This item is only available as the following downloads:
EATING DISORDERS AND PERSONAL CONSTRUCTS:
THE EFFECTS OF ANTICIPATED WEIGHT GAIN ON WOMEN'S
PERSONAL, INTERPERSONAL, AND VOCATIONAL CONSTRUCT DOMAINS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
This dissertation is dedicated to my parents,
John and Karen,
whose lives are devoted to inspiring and supporting others,
especially my brother and me;
and to my husband, Stephan,
for providing his endless strength, humor, and encouragement
throughout our years together.
There are many people whose personal and professional talents assisted me in the
many steps necessary for completing this academic endeavor.
Foremost, Greg Neimeyer, my doctoral chairperson as well as mentor, was pivotal
not only to the successful completion of this manuscript but to my continued growth as a
future psychologist. I was privileged to learn from Greg by observing him in many roles
including instructor, researcher, and clinical supervisor. He is a truly gifted educator, and
I am grateful to have been his student throughout my graduate schooling.
I am also thankful to Lisa Brown for agreeing to serve on my doctoral committee
She is a remarkable woman, and her keen critical thinking skills and sharp wit validated
my work over the many years. Lisa fueled my enthusiasm and dedication to the field of
psychology by always demonstrating her own commitment and energy to learning and
enjoying teaching and "testing" hypotheses.
I would also like to thank Dorothy Nevill for her positive, energetic, and consistent
support throughout my graduate training. I respect her many contributions to psychology
and to my development as a psychologist-in-training. She was always accessible,
receptive, and responsive to my professional needs, questions and curiosities. Dr. Nevill is
an outstanding role model for many.
Jaquie Resnick's assistance to my professional growth has been invaluable. I am
appreciative that Jaquie graciously agreed to allow me to serve as her co-therapist for a
disordered-eating client during my first practicum placement. Thereafter, we developed a
rapport that greatly contributed to my interest in women's issues and to the design of this
study Jaquie provided a sense of realism to the goals of this research.
I am grateful that Connie Shehan agreed to serve on my doctoral committee. She
is a selfless, creative, and a faithful advocate for teaching and learning. We worked
together on an independent research study, and I again benefitted immensely from her
enthusiasm, knowledge, and appreciation for women's issues. My graduate work was
deeply enriched because of the strengths that Connie possesses and shares with others.
In all, these doctoral committee members exhibit a passion for teaching and
inspiring others and for this I am forever grateful.
I would also like to express my gratitude to Kurt Boniecki for providing me with
statistical assistance and consultation during my prescreening and data analysis.
Lastly, I thank my husband, Stephan Eisenschenk. Words can not express the
strength, support, patience, and encouragement he has provided me throughout my
graduate school years. To him, I am most grateful
TABLE OF CONTENTS
Understanding Eating Disorders ....
Current Study ....
2 REVIEW OF THE LITERATURE .............
Historical Origins of Eating Disorders .. ....... .
Sociocultural Factors Associated with Eating Disorders ....
Cognitive Features Associated with Eating Disorders .......
Personal Construct Literature as Applied to Eating Disorders .
Purpose and Hypotheses of the Study .........
Instruments and Procedure
5 DISCUSSION ..
Limitations of the Study
Future Directions ..
. . . iii
. . 1 1
. 3 0
.. .. 34
. 3 8
S. ... 48
. .. .. 4 8
. . 5 0
A SELECTED EATING DISORDERS INVENTORY (EDI) SUBSCALES
B INFORMED CONSENT ........... .... ......
C REPERTORY GRID--PERSONAL CONSTRUCTS ...... ... .
D REPERTORY GRID--INTERPERSONAL CONSTRUCTS ..........
E REPERTORY GRID--VOCATIONAL CONSTRUCTS .............
F MODIFIED CONSTRUCT IMPLICATIONS GRID ............. .
G CONSTRUCT IMPORTANCE FOR EVALUATING WOMEN .......
H CONSTRUCT IMPORTANCE FOR EVALUATING MEN .....
R E FE R E N C E S ....... .....................................
BIOGRAPHICAL SKETCH ...... ....
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
EATING DISORDERS AND PERSONAL CONSTRUCTS:
THE EFFECTS OF ANTICIPATED WEIGHT GAIN ON WOMEN'S
PERSONAL, INTERPERSONAL, AND VOCATIONAL CONSTRUCT DOMAINS
Chairman: Dr. Greg Neimeyer
Major Department: Psychology
This study sought to explore the relationship between disordered-eating and
structural aspects of personal construct systems. The study was designed to test whether
persons exhibiting disordered-eating would exhibit greater levels of construct integration
but lower levels of differentiation, compared to persons not exhibiting disordered-eating.
Additionally, it was hypothesized that the former group (High Eating-Disordered) would
anticipate significantly greater degrees of change in relation to their personal,
interpersonal, and vocational functioning should they suddenly gain 20% of their current
weight compared to the latter group (Low Eating-Disordered). Participants (n=40) were
selected from Introductory psychology classes and were assigned to either the High
Eating-Disordered group (n=20) or Low Eating-Disordered group (n=20) depending on
their prescreening scores on selected subscales of the Eating Disorders Inventory (i.e.,
Drivefor Thinness subscale, Bulimia subscale, and Body Dissatisfaction subscale).
Participants completed a series of six pen and paper measures that included three repertory
grids that varied the content of the constructs (i e., a personal grid form, an interpersonal
grid form, and a vocational grid form), a modified implications grid, and two forms that
assessed the construct importance that the participants used in their evaluations of other
persons (i.e., one form assessed women, another form assessed men) across personal,
interpersonal, and vocational domains. Results failed to replicate past research findings
that demonstrated disordered-eating persons as having more integrated yet less
differentiated construct systems. However, as predicted, the High Eating-Disordered
group anticipated greater degrees of change to occur across personal, interpersonal, and
vocational domains of experience should they suddenly gain weight compared to the Low
Eating-Disordered group. There was no main effect between the two groups in relation to
construct importance when evaluating other persons across personal, interpersonal, and
vocational domains. However, there was a main effect for construct domain with the
interpersonal domain being the most important in the participants' evaluations of others
(both men and women) These results were discussed in relation to previous literature
findings and in a sociocultural context Limitations to the study as well as
recommendations for future research directions are also provided.
The term "Eating-Disordered" is no longer unfamiliar. Eating disorders are
currently recognized by several fields of study, including psychology and psychiatry, as
potentially life threatening sets of behavior patterns. They are believed to affect over 8
million women in the United States (Wolf, 1991). More specifically, it is estimated that
approximately 2% of adult females in our country are suffering from either anorexia
nervosa or bulimia nervosa (Fairburn & Beglin, 1990), while adolescents are reporting
even greater prevalence rates for these two eating disorders (Lucas, Beard, O'Fallon, &
Kurland, 1991; Powers, 1996)'. Perhaps even more alarming is that disordered-eating
exists on a continuum (Drewnowski, Yee, Kurth, & Krahn, 1994; Neimeyer & Khouzam,
1985; and for a review of related literature, see Levine, Smolak, & Striegel-Moore, 1996),
suggesting a much higher prevalence rate still when "subclinical" versions of eating
disorders are taken into consideration. In fact, there is a developing opinion that all
women are susceptible to restrained eating (Orbach, 1986). Consequently, there are
significant numbers of individuals, including clinicians, researchers, affected families, and
victims, who are invested in identifying the factors involved in the etiology as well as the
maintenance of eating disorders.
'Fombonne (1996) reviews the epidemiological data that suggest changes in
diagnostic and referral practices account for much of the increase in the number of bulimia
nervosa diagnoses; and see Heatherton, Nichols, Mahamedi, Faru, & Keel (1995) for
additional explanations of varying prevalence rates.
The current project was designed to address these factors and to identify the
generalizability of distinctive cognitive features associated with disordered eating.
Specifically, the study tests the extent to which women are implicativelyy bound" by the
restricted nature of their food-related constructions, and the extent to which these
constructions are reflected across a wide variety of personal, interpersonal, and vocational
domains of experience.
Understanding Eating Disorders
Eating Disorders Defined
The fourth edition of the Diagnostic and Statistical Manual of Mental Disorders
(DSM-IV; American Psychiatric Association, 1994), defines disordered-eating as
abnormal behaviors such as restricting one's food intake, engaging in purging behaviors
(e.g., self-induced vomiting, or misuse of laxatives or diuretics), as well as binge-eating
("eating within a discrete period of time an amount of food that is definitely larger than
most people would eat during a similar period of time and under similar circumstances";
also, sensing a "lack of control over eating during the episode," p. 549). More
specifically, the current DSM classifies these eating disorders into three distinct categories:
Anorexia Nervosa, Bulimia Nervosa, and Eating Disorder Not Otherwise Specified. Since
the 1970s, researchers, clinicians, the media, and society have increased their attention to
all eating disorders, yet have particularly focused on the former two, anorexia nervosa and
Anorexia nervosa and bulimia nervosa share some defining criteria as described by
the DSM but are clearly two distinct eating disorders. Specifically, both disorders involve
the undue influence of "body weight or shape on self-evaluation" (APA, 1994, p. 545) as
well as a desire to prevent weight gain. However, while persons diagnosed with anorexia
nervosa have a body weight that is 15% less than that expected for age and height,
persons diagnosed with bulimia nervosa typically appear "normal" (i.e., not emaciated) in
weight for their age and height Individuals with the latter diagnosis also engage in out-of-
control binge-eating behaviors. While these are not the only distinguishing features of
these two disorders, they serve to illustrate the broadest differences between their defining
Societal Causes of Eating Disorders
Researchers and clinicians alike are sensitive to the role that Western society's
ideals have in the development and maintenance of eating disorders. They view harsh,
gender-rigid messages and proscriptives that define the ideal woman as thin as important
contributing factors to the prevalence of eating disorders. They also observe that some
women "hear and accept" these restrictive sociocultural messages more so than other
women, but, argue that all women are susceptible to the dangers of the message
(Anderson & DiDomenico, 1992; Brumberg, 1988; Bushnell, 1995; Solomon, 1996;
Garner, Garfinkel, Schwartz, & Thompson, 1980; Keas & Beer, 1992; Orbach, 1978,
1982; Raphael & Lacey, 1992; Sanford & Donovan, 1984; Spillman & Everington, 1989;
Stice, Schupak-Neiberg, Shaw, & Stein, 1994; Waller et al., 1994; White, 1992; Wolf,
1991; Worrell & Remer, 1992).
