u M. Wm. Mn. NW. an m
mm» m mm m mummumammm
o
2
o
8 w
fl
8 M
a. m
m
cm
8

80620 fig u 5 B:o.a:o§=D
98 A? u 5 B=ob=8uo>o Amm n 5 “com—Bum 383‘ 8:335 Gag—.55 he 2.32 N ens—ME

145

   

IIIIjIgIIIIIijIIIjijil