Biological. Familial, and Cognitive Causes of Eatine Disorders
Much progress has been made in regards to clarifying the defining criteria of
anorexia nervosa and bulimia nervosa. As a result, one might conclude that the origins of
these criteria are also known and strongly validated. Unfortunately, this is not true
(Brumberg, 1988; Wilson & Pike, 1993). Neither the psychological nor psychiatric
(medical) communities have conclusively determined why some persons are at a greater
risk of developing an eating disorder (or characteristics of an eating disorder) than other
persons. The only consensus existing among mental health professionals is that eating
disorders and eating disorder symptomatology occur largely in women and that eating
disorders are likely best understood and explained by a multidimensional perspective
(Button, 1993; Brownell & Fairburn, 1995; Garfinkel & Garner, 1982)
Appreciating the need for a multifaceted perspective regarding the etiology and
maintenance of eating disorders, researchers and clinicians contribute to the eating
disorder literature by designing research and clinical treatments that reflect their particular
field's hypotheses. For example, biomedical researchers interested in disordered-eating
might explore its relationship with nutritional and anatomical deficits (i.e., serotonin
deficits), as well as hormonal imbalances, dysfunctions of the hypothalamus, and excess
cortisol productions. They might also examine the relationships between the genetic and
biological factors of eating disorders (Strober, 1995).
Family systems-oriented psychologists, however, view the family and problematic
parent-child relationships (i e., mother-daughter) as responsible for the daughter's eating
problems (Bruch, 1973, Minuchin, Rosman, & Baker. 1978). Specifically, therapists
espousing this theoretical orientation address the familial themes of control, enmeshment,
perfectionism, and overprotection, as well as the family's conflict resolution styles. More
recently, interest in the disordered-eating person's siblings and choice of significant others
has arisen yet this research direction is still in its inaugural stages (Vandereycken, 1995).
Other investigators of eating disorders primarily view the differing cognitive styles
(i.e., rigid thinking, overgeneralization, selective abstraction, all-or-none thinking) existing
between women with disordered-eating behaviors and women with healthy eating
behaviors as the critical variable that distinguishes between more susceptible and less
susceptible women (Fairbur, 1981; Wilson & Pike, 1993). Treatment from this
perspective attempts to modify the person's cognitive style (i.e., challenge the cognitive
distortions; clarify and strengthen the person's identity) that positively affects their self-
esteem and (often comorbid) depressive state (Weinreich, Doherty, & Harris, 1985). In
all, the processes of research and treatment tend to mirror how a researcher or clinician
conceptualizes the contributing and maintaining factors of eating disorders.
Personal Construct Theory
One expression of the cognitive approach to eating disorders can be found in
Kelly's (1955) personal construct theory The most fundamental of Kelly's assumptions
was that humans are continually processing the world around them so as to generate and
refine predictions about future events, feelings, and behaviors. Individuals vary in both the
number of dimensions they use for making an evaluation or prediction (i.e., "differentiation"),
as well as the degree to which there is organization or correlation among the dimensions
(i.e., "integration"). Thus, the greater number of dimensions available for making
predictions and evaluations, the more differentiated is the person, and the more tightly
organized those dimensions are, the more integrated is the person.
The application of these structual dimensions can be extended to diverse areas of
interest, including the field of eating disorders (Button, 1993). Specifically, a person with
anorexia nervosa or bulimia nervosa would likely exhibit low levels of differentiation yet
high levels of integration In other words, this low-differentiated/high-integrated eating-
disordered individual has few dimensions in which to construe the world, but these
dimensions are tightly organized Thus, this person's highly integrated cognitive system
greatly assists him or her in making decisions and predictions about the world.
When faced with the more complex and variable construing to be found in the
wider and adult world such limited construing as they have available is likely to
be inadequate. It is perhaps not surprising that they constrict their world in search
of greater predictability and control. (Button, 1993, p.100)
The most common method of assessment used in the application of personal
construct theory is the Role Construct Repertory Test (Neimeyer, 1987). As part of the
administration of the "reptest," an individual is asked to consider the wide variety of
people in his or her life. For example, a person might be asked to write down the names
of his or her mother, father, spouse, least liked person, most liked person, self at heaviest
weight, self at lightest weight, self at ideal weight, etc These names constitute the
repertory grids' elements.
In addition to a set of personal elements, a set of "constructs" is also necessary for
completing a repertory grid Constructs seek to capture how a person construes a
particular experience and they are presented in contrasting pairs to the person completing
the repertory grid. For example, if we seek to explore how a restrictive eater construes
interpersonal dimensions we might provide him or her with the following set of constructs:
happy/sad, anxious/relaxed, spontaneous/deliberate, angry/calm, outgoing/introverted.
The person would then be asked to rate each of his or her elements along each of the
provided construct dimensions on a 5-point Likert-type scale.
Once the person has rated all elements across all pairs of constructs, then the
reptest is completed and it consists of a matrix of numbers. This matrix of numbers is then
analyzed for levels of cognitive differentiation and integration (Landfield, 1977) which can
be interpreted according to personal construct theory.
The application of personal construct theory to eating disorders was pioneered by
Crisp and Fransella (1972). In their case-study approach, Crisp and Fransella examined
the cognitive changes associated with the recovery process of two women suffering from
anorexia nervosa. They did this by having the women complete a series of repertory grids
on seven testing occasions. The researchers observed that significant changes occurred in
the relatively rigid cognitive system of one of the women ("Miss M. H.", p. 395), as
indicated by the change in correlations between two of her personal constructs.
Specifically, at the beginning of treatment, when the woman was very emaciated (i.e., had
a very low body weight), the constructs "mature" and "likely to become pregnant" were
perceived as very related. Thus, Miss M. H. associated these two constructs as basically
interchangeable This strong association denotes the woman's cognitive system as being
very integrated and therefore, very resistant to change. And, when a disordered-eating
person's construct system is resistant to change, a poorer clinical prognosis is expected
because the person is not receptive to alternative (i.e., more flexible or loose) ways of
construing her world (Button, 1983)
Later, however, as Miss M. H.'s recovery progressed (i.e., weight increased), the
relationship between these constructs diminished, which is a positive treatment indication.
In other words, her construct system had become permeable to other possibilities of
construing her world Miss M H.'s cognitive system was no longer poorly differentiated
and highly integrated and her style of construing had become more flexible.
Crisp and Fransella's other patient, "Miss A.W." (p. 395) also exhibited some
unique cognitive features after having been administered a series of 7 reptests at different
testing times. They found that until her treatment was complete, Miss A.W.'s
preoccupation with weight greatly influenced her self-view. Once her weight became less
meaningful to her, Miss A W. experienced significant disorder in other areas of her life,
including interpersonal relationships. She no longer exhibited a highly integrated cognitive
system which resulted in her needing to adapt to a less predictable albeit a more healthy
physical and mental state of being. As Crisp and Fransella conclude,
The use of repertory grids in the study of these patients has allowed exploration
and quantification of some aspects for these patients' construct systems,
including the ways in which these were organized and ways in which they changed
during the course of treatment. (p.405)
A researcher and clinician who has replicated and extended upon the pioneering
work of Crisp and Fransella (1972) is Eric Button In his Eating Disorders: Personal
Construct Therapy and Change (1993), Button states that "a common theme that runs
through the lives of people afflicted with an eating disorder .. is that they have one
interpretation of the world ... this focuses on the importance of weight" (p. 32). Thus,
Button, fine-tuning the research method initiated by Crisp and Fransella (1972), is one of
many researchers whose work has found that individuals with eating disorders tend to
have highly integrated but poorly differentiated cognitive systems (Button, 1983; Coish,
1990; Crisp & Fransella, 1972, Heesacker & Neimeyer, 1990; Mottram, 1985; Neimeyer
& Khouzam, 1985). Again, the implication for an individual exhibiting this type of
restrictive cognitive system is that the person may be resistant to treatment insofar as
treatment may require radical reorganization and elaboration of the individual's current set
The present study sought to extend this line of research concerning the relationship
between disordered-eating and structural aspects of the personal construct system.
Specifically, it explored the universality of the cognitive/structural differences associated
with eating disorders across different domains of experience In particular, this study
addressed the generalizability of these structural features across personal, interpersonal,
and vocational domains To date. research has largely concentrated on the structural
features concerning weight-related constructs (i.e, as reflected by the frequent use of
eating and appearance constructs). However, obtaining insight regarding the extent to
which disordered-eating may interfere with other areas of functioning (e.g., interpersonal
and vocational) may be warranted as we seek a more thorough understanding and respect
for the repercussions of eating disorders:
I think there is a good case for broadening the context of enquiry (sic) Up to
now, personal construct exploration of people with eating disorders has mainly
centred (sic) on the construing of people or eating situations ... I think there is a
good case for more exploration of the construing of relationships in eating
disorders. (Button, 1993, p. 210)
The purpose of this study is to determine the extent to which these structural
features present themselves across different domains of experience. In other words, this
study tests the generalizability of these effects across personal, interpersonal, and
vocational domains. The hope is that we can either narrow the range of significance of
these structural features (i.e., determine that they are evident only in interpersonal
construing), or document the extent to which they are present across a broad range of
other realms of construction.
In addition, this study examined the relative importance that weight-related
personal constructs have on the cognitive systems of disordered and nondisordered eaters.
It was hypothesized that the disordered-eating women would exhibit implicativelyy bound"
cognitive systems; i e., they would anticipate significantly greater changes than the
nondisordered-eating women in their personal, interpersonal, and vocational domains of
functioning if they were to suddenly experience a change in their personal weight.
The following chapter will review the literature in support of this study. In
particular, it will examine the cognitive features and sociocultural factors associated with
disordered-eating, as well as the studies that have applied personal construct theory to the
exploration of eating disorders. The chapter then concludes with a set of hypotheses that
will be examined through the methods outlined in the subsequent chapter (Chapter 3).
REVIEW OF THE LITERATURE
Following an introduction to the historical development of eating disorders, this
chapter reviews sociocultural influences on disordered-eating and then concludes with a
review of personal construct contributions to this literature. An understanding of the
historical development, sociocultural contributions, and personal construct research
literature will set the stage for the specific hypotheses that conclude this chapter.
Historical Origins of Eating Disorders
Anorexia nervosa was likely first recorded around 1689 (Silverman, 1995) when
a physician's notes on two patients were found to provide the first descriptions of this
particular eating disorder. Similarly, the presentation ofbulimia nervosa symptoms have
been reported as early as ancient times (Habermas, 1989; Parry-Jones & Parry-Jones,
Contemporary attention to eating disorders is great. It can be measured by the
reporting of eating disorders' prevalence and incidence and is highlighted by the growing
number of professionally-led disordered-eating psychotherapy groups, the creation of
"Eating Disorders' Coordinator" positions on college campuses, and the establishment of
the "Academy for Eating Disorders."2 The development of specialty journals devoted
A review by Ziolko (1996) offers excellent material for learning more about the
origins of these two eating disorders.
2 Established in 1993, this academy combats eating disorders primarily through
education, the defining and promoting of preventative interventions, and the identifying
and rewarding of outstanding and achievement or service to the study of eating disorders.
solely to eating disorders (e.g., the International Journal of Eating Disorders, Eating
Disorders Review, the British Review of Bulimia and Anorexia Nervosa, and Eating
Disorders: The Journal of Treatment and Prevention) also attests to the increasing
popularity of eating disorders as a subject of interest and professional study.
Some observe (Beumont, 1995) that the disorders characterized as eating
disorders would be more accurately described as dieting disorders, primarily due to the
high fatality rates particularly associated with anorexia nervosa which is approximately 5%
(Hsu, 1980; Neumarker, 1997). Others bluntly note that "dieting" is a very valid name
given that "die" is its root word (K. Kratina, personal communication, April 28, 1995).
Relatedly, and many years ago, Hilde Bruch (1978), often considered the founding
researcher of anorexia nervosa, observed that the term "anorexia" is misleading because its
translation is "poor appetite Bruch contended that its victims do not suffer from lack of
an appetite, rather, they are "ravenous." In all, it is clear that anorexia nervosa and bulimia
nervosa are disorders related to dangerous and extreme attempts to restrict eating.
Researchers and clinicians studying and treating individuals with mental disorders,
including eating disorders, use the Diagnostic and Statistical Manual of Mental Disorders-
IV (American Psychiatric Association, 1994) as a guide to accurately defining the disorder
in question. While the DSM-IV does not provide any treatment information, it is the
guiding manual used in procuring a diagnosis. The DSM has endured many revisions
since its inception in 1952 and only gradually has begun to differentiate and define the
criteria required for specific eating disorders.
In regards to eating disorders, the criteria have been revised at least three times.
These revisions reflect the continued advances in criterion-specification for eating
disorders. Thus, disordered-eating originated as an "Other" diagnosis (APA, 1968, p.80),
was then known as a "Feeding Disturbance" (APA, 1968, p.80), and later, a person with
disordered-eating was diagnosed as having either "Anorexia Nervosa" or "Bulimia"
because of "differing clinical pictures, courses, and treatment implications" (APA, 1980,
p.383). Later still, bulimia was renamed, "Bulimia Nervosa" (APA, 1987).3
In addition to the global name changes that have been made to the diagnostic
categories associated with eating disorders, specific changes in the criteria for anorexia
nervosa and bulimia nervosa have also occurred. First, a person currently diagnosed with
anorexia nervosa should weigh 15% below his or her normal body weight (APA, 1987,
1994), a change from the 25% criterion previously necessary for diagnosis (DSM-III;
APA, 1980). Second, a current diagnosis of bulimia nervosa (DSM-IV) mandates an
individual as having shape and weight concerns.' Lastly, various subcategories or
classifications exist for anorexia nervosa and bulimia nervosa (e.g., "Restricting" or
"Binge-Eating/Purging" type. In all, these criterion specifications, in addition to the name
changes for eating disorders, reflect the increased attention and research that these
disorders have demanded and continue to receive.
Sociocultural Factors Associated with Eating Disorders
This section seeks to examine the impact that sociocultural influences have on a
woman's personal, interpersonal, and vocational functioning as they may relate to
3 And, the most recent edition of the DSM (DSM-IV; APA, 1994) also differs
from prior editions: Disorders previously included within the Eating Disorders of Infancy,
and Atypical Eating Disorder categories) are now identified as "Feeding and Eating
Disorders of Infancy or Early Childhood."
4 This was not a requirement in the DSM-III's description of necessary criteria
(APA, 1980) Thus, a diagnosis of bulimia nervosa is more restrictive and is differentiated
from "binge eating" (a behavior, not a disorder, that can occur independent of weight
disordered-eating. First, however, it is important to highlight that there are many
researchers concerned about the media's role in the development of disordered-eating and
pursuit of thinness, as reflected by the growing number of studies and reviews on the
subject (Kilbourne, 1994; Levine & Smolak, 1996, Raphael & Lacey, 1992, Spillman &
Everington, 1989, Stice et al., 1994; Waller et al., 1994; White, 1992; Wolf, 1991). They
argue that it is necessary to study the media and its influence on the development of eating
disorders among young women by observing that television viewing, popular magazines,
and marketing techniques in general have a tremendous power to influence the emotional
and physical well-being of the young. The potential health risks of the young, who are
widely and frequently exposed to the media, have been observed and argued by many.5
Recognizing the general effects the media can have on individuals, we will now
discuss how sociocultural factors can influence a woman's physical or personal sense of
self (By personal sense of self, we mean the way in which a woman construes her looks or
appearance to others and any other overt characteristics of self). As highlighted earlier,
there are a growing number of approaches by which a woman can recreate her bodily self
so as to minimize, hide, even reverse or undo the natural body aging process.
The ability to reconstruct oneself, through cosmetic surgery for example, used to
be seen as something only wealthy women could afford. Thus, the pursuit of thinness
s Recently, in fact, state legislatures have contended that tobacco companies have
knowingly influenced youth to purchase their health-adversive products. One of these
companies profited greatly because of the marketing appeal that a cartoon camel has had
on adolescents. This company has (reluctantly) conceded as part of a landmark settlement
agreement to stop marketing the cartoon animal. See also Pierce, J P., Gilpin, E., Bums,
D. M., Whalen, E., Rosbrook, B., Shopland, D., & Johnson, M. (1991). Does tobacco
advertising target young people to start smoking? Evidence from California. Journal of the
American Medical Association. 266, 3154-3158.
through dieting or restricting or bingeing and purging allows women of all socioeconomic
backgrounds to achieve this desired and socially-approved physical goal.
In fact, some researchers have suggested that a relationship between a culture's
financial resources and thinness as a desired shape for women not only exists but
contributes to the etiology and maintenance of eating disorders. For example, M. Alteby
(personal communication, July 23, 1997) contends that a woman can never be too thin in a
society where food is abundant. She further elaborates that because individuals typically
show their wealth and status through their body,' a thin woman may be indirectly voicing
the position that while she can afford to purchase food and has the availability of food, she
chooses to restrict her food intake ("I'm so good and so confident, that I can be thin").
Thus, the thin ideal is not a physical state only afforded and accessed by the wealthy.
And, a woman is not alone in her striving for thinness. Importantly, the larger
sociocultural context broadly supports and sanctions her pursuit of the "obtainable" body
image ideal For example, Freedman (1986) notes that even though models of art in past
centuries were deemed as representations of the ideal, desired female shape, it was
recognized that those physiques were unattainable. Now, however, modern-day media
"blurs the boundaries between romanticism (i.e., obtaining the ideal body) and realism" (p.
A well-known and frequently cited study by Garner and his colleagues (Garner,
Garfinkel, Schwartz, & Thompson, 1980) underscored the misleading messages produced
6 For example, having tanned skin in modem times is often an indicator of a
Caucasian's financial ability to vacation; and yet this contrasts with other times when
having fair skin was a barometer of wealth and status because it indicated that the person
did not have to engage in any outdoor labor.
by the media regarding the likelihood of obtaining the ideal image. They reviewed the
shapes and sizes of Playboy centerfolds and Miss America pageant contestants for a
twenty year period. These researchers found that these two visible groups of women had
been getting thinner over the years, a finding particularly alarming as the results of another
study indicated that the average woman in America was becoming heavier (Society of
Actuaries and Life Insurance Medical Directors, 1980) during those same years.
Garner et al (1980) also discovered that during the same period (1960-1980)
popular women's magazines increased their number of articles on dieting. Given that
excessive dieting is often a precursor to the development of an eating disorder (Hsu,
1989), the results of the Garner et al. study (1980) have often been cited in the eating
disorders literature as indicative of the problematic sociocultural messages targeting young
The depth of this problem is further underscored by the quest for perfectionism as
an ideal and obtainable state.7 Women are continually barraged by advertising in which
youth and thinness are seen as obligatory aspirations in their lives. Thus, the cosmetic,
fashion industries, and more recently medical communities, take advantage of women'
"normative discontent" (Rodin, Silberstein, & Striegel-Moore, 1985) by promoting the
idea that women can and need to recreate themselves (Wolf, 1991). These groups
reinforce the belief that a woman is incomplete and unfeminine (both of which are
descriptions created and reinforced by particular political and social groups that prey on a
vulnerable woman's sense of identity) unless she has the "right" look (Mahowald, 1995;
7 See Wolfs (1991) first chapter, specifically pp 16-17, that describes the
substantial amount of money that women provide to the cosmetics and other-related
industries in their quest for beauty.
Timko, Striegel-Moore, Silberstein, & Rodin, 1987).8 Women learn that how they look
can be manipulated and should be manipulated.
In light of the sociocultural pressures regarding physical appearance, it may be
understandable that women are also seeking elective cosmetic surgery in record numbers
as they seek the perfect shape, size, and image. Women endure numerous risky
procedures in order to more closely resemble the appearance of an ideal woman. Thus,
while some women may restrict their pursuit of physical idealism to the wearing of
(colored) contacts, the highlighting, perming, straightening, weaving, or use of hair
extensions, and bleaching of teeth (services available on one end of the "makeover"
continuum), others, however, may choose to endure dangerous cosmetic surgeries 9
These latter more costly and dangerous procedures are becoming more acceptable, more
familiar, and more accessible than in any years past (Solomon, 1996).
Ironically, young girls seek the opposite. They yearn to look older than their years
by manipulating their appearance with makeup and dress and are rewarded for their
attempts. For example, last year one young girl in particular received much posthumous
notoriety as the media presented numerous photos and videos of the young girl performing
during state and regional beauty pageants. The girl was photographed and seen in a
8 And, this look is all encompassing: from the hairstyle (including cut and color),
to the scent (including perfumes, powders, and lotions), to the cosmetics (including
foundation, eyeliner, eyeshadow, eyebrow pencil, eyelash curler, mascara, blush, powder,
lipliner, and lipstick), to the clothing (including undergarments and hose), and, finally, to
the accessories (including shoes, handbag, jewelry, and briefcase).
9 Specifically, this latter group of women can choose to have their noses reshaped
(a painful procedure in which the physician must sometimes first break the nose), faces
lifted, eyes resculpted, lips enlarged (with collagen injections), chin and cheekbones
resurfaced, thighs liposuctioned, tummies tucked, intestines stapled, ribs removed, breasts
siliconed, buttocks tightened, or their bodies laser-smoothed
videotape with color processed hair, professional makeup coverage, and was adorned with
provocative clothing, the combination of which contributed to the projection of a persona
much older than her actual 6 years of age. Eventually, debate concerning the
appropriateness of the deceased girl's parents decision to "age" their daughter arose and
initiated (short-lived) concerns about the societal demands and messages that target young
girls and their appearance.
Feminist scholars and others observe that the phenomena of being discontent with
how old one looks extends to women of all ages; women are seeking to either look older
or younger. They believe this condition contributes greatly to the negative self-image and
self-esteem that many women have which in turn has been associated with the
development of eating disorders (Downs, 1997; Orbach, 1978, 1982; Rodin et al., 1985;
Sanford & Donovan, 1984; Stice et al., 1994; Striegel-Moore, 1993, 1995; Wolf, 1991).10
Consequently, many question why society (e.g., media) would perpetuate these unhealthy
proscriptions via advertising as it contributes to the dissatisfaction that women have with
their bodies, which in turn is a core (if not cardinal) feature of eating disorders."
10 Mary Pipher's "Reviving Ophelia: Saving the Selves of Adolescent Girls" (1994)
provides an excellent description and commentary about the difficulties and potential
dangers associated with the social development of young women.
The construction of the "Sociocultural Attitudes Towards Appearance
Questionnaire" (SATAQ) by Heinberg, Thompson, and Stormer (1995) reflects the need
for useful instruments for those individuals researching or working with individuals
presenting with eating disorders, eating disorder symptomatology, and/or body image
concern or preoccupation. Its 14 items are rated on a Likert-type scale which asks
respondents to identify whether they (1) Completely Disagree, (2) fall in-between
categories (1) and (3), (3) Neither Agree nor Disagree, (4) fall in-between categories (3)
and (5), or (5) Completely Agree with the statement. The 14 items are as follows. (1)
Women who appear in TV shows and movies project the type of appearance that I see as
my goal, (2) I believe that clothes look better on thin models, (3) Music videos that show
Thus, societal messages not only can dictate a woman's personal area of
functioning, but they can also prescribe her interpersonal relationship style For example,
women are taught to be responsible for developing and establishing social relationships.
Young girls "are socialized to be more interpersonally oriented than men" (Striegel-Moore
& Marcus, 1995, p. 447, see also Rodin et al., 1985). Consequently, a girl acquires skills
that are deemed necessary for sustaining relationships yet also may serve to devalue her
self-worth, self-esteem, and sense of individuality (Bruch, 1978; Mahowald, 1995).
Some writers conceptualize anorexia and bulimia nervosa as problems of identity
(Mahowald, 1995; Weinrich et al., 1985). One of these individuals, Mahowald (1995),
has chosen to explore a subcomponent of identity, gender identity In her writings, she
delineates between gender identity and gender role by describing the former as "the private
experience of gender role," and gender role as "the public expression of gender identity"
(p. 292) Some women may vehemently pursue the ideal feminine gender role, a pursuit
that can take the form of being excessively thin (e.g., anorexia nervosa). Should a woman
develop anorexia nervosa in her quest for thinness, she will also develop amenorrhea.
Some argue, including Mahowald, that the progression of amenorrhea can be interpreted
as a woman's attempt to avoid fulfilling the demands of her perceived gender role. Thus,
regardless the reason behind the development of anorexia nervosa, Mahowald contends
thin women make me wish that I were thin, (4) I do not wish to look like the models in the
magazines, (5) I tend to compare my body to people in magazines and on TV, (6) In our
society, fat people are not regarded as unattractive, (7) Photographs of thin women make
me wish that I were thin, (8) Attractiveness is very important if you want to get ahead in
our culture, (9) It's important for people to work hard on their figures/physiques if they
want to succeed in today's culture, (10) Most people do not believe that the thinner you
are, the better you look, (11) People think that the thinner you are, the better you look in
clothes, (12) In today's society, it's not important to always look attractive, (13) I wish I
looked like a swimsuit model, (14) I often read magazines like Cosmopolitan, Vogue, and
Glamour and compare my appearance to the models.
that anorexia nervosa is an expression of identity, an identity that society seeks to (rigidly)
define for a young woman
One study in particular exposes the difficulty in which a woman can experience
when seeking to develop and strengthen a healthy, feminine identity. In their classic 1970
study, Broverman and her colleagues (Broverman, Broverman, & Clarkson) sought to
identify the distinguishing characteristics and behaviors perceived as healthy for men, for
women, and more generally, as an "ideal standard of health" (p 1) They found that the
traits associated with ideal health in general were fairly similar to those traits associated
with ideal health for men, yet contrasted with those characteristics and behaviors equated
with ideal health for women.
The findings of a study by Basow and Kobrynowicz (1993) provides yet another
example of how women are evaluated on wide range of dimensions Specifically, these
researchers sought to determine whether the size (small and large salad, small and large
meatball hoagie) and gender connotation (with the salads being perceived as very feminine
meals and the meatball hoagies perceived as very masculine meals in a pilot study) of a
meal affects a woman's social appeal. The 113 college student participants (51 males, 62
females) viewed one of four videos that showed a woman eating either the small salad,
large salad, small hoagie, or large hoagie. They were then asked to rate her social appeal
as measured by a series of five questions on a 5-point Likert type scale developed by the
authors. Results found that the woman eating the small, feminine meal (salad) was
perceived as much more socially appealing than the same woman seen in the videotape
eating the large, masculine (meatball hoagie) meal. The authors interpreted this finding as
offering insight into why some women vary their eating behaviors depending on the
presence of others. Furthermore, they believe that if these same women seek to also
control their weight they may be at a heightened risk for developing an eating disorder. In
all, this study speaks to the social power that eating and size meal has on a woman's
Other researchers have investigated the characteristics deemed desirable in women
in personal ads. For example, in a study by Andersen, Woodward, Spalder, and Koss
(1993) 481 personal ads were analyzed "to determine self-report characteristics of males
and females seeking companionship as well as the desired height, weight, and shape
characteristics of the person being sought" (p. 111). They found that the female searchers
were more likely to describe themselves as thinner than average in weight, and either
smaller or taller than average in height. Male searchers also sought women different from
the norm. Their ads called for women who were "5 years younger, thin, fit, petite, or tall"
(p. 111). These descriptive findings reflect the continued sociocultural message that
desirable women are thin and fit. And, this message is known by women as indicated by
the characteristics the women in the personal ads used to "market" themselves to others in
the hopes of obtaining interpersonal happiness.
Some authors have observed that specific personality traits are often exhibited by
individuals with an eating disorder. For example, Button (1996) has recently completed a
prospective study that examines the role of self-esteem in the etiology of an eating
disorder. He measured the self-esteem in 594 girls at age 11-12 and again later when the
girls were 15-16 years old. Button found that those girls who showed low self-esteem at
the time of the first testing were significantly at greater risk of developing more severe
symptoms of an eating disorder four years later.
Other characteristics that have been associated with individuals with eating
disorders include hesitancy, insecurity, anger, stubbornness, awkwardness, withdrawnness,
low self-esteem, overly-compliant and polite behavior, as well as hard-working and high-
achieving abilities (Bruch, 1973; Button, 1996; Crisp and Fransella, 1972; Downs, 1997).
Katzman and Wolchik (1984) compared the personality and behavioral characteristics of
bulimics with controls. Their results indicated that bulimics were more depressed,
experienced poorer body image, reported higher self-expectations, a higher need for
approval, and greater restraint than participants who did not engage in binge eating. And,
individuals with anorexia nervosa have been described as chameleon-like in their social
interactions as they are facile in their ability to adapt their personalities to complement the
situation or persons present (Bruch, 1988). Bruch (1978) also notes that women afflicted
with anorexia nervosa lack assertiveness and have difficulty making decisions, yet there
has been some evidence contrary to Bruch's observation (Mizes, 1988).
The work of somatotypes, first introduced by W H. Sheldon (1940) has frequently
been used in relation to a wide variety of areas including personality, potential for self-
actualization, temperament, as well as eating disorders Sheldon's theory classifies human
physiques into three categories: endomorphy, mesomorphy, and ectomorphy. The first
two categories, endomorphy and mesomorphy, are both described as "compactness of the
body" (Hartl, Monnelly, and Elderkin, 1982, p. 5). Endomorphs, however, tend to be
individuals who produce surplus fat and store it throughout their body, thus they become
round in shape. Mesomorphs, on the other hand, have an athletic body image as reflected
by their tightly compacted bones, muscles, and tissues which provides them with strongly
developed arms and legs. The other somatotype, ectomorphy, is described as a physique
that is elongated; their limbs are often stretched out, and subsequently they are viewed as
thin individuals (Hart] et al., 1982).
For the past several years, the media has portrayed the thin body or what Sheldon
(1940) defines as the ectomorph physique, as the ideal female figure (Sanford & Donovan,
1985; Wolf, 1991). Spillman and Everington (1989) explored the characteristics assigned
to the somatotypes and found that individuals associate "strength, happiness, and
dominance" (p. 887) to the mesomorph physique, "social aggression, laziness, and
unattractiveness" (p. 887) to the endomorph build, and "nervous(ness), submissive(ness),
and socially withdrawn(ness)" (p. 887) to the ectomorphy physique.
In a similar study (Keas & Beer, 1992) participants were asked to identify which
somatotype endomorphh, mesomorph, or ectomorph) would best represent a series of 30
statements, some of which were. "The woman who is to be most wanted as a friend," "The
woman who would be the most aggressive," "The woman who would assume leadership"
(p. 225). Participants chose the ectomorphy (thin shape) somatotype most often for the
following statements (all of which begin with "The woman who"): smokes 3 packs of
cigarettes a day; eats the least often, would be most likely to have a nervous breakdown,
would make a poor mother; would be the least aggressive; would endure pain the least;
eats the least" (p. 225).
In light of the results of these two studies (Keas & Beer, 1992; Spillman &
Everington, 1989) one may be more likely to accept the argument that sociocultural
messages have enormous influence on young women and their vulnerability to eating
disorders. These powerful and pervasive messages that target young women to be thin
serves as one likely explanation as to why women would seek to alter their body build to
that of an ectomorph, despite the association of that body shape with otherwise negative
In all, there is significant quantitative and qualitative research that supports the
belief that sociocultural factors influence a woman's interpersonal traits and behaviors.
This assumption, coupled with the research findings that suggest these same factors also
negatively affect a woman's personal functioning (e.g., leads to a drive for thinness and
overall quest for ideal body image), introduces the next section that explores how
sociocultural factors influence a woman's vocational functioning.
Possessing a drive for thinness (which affects one's personal functioning) and being
dissatisfied with one's body (which interferes with one's interpersonal functioning) are but
two of many sociocultural factors that are believed to contribute to the prevalence of
eating disorders. Researchers are also cognizant of how the changing roles demanded of
women in this society may also contribute to the phenomena of the growing prevalence of
Specifically, as women continue to progress through this modern era, they have
similarly experienced many shifts in vocational roles Worrell and Remer (1992) recognize
these changes and observe that shifts need to be explored in the contexts they have
developed. First, more women than ever before are in the workplace and the demands in
the workplace can negatively affect all aspects of a woman's functioning (personal,
interpersonal, and vocational). For example, there have been growing reports of sexual
harassment being an experience that many women confront at work. Should a woman be
sexually harassed, she may be reluctant to confront her harasser fearing escalation of
harassment or even employment loss. However, should she choose to remain in the
problematic work environment she may experience depression, anxiety, helplessness, low
self-esteem, and anger as a result, and possibly exhibit diminished work capability. If a
woman does not possess adequate coping strategies or have access to a strong support
system it is likely that the aforementioned list of symptoms will be channeled in less
effective coping styles. For example, the powerless woman may seek control in her life
through her eating behaviors.
Second, while much has improved regarding employment discrimination in relation
to the hiring of women, it continues to exist and consequently affects a woman's
interpersonal and vocational functioning. Specifically, a woman may experience
employment discrimination should she seek employment in an environment in which her
body-build does not "match" the setting. For example, in a review of sex stereotype
research findings, Ruble and his colleagues observed that people perceive suitable jobs for
women as those that require "gentleness, understanding, and warmth" (Ruble, Cohen, &
Ruble, 1984, p. 342).
This review may have inspired Keas and Beer (1992) to design a survey that
incorporated the application of Sheldon's (1940) theory of somatotypes with stereotypes
held in general regarding the gender suitability of ajob They found the following jobs to
be perceived as most suitable for an ectomorph (thin build) somatotype stenographer,
private secretary, dental assistant, telephone operator, florist, piano tuner, beautician,
waitress, shoe salesperson, office secretary, bookkeeper, file clerk, private secretary,
typist, cashier, and artist Thus, body-build stereotypes were found to exist among the
group of participants. The mindful reader observes that the occupations deemed most
appropriate for an ectomorph body shape are also traditionally female occupations.
In fact, in a review of sociocultural factors associated with the risk of developing
eating disorders, White (1992) identified dancing, modeling, and gymnastics as professions
with the greatest incidence of disordered-eating These careers are breeding grounds for
the development of anorexia nervosa or bulimia nervosa because of their emphasis on ideal
body image and shape. Thus, the young girls that enter these professions soon learn to
pursue thinness, monitor their body shape, and may diet as a result of the demands for
maintaining that "perfect" body shape. And, again, dieting is often a precursor to the
development of eating disorders (Hsu, 1989).
Cognitive Features Associated with Eating Disorders
This review now turns from addressing the main sociocultural factors associated
with eating disorders, to an examination of the cognitive features of eating disorders. One
of the main (dysfunctional) cognitive features associated with individuals diagnosed with
either anorexia nervosa or bulimia nervosa is a "disturbance in perception of body shape
and weight" (APA, 1994, p.539). Two other factors also appear to be forerunners in
contributing to, as well as maintaining, these eating disorders (White, 1992): pursuing
thinness and perfectionism. The primary difference between clinical and subclinical groups
is the greater extent to which the former group is willing to sacrifice components of
oneself (e.g., mental and physical health) in order to be "successful" in maintaining a
specified body weight or size. In fact. Butow, Beumont, and Touyz (1993) found that in
comparison to nonrestrained eaters, anorexic patients were likely to evaluate their self-
worth almost entirely in terms of self-control.
The first of these cognitive characteristics has generated significant attention,
research, and debate in the past several years (Hsu & Sobkiewicz, 1991; Williamson,
Cubic, & Gleaves, 1993). While the concept "body image disturbance" has been espoused
by many as a cardinal feature of anorexia nervosa and bulimia nervosa (APA, 1994, Bruch,
1962), some argue that it is incorrectly being interchanged with "body size
overestimation" and "body disparagement" (Hsu & Sobkiewicz, 1991; Williamson et al.,
For example, Hsu and Sobkiewicz (1991) reviewed 19 studies that explored "body
image" of individuals with anorexia nervosa or bulimia nervosa. They argue that these
studies explored "body image disturbance" by focusing on the perceptions and attitudes or
emotions that one has of his or her physical self Perception distortion was usually
measured by techniques that included image marking, analogue scales, optical distortions,
and silhouette-card sorting, whereas attitudes toward one's body were typically measured
through the use of questionnaires (e.g Eating Disorder Inventory, Garner, Olmsted, &
Hsu and Sobkiewicz (1991) found that while some anorectics and bulimics did
overestimate their body width (i.e., exhibited body perception distortion), this finding did
not extend to all or even most anorectics or all or most bulimics. And, in regards to
attitudes and affect towards body, "many bulimics and some anorectics (were) more
dissatisfied with their bodies (yet), not all patients showed) such dissatisfaction" (p.
24). Thus, these authors argue that body dimension estimation cannot measure body
image' and therefore, argue that Bruch (1973) initiated the erroneous leap in linking fear
of fatness and pursuit of thinness with distorted body image.
Regardless, it is agreed that individuals with disordered-eating are hindered in their
ability to accurately gauge their body size and exhibit dysfunctional cognitions in relation
2 Readers might be interested in Hsu and Sobkiewicz's (1991) literature review on
variables that appear to influence one's body size estimation. For example, they found that
"sex age ... actual size of subject ... self-esteem ... masculinity and feminity
whether the subject estimated according to what they thought or how they feel" to
influence a person's body size estimation
to the meaning of body shape and weight, and meal size and frequency (Fairburn, 1981;
Garner & Bemis, 1982, Mizes & Christiano, 1995).
Another cognitive factor associated with eating disorders, valuing thinness, has
been endorsed by society periodically throughout history. In a 1996 documentary,
Solomon outlined the historical changes and patterns of ideal female body images from
1400 BC until present day. Its descriptions of the changing ideal female shape
underscores the waxing and waning of valuing thinness: Specifically, the documentary
reported that the ideal female shape in 1400BC was influenced by a 5'2", 110 pound
Egyptian queen. Later, in 500 BC, a more muscular version of the ideal female shape was
introduced and reportedly resembled the Roman goddesses who typically possessed a 5'9",
150 pound frame. These ideal muscular physiques were replaced in the 1300s by more
voluptuous ones and were characterized by 5'8" and 180 pound women. In the
seventeenth and eighteenth centuries, the true shape of women was unknown because
women's bodies were mostly camouflaged by layers of lace and skirts. However, women
embraced the thin appeal to a large degree as reflected by their enduring steel corsets in
order to procure a fashionable 18 inch waist." In the early part of the twentieth century
(e.g., 1914), dieting was the norm as women continued to seek a thin ideal. In the 1930s,
however, the ideal female shape was dichotomous in nature as both the petite and more
mature shape was accepted. Twenty years later, the ideal female shape was no longer
petite and Marilyn Monroe's full figured shape represented the ideal. In the late 1950s,
thinness was again the desired shape, yet by the 1960s, it was replaced by an even
scrawnier physique as reflected by the mass appeal and success of Twiggy, a 5'6", 97
"1 And, these corsets were drawn so tight that bones might break and internal
organs might be injured.
pound British supermodel Currently, thinness continues to be pursued by those seeking
to possess the ideal female shape
While individuals diagnosed with anorexia nervosa have a pursuit of thinness and
fear of fatness, some researchers have argued that their true fear is of being normal
weight. Specifically, Crisp (1967) initially observed (and Hsu & Sobkiewicz, 1991
concurred) that fear of fatness is not a psychopathological fear, as it is normal in our
society for individuals to fear fatness, but that it is abnormal to fear normal weight.
Because there is either a fear of fatness or an exaggerated concern or influence of
body shape and weight on self-evaluation by eating-disordered individuals, these same
persons typically seek to alter their body size and shape. This can be particularly
destructive when considered in relation to the third feature associated with eating
disorders, a quest for perfectionism
More specifically, the trait of perfectionism is a complementary behavior to the
valuing of thinness and dieting as it assists the individual in engaging, developing, and
ultimately strengthening the rigid thinking and behaviors necessary for achieving ultra
thinness. An exchange between an eating disorders researcher and clinician (Button,
1993) and a woman who had recovered from anorexia nervosa illustrates the strength and
dangers that perfectionism can bring to this eating disorder:
Button: What helped you to be more able to handle these things (social
Woman: Confidence in myself .. basically now I like myself .. the
perfection is not there anymore.
Button Why do you think perfection was so important to you?
Woman: It was this failure thing .. there are only two sides to an
anorectic--you are either perfect or a failure. My personality was
very black and white. People either loved me or hated me. There
was nothing in between those-that is how I viewed my life (Button,
1993, pp. 114-115).
Thus, anorexia nervosa (or bulimia nervosa) can easily develop when a woman has
accepted thinness as an ideal, begins to restrain her eating, and uses her skills of
perfectionism in assisting her in her goal of thinness, for example.
Personal Construct Literature as Applied to Eatine Disorders
Since distorted self-perceptions play a significant role in the development of eating
disorders it would be advantageous to examine how an individual perceives himself or
herself, as well as others. Several psychological interventions could be used to obtain this
information: clinical interviews, surveys, projective tests and other personality tests; but
one instrument in particular, the repertory grid test, appears to be very favorable for
eliciting this desired knowledge (Mottram, 1985). This instrument, developed by George
Kelly (1955), has been applied to a variety of fields, in both clinical and educational
settings (see Neimeyer, 1988; and, Neimeyer & Neimeyer, 1987).
The repertory grid can be used to obtain differentiation and integration scores for
an individual. Differentiation is defined as the number of different dimensions of
judgement an individual uses when making a decision (Bodden, 1970), whereas integration
is explained as the level of organization the cognitive system can incorporate in the
decision-making process (Cochran, 1977). The repertory grid's objective yet personal
approach has significant benefits when seeking to apply it to the sensitive and complex
issue of eating disorders (see Mizes and Christiano, 1995, for a review of questionnaires,
procedures, and methods "used in the assessment of eating disorder cognitions," p.95).
Fransella and Crisp (1970) and Crisp and Fransella (1972) were the first to apply
the use of a repertory grid to the area of eating disorders and did so in a case study
context. More specifically, Crisp and Fransella (1972) administered a series (seven testing
occasions) of repertory grids to two anorexic women during their course of treatment at a
hospital. The two grids were constructed with differing elements, and sometimes with
different constructs. Crisp and Fransella named one grid a "people" grid, the other a
"fashion model" grid. The former grid included elements of a variety of people including
the patient at specific principal weights (e.g., her weight before puberty). Its constructs
were both elicited and imposed (provided) in nature. The latter grid was comprised of 10
elements that were 10 pictures of women from a fashion magazine and all of its constructs
were imposed (e.g., helpful and kind, mature, irritable).
Their results suggested that those suffering from this particular eating disorder are
not likely to improve until their weight is no longer a significant determining factor to their
sense of self During the recovery process significant changes became evident among the
correlations of constructs. For example, one of the women initially construed the
construct "mature" as very related to the construct "likely to become pregnant." Later, at
the end of treatment, the same woman exhibited negative correlations between these
constructs. Crisp and Fransella interpreted this change as reflecting the woman's growing
sense of identity and her ability to discriminate more broadly among constructs. "She
appears to have swung from one pole of a single limited way of construing the world to
the other pole" (p. 405)
Button (1983) elaborated on Crisp and Fransella's approach and used a repertory
grid technique both before and after treatment with a subject pool of 20 inpatient
anorexics. Using a repertory grid consisting of 20 elements (comprised of the names of 10
important people in the patient's life and 10 provided elements such as "Me," "Me if I Was
Overweight," and "Me as I Would Ideally Like to Be") and 20 constructs, he found that,
overall, these women construed themselves in relation to their weight more heavily prior
to treatment than after treatment. However, these same women finished treatment
without any alternative self-construction. Their post-treatment view of self was
ambiguous, and subsequently, they were more likely to relapse.
In 1984, Hutton sought to draw upon Crisp and Fransella's (1972) and Button's
(1983) research aimed at applying repertory grid methodology to the study of eating
disorders. She administered and then analyzed the cognitive measures associated with 4
groups of participants: anorexics, bulimics, recovered anorexics, and a control group.
Hutton observed that the former two groups (those representing disordered-eating
symptomatology) exhibited a "tendency to attribute different personality traits to
themselves at different body weights" (p. 353). This work supports the current's study's
goal that seeks to document the nature and extent to which perceived weight influences
personal, interpersonal, and vocational domains of function.
An often-cited study by Neimeyer and Khouzam (1985) provides additional
evidence regarding the utility of the repertory grid technique as applied to the area of
eating disorders. They examined the cognitive processes of high and low "restrained
eaters" by using repertory grids to obtain cognitive complexity scores (i.e., differentiation
scores). Neimeyer and Khouzan then compared the complexity scores of both these
groups and found that high restrainers exhibited less complex systems in a variety of eating
situations compared to the low restrainers. In other words, the high restrainers possessed
fewer dimensions (i.e., were less differentiated) which limited their ability to flexibly
engage in interpersonal situations: "More restrained eaters see themselves as having fewer
alternatives available in relation to eating and therapy may be aimed, in part at
expanding their repertoire of self-constructions" (p. 368).
Other personal construct research that underscores the applicability of personal
construct theory to eating disorders can be found in Button's 1993 work. In this
publication, Button provides an instructive narrative regarding the interplay of these two
fields of study. He describes how eating-disordered individuals typically have one broad
interpretation of the world: they focus on weight. Button further notes that they are
similar to noneating-disordered people in that they will envision particular events in
relation to domains of interpersonal interactions and vocation, yet they might "also have
particularly important anticipations with regard to weight, food and eating" (p.34). His
description of this phenomena is important:
faced with an invitation to go out to a party, this is interpreted less as an
opportunity for fun, meeting someone, letting one's hair down, getting drunk,
exploring ideas, making love, etc, than as a matter of weight/eating: 'I'm too
fat'. .'I'd have to eat' (p.32).
Thus, Button uses personal construct theory to assist him in conceptualizing
eating-disordered individuals He views these individuals as those who use food, weight
and eating behaviors as a way of predicting their intrapersonal and interpersonal worlds.
Each person is unique in their various interpretations of self and others, and, a personal
construct orientation respects this individualism.
Given the role that sociocultural and personal construct theory factors have in the
understanding of eating disorders, the current study sought to extend previous work in
this area in two ways. In particular, we wanted to test the (1) generalizability of
cognitive/structural features across personal, interpersonal, and vocational domains; and,
(2) the implications that perceived weight gain may have on personal, interpersonal, and
vocational areas of functioning (suggesting the centrality or importance of weight in
various domains of the individual's experience).
Purpose and Hypotheses of the Study
The studies discussed in this chapter suggest the likelihood that the cognitive
features associated with an eating disorder extend beyond the domain of eating behaviors,
per se. In other words, these effects are likely to extend across numerous domains,
including overt physical characteristics, interpersonal functioning, and vocational pursuits.
Thus, this study explored these potential widespread effects and used previous work in
this area (personal construct theory as applied to eating disorders) as its foundation
(Button, 1983, 1993, 1996; Crisp & Fransella, 1972, Fransella & Crisp, 1970; Neimeyer &
Khouzam, 1985). In so doing, it examined self as well as other-construing differences that
might exist between restrained eaters and non-restrained eaters. This study compared the
cognitive complexity scores (differentiation and integration) of High and Low eating-
disordered individuals, as determined by high and low scoring on the Eating Disorders
Inventory (Garner et al., 1983).
Two main hypotheses were central to the study. First, we hypothesized that
women exhibiting greater eating-disordered characteristics would be more integrated and
less differentiated than women exhibiting fewer eating-disordered characteristics, and that
this finding would extend across all three domains of experience (personal, interpersonal,
and vocational). The former part of this hypothesis was supported by previous personal
construct and eating disorders research findings, and the latter part was an attempt to
generalize those findings
We also hypothesized that the group characterized by more eating disorder
symptomatology would anticipate greater changes across personal, interpersonal, and
vocational domains if confronted with sudden weight gain (20% of their current weight).
This second hypothesis was based on previous work that revealed eating-disordered
women possess more integrated cognitive systems than noneating-disordered women. In
other words, a significant change in a disordered-eating woman's tightly construed
construct system (e.g., weight gain) introduces many implications for other constructs in
the system (e.g., personal, interpersonal, and vocational functioning).
Given these two main hypotheses, we also predicted significant correlations
between levels of Eating Disorders and measures of cognitive differentiation (negative
correlation) and integration (positive correlation). Additionally, EDI scores were
expected to be correlated with the degree of change implied by weight gain, with higher
EDI scores being positively correlated with the greater degree of implied changes in
personal, interpersonal, and vocational domains of experience.
In order to assess further the importance of these constructs in relation to a
woman's personal, interpersonal, and vocational functioning, we also wanted to assess the
importance of weight-related constructs in the systems of eating-disordered women.
Given that women exhibiting disordered-eating perceive themselves as needing to
strive for perfection or "success" in many areas of their lives (Bruch 1978; Sanford &
Donovan, 1984; Wolf, 1991), this "superwoman" expectation of self may extend to others
as well.". In fact, Beebe and his colleagues (1996) investigated the extent to which this
superwoman phenomenon may exist as related to weight by comparing whether "women
who focus on their own bodies place a similar focus on body shape when evaluating
others" (p. 415). They found that when presented with picture slides of women in various
situations, high scorers on the Eating Attitudes Test (Garner & Garfinkel, 1979) tended to
"4See Hayes' (1986) "The Superwoman Myth" in which this so-called myth is still
endorsed by many women who have high expectations for their professional and personal
focus on the weight-related and body shape features of women in the pictures. While this
finding did not reach statistical significance, results were in this direction.
Thus, in light of the research studies that have revealed the extent to which
disordered-eating women may focus on others' body shapes as well as their own, the
authors of this study anticipated that the High Eating-Disordered group's construct
importance for evaluating other women would be in the order of personal, interpersonal,
and vocational domains of experience. The direction of this hypothesis was also
influenced by the sociocultural recognition that women are socialized to sharpen their
social and interpersonal skills, while maintaining a "perfect" personal physical image
(Bruch, 1978; Hayes, 1986; Wolf, 1991).
Relatedly, results of a study by Broverman and her colleagues (1970) indicated
that interpersonal constructs are prominent to a women's perceived sense of well-being.
Interpersonal characteristics that these researchers found to be typically associated with
healthy women included items such as "very talkative ... very tactful ... very gentle
very aware of feelings of others ... very quiet" (p. 3).
And, while there has been some inquiry as to how women unduly concerned with
weight and body shape regard others' physical selves, there has been little interest or
attention regarding how much these women might consider weight-related constructs as
central to their evaluations of men. In other words, while preliminary research is tending
to suggest that women who evaluate the many parts of themselves in relation to weight
may also hold other women to a "weight-comparison" stick, research has not explored
whether or not this comparison also extends to their appraisal of men.
Although speculative, we expected women to place greater importance on
constructs concerning vocations when construing men, followed by interpersonal, and
personal constructions. This expectation was based on results of research such as that
conducted by Broverman et al, (1970). Specifically, they found that healthy maie
attributes were typically associated with power and autonomy, both of which are traits
most often obtained via a man's occupation.
In summary, a series of predictions were made regarding the two groups of Eating-
Disordered women (High and Low) across the three domains of experience (personal,
interpersonal, and vocational). Specifically, main effects for EDI score were predicted
along structural measures of differentiation and integration, construct implications, and
construct importance, Individuals scoring high on the EDI (High Eating-Disordered) were
expected to exhibit significantly higher levels of integration, implications, construct
importance, and lower levels of differentiation compared to individuals scoring low on the
EDI (Low Eating-Disordered). These hypotheses were tested within the context of a
quasi-experimental research design, as detailed in the following chapter (Chapter 3).
The primary purpose of this study was to describe and explore the cognitive
features and differences between females who differ in their eating and body shape
attitudes, feelings, and behaviors In particular, this research sought to determine the
extent to which these differences may generalize across domains of personal construction
(personal, interpersonal, and vocational domains)
Participants consisted of 40 female students enrolled in undergraduate
Introductory Psychology classes at a large southeastern university. These students were
selected from a larger sample (n=671) of undergraduates based on their extreme scores on
subscales of the Eating Disorders Inventory (Garner et al., 1983; and see Pre-Screening
below). Students voluntarily participated in the study with the understanding that its
intent was to investigate decision-making in relation to appearance, personality, and
vocation. Students were able to partially fulfill course requirements by participating in the
study. The mean age of the group was 18 years with a range of 17 to 21 years
Instruments and Procedure
In order to secure samples of high and low eating disordered women, three
subscales of the EDI were administered to a total of 671 undergraduate students as part of
a large protesting session Ensuring participation of both high and low scorers on the
selected EDI subscales was necessary for posttesting (see below). Participants (n=671)
were arranged according to their preadministration testing scores on the Drive for
Thinness subscale, Bulimia subscale, and Body Dissatisfaction subscale, from the highest
to lowest scores (See Table 3-1 for relevant means and standard deviations). Specifically,
a score for each participant for each of the three subscales was obtained. Only those
participants (n=182) who scored highest (top 33 percentile; n=98) or lowest (bottom 33
percentile, n=84) on all three of the selected EDI subscales were selected for potential
inclusion in the study (Garner et al., 1983).'
SSpecifically, for each of these three subscales, participants were asked to respond
to a series of related statements with either an "Always," "Usually,""Often," "Sometimes,"
"Rarely," or "Never" answer (See Appendix A). This 6-point response system is
generally scored by weighing the responses from 0 to 3, with a score of 0 being assigned
to those 3 responses farthest away in directions from the symptomatic direction, whereas a
score of 1,2. or 3 is assigned to those responses in the symptomatic direction.
The basis for this 0-3 scoring system as opposed to a 1-6 scoring system is
grounded in a theoretical discussion (see Garner & Olmsted, 1991) It is believed by some
that evidence of symptomatology exists only for those responses weighted 1-3. All
responses that might have otherwise received a 4-6 rating if using a 1-6 scoring system,
are instead all weighted a 0, as they are assumed to be nonsymptomatic, and hence should
not contribute to the total subscale score that reflects psychopathology.
Another scoring system, however, is supported by Schoemaker, Strien, and Staak
(1994). They recognized that the EDI is becoming more frequently and widely used as a
screening instrument for nonclinical populations and subsequently, questioned the integrity
of the psychometric properties of the transformed scoring system (transforming the 6
point responses into 4 point responses) when studying a nonclinical population. These
researchers found that a nontransformed scoring system (using a 1-6 scoring system and
not a 1-3 system) yields higher validity scores which also reflects increased sensitivity of
the instrument. Other researchers have used untransformed subscale responses and argue
that in so doing they obtain "a full range of scores" and that the EDI becomes more
sensitive to differences between participants (Thompson, Berg, & Shatford, 1987, p. 230).
Thus, without transforming the point-response items, some researchers believe that the
EDI "would not differentiate the subject responding 'sometimes' from another who
responded 'never'" (Thompson et al., 1987, p. 230).
In this study, an untransformed (0-5) scoring system was implemented, for both
theoretical as well as empirical reasons. First, researchers sought to respect the
delineation of symptom-evidence responses from nonsymptom-evidence responses (which
is congruent with the 0-3, transformed scoring system) by selecting only those participants
who scored either 0-1 on all three subscales to be included as part of the most extreme
symptomatic group. Similarly, we wanted only those participants who scored either a 4 or
Potential Participants' Prescreening Means (and Standard Deviations) Organized by
EDI Subscale Bottom 33% Middle 33% Top 33%
(n=84) (n=489) (n=98)
Drive for Thinness .61 (.41) 2.11 (1.14) 4 08 (.53)
Bulimia 29 (.15) .79 (.48) 1.78(73)
Body Dissatisfaction 1.10(.58) 2.67(1.03) 4.23 (.55)
These prospective participants (n=182) were then called by a research assistant
who started calling those participants who scored in the extreme (i.e., highest and lowest
scorers). This assistant informed these prospective participants of a study they might wish
to participate in as part of their Introductory Psychology research requirement. Calling
continued in stepwise progression until 40 volunteers (20 participants who scored in the
top third percentile or "high" EDI scorers, and 20 participants who scored in the bottom
third percentile or "low" EDI scorers) were obtained for the study. (See Table 4-2,
Results, for relevant means and standard deviations).
Eating Disorders Inventory
The original Eating Disorder Inventory (Garner et al., 1983) is a widely used
(Rosen, Silberg, & Gross, 1988), 64-item self-report index that is designed to provide
descriptive information relevant to psychological functioning and eating disorders Its
eight subscales (Drive for Thinness, Bulimia, Body Dissatisfaction, Ineffectiveness,
Perfectionism, Interpersonal Distrust, Interoceptive Awareness, and Maturity Fears)
5 on all three subscales to be included as part of the least extreme symptomatic group.
However, due to sample size problems, we instead conducted a tertile split on all three
subscales. We then identified those participants scoring highest (top 33% on each of the
three subscales) and those participants scoring lowest (bottom 33% on each of the three
subscales). Thus, those exhibiting the most extreme and least extreme responses were
identified as potential testing participants
"have been used to select or define criterion groups in studies of body
satisfaction/dissatisfaction, weight preoccupation, and perfectionism" (Garner & Olmsted,
1991, p. 5) Additionally, the EDI has been used with nonclinical populations using
transformed and untransformed responses (see Schoemaker et al., 1994, for a discussion).
The first three of the EDI's subscales, Drive for Thinness, Bulimia, and Body
Dissatisfaction, are recognized as those EDI subscales that assess attitudes, feelings, and
behaviors related to eating and body shape As other researchers have done (see
Heatherton et al., 1995), we determined that these three subscales were most applicable to
our prescreening needs. The first of these, the Drive for Thinness subscale (7 items, e.g.,
"1 am terrified of gaining weight," "I am preoccupied with the desire to be thinner"), is
designed to measure a person's undue concern with weight (including fear of weight gain
and regard for dieting). High scores on the Drive for Thinness subscale reflect excessive
attention to dieting and weight as related to an unrelenting desire for thinness.
Relatedly, the Body Dissatisfaction subscale (9 items, e.g., "I think that my
stomach is too big," "I think my hips are too big") seeks to identify individuals who are
displeased with their body shape or size, whereas the Bulimia subscale assesses the
"tendencies to think about and engage in bouts of uncontrollable overeating (bingeing)"
(Garner, 1991, p. 5). Thus, high scores on the Body Dissatisfaction subscale reflect a
greater unhappiness with one's physique, and high scores on the Bulimia subscale (7 items,
e g., "I have gone on eating binges where I felt that I could not stop," "I eat moderately in
front of others and stuff myself when they're gone") identify individuals likely to binge
which may be followed by self-induced purging.
Clinicians and researchers alike believe that the five remaining subscales
(Perfectionism, Ineffectiveness, Interpersonal Distrust, Interoceptive Awareness, and
Maturity Fears) assess more general psychological traits that are clinically relevant to
eating disorders. We chose to also include the first of these, the Perfectionism subscale, in
our testing administration because of its critical role in the development and maintenance
of eating disorders (Bruch, 1978) Also, it is recognized as important to the assessment of
eating disorders (Button, 1993; Cooper, 1995; Heatherton et al., 1995; Hewitt, Flett, &
Ediger, 1995; Slade, 1982; Srinivasagam et al., 1995; Terry-Short, Owens, Slade, &
Dewey, 1995; Thompson et al., 1987, White, 1992) and hence its inclusion in the Eating
Disorders Inventory (Garner et al., 1983). Items on this scale (6 items, e.g., "I feel that I
must do things perfectly or not do them at all," "I have extremely high goals") measure the
extent to which one believes "that only the highest standards of personal performance are
acceptable" (p. 6).
The EDI generally possesses favorable psychometric values, with an average item-
total scale correlation of 63 (Garner et al., 1983), an internal consistency for the subscales
reported as Cronbach alphas all above .80 (Garner et al., 1983), and evidence for
construct validity as demonstrated by "congruence between clinicians' ratings and patients'
subscale scores" (p. 28) Additionally, evidence of convergent and discriminant validities
is indicated by the high correlation of the Drive for Thinness subscale with the Eating
Attitudes Test (Garner & Garfinkel, 1979), r = 88, p<0.001, as well as the significant
correlations of both the Bulimia and Body Dissatisfaction subscales with "restraint"
(Herman & Polivy, 1975), r = .44, p<0.001 and r = .42, p<0.001, respectively.
Research assistants verbally informed all participants about the study (see
Appendix B). Specifically, they were told that they would be asked to complete 6 pen and
paper forms, that their responses were anonymous, and that they should not write their
name or any identifying information on the forms so as to protect their identity
Participants did not have to answer any question they did not wish to answer and could
choose not to participate or could end their participation at any time. They were
instructed about the expected length of time needed to complete the study and were asked
if they had any questions. They were provided with the names and telephone numbers of
the researchers should they have any questions about the study. Finally, they were
reminded that the study was confidential and were asked not to discuss it with anyone. All
participants chose to continue to participate in the study
All participants received three different Grid Forms in randomized order: a
Personal Grid Form (see Appendix C), an Interpersonal Grid Form (see Appendix D), and
a Vocational Grid Form (see Appendix E). In each case, the administration of the
repertory grid consisted of two stages. In the first stage, participants were provided with
a list of 10 elements (role titles), "for which the (participants) ha(d) to supply the names of
people for each role title" (Button, 1993, p. 51). These elements included Yourself,
Spouse or Significant Other, A Same-Sex Friend, An Opposite-Sex Friend, A Parent, A
Person You Dislike, An Overweight Person, Someone in Authority, A Successful Person,
A Favorite Professor. The selection of these 10 role titles was guided by other research
that implemented use of repertory grid procedures with eating disordered women as the
population of interest (Butow et al., 1993; Clark, Hershgold, & Rigdon, 1984; and, see
Button, 1993 for a discussion on element selection).
In the second stage of the repertory grid administration, the participants were
asked to rate each of the 10 persons (as identified and supplied by the participants in the
previous step) along three sets of 10 constructs: one set of 10 personal constructs, one set
of 10 interpersonal constructs, and one set of 10 vocational constructs in randomized
order. For example, on the Personal Grid Form participants were asked to rate each of
the 10 people on primarily physical attribute constructs that included
Underweight/Overweight. Attractive/Unattractive, Firm/Flabby, Feminine/Masculine,
Tall/Short, Graceful/Awkward, Photogenic/Not Photogenic, Well-groomed/Sloppy, Well-
Proportioned/Ill-Proportioned, Good Complexion/Poor Complexion. The majority of
these constructs were generated by the author because they reflect the physical attributes
commonly regarded as important factors in the favorability of a personal image However,
some of these constructs have also been included in other studies (Butow et al., 1993;
Similarly, on the Interpersonal Grid Form participants were asked to rate each of
the 10 people on the following constructs. Stubborn/Not Stubborn,
Competitive/Cooperative, Confident/Insecure, Friendly/Hostile, Authentic/Fake,
Angry/Not Angry, Energetic/Not Energetic, Structured/Spontaneous,
Optimistic/Pessimistic, Intelligent/Not Intelligent This set of constructs was also created
by the author who sought to include constructs that could be characteristics of an
individual's personality. However, some of these elements have been found in other
personal construct research as applied to eating disorders (Butow et al., 1993). Hence,
they are also important traits that characterize a person's interpersonal style.
Lastly, all participants completed the Vocational Grid Form. They were asked to
rate one of their parent's vocations on the construct "High vs. Low Salary" on a 6-point
Likert-type scale. They then continued to rate each of the 10 people on the remaining
bipolar constructs, including Much Education/Little Education, High Prestige/Low
Prestige, Influences People/Doesn't Influence People, Creative/Not Creative, Helps
People/Doesn't Help People, Much in Demand/Seldom in Demand, Interesting Work/Dull
Work, Works with Thoughts/Works with Hands, Offers Much Security/Offers Little
Security These constructs were derived from Bodden's (1970) Cognitive Differentiation
Grid, the single most commonly used repertory grid in the study of career counseling and
Two structural measures, differentiation and integration, were then derived from
each of these three grid from types for each participant using a computer program by
Landfield (1977). Differentiation refers to the number of different dimensions an
individual uses when making decisions, whereas integration refers to the degree of
correlation among these dimensions (Kelly, 1955). The reliabilities associated with these
two measures have been reported as generally high across various adaptations of the
reptest, with a one-week test-retest reliability for integration reported as r = .75, and a
one-week test-retest reliability for differentiation scores as ranging from r = .73 (Russ-
Eisenschenk & Neimeyer, 1996) to r = .89 (Feixas, Moliner, Montes, Mari, & Neimeyer,
After having completed the three grids (Personal, Interpersonal, and Vocational) in
randomized order, participants were asked to complete a fourth grid, a modified
Implications Grid (see Appendix F). The Implications Grid (Hinkle, 1965) is based on
Kelly's (1955) assumption that personal construct systems are hierarchically organized,
with some constructs occupying relatively important, central, or superordinate positions
within the system. The implications grid provides a direct measure of this hierarchical
organization by asking the participants to identify the degree of anticipated changes in
their personal construct system associated with a specific change in themselves. For
example, a participant might be asked to imagine changing from being an "outgoing"
person to a relatively "shy" person. They would then be asked, "What other constructs
would you imagine that you would change along if you were to change from being
outgoing to relatively shy?"
In this study, a modified version of the implications grid was used. Specifically,
participants were asked to imagine that they had experienced a 20% increase in their body
weight (i.e., a 150-pound participant was told to imagine that she now weighs 180
pounds). In order to ensure that they performed this task correctly, they were then asked
to write down both their current weight and newly imagined increased weight. To assist
participants in determining their new weight, a conversion chart was included on the same
page as the implications grid they were to work on next (see bottom of Appendix F).
Research assistants were careful to assure participants of the confidentiality of their
responses in this, and other phases of the study.
Participants were then given the three sets of 10 constructs already used in the
personal, interpersonal, and vocational grids in random order and asked to indicate, given
a 20% increase in their weight, "How much would you expect to change along each of
these dimensions?" Participants were instructed to respond to this question by placing a
checkmark next to "No Change," "Little Change," "Moderate Change," or, "A Lot of
Change" for each of the thirty constructs
Three dependent measures for each participant were calculated, reflecting (1) the
degree of anticipated changes in personal constructs (personal implications), (2) the
degree of interpersonal changes (interpersonal implications), and (3) the degree of
vocational changes (vocational implications).
And finally, in order to explore the relative importance of personal, interpersonal,
and vocational constructions for judging other people, we asked High and Low Eating-
Disordered groups to rate the importance of each of the three Specifically, participants
were asked to place a checkmark next to each of the 30 constructs (10 personal
constructs, 10 interpersonal constructs, and 10 vocational constructs) indicating whether
they viewed it as being "Not Important," "A Little Important," "Moderately Important,"
or, "Very Important" when they judge individuals of their own sex (see Appendix G) and
when they judge individuals of the opposite sex (see Appendix H). Mean ratings of
importance were used to reflect the relative construct importance of personal,
interpersonal, and vocational constructs in the groups' construct systems
Participants were placed in one of two groups (see Methodology) based on their
pretest scores on three subscales of the Eating Disorders Inventory (Garner et al, 1983).
Participants scoring lowest (n=20) during protesting on the Drive for Thinness, Body
Dissatisfaction, and Bulimia subscales had a mean age of 18.35 (SD= 745; range=18-21),
weight of 124.95 pounds (SD=16 9; range=95-165 lbs.), and were primarily Caucasian
(70%), followed by Hispanic (10%), Other (10%), Black (5%), and Asian (5%). For
clarity purposes, these low-scoring participants were identified as "Low Eating-
Disordered." It is important to note, however, that this label is used descriptively, not
diagnostically, and reflects a group of participants who evidence few characteristics
associated with eating disorders, such as drive for thinness (M=.59, SD=.43), body
dissatisfaction (M= 98, SD=.51), and bulimic features (M=29, SD=. 15), as measured by
the three subscales of the EDI.
Similarly, participants scoring highest (n=20) during protesting on the three EDI
subscales had a mean age of 18.55 (SD=28.95, range=17-21), weight of 139.45 lbs.
(SD=28.95; range= 12-250), and were primarily Caucasian (65%), followed by Other
(25%), Hispanic (5%), and Asian (5%). These participants, identified as "High Eating-
Disordered," should not be interpreted as participants who have been clinically diagnosed
as having an eating disorder. Rather, these high scoring individuals are those reporting
more eating disorder characteristics than the Low Eating-Disordered participants, along
measures of Drive for Thinness (M=3.04, SD=52), Body Dissatisfaction (M=4.29,
SD=.59), and Bulimia (M=l 76, SD= 89) subscales of the EDI. (See Table 4-1 for
relevant means and standard deviations).
Actual Participants' Prescreening Means (and Standard Deviations) on Selected EDI
Subscales and Organized by High and Low Scores
Low Scorers High Scorers
EDI Subscale (n=20) (n=20)
Drive for Thinness 57 (42) 3.04 (.52)
Bulimia .29( 15) 1.76 (.89)
Body Dissatisfaction .80 (.50) 4.29 (.59)
Note that it is difficult to compare these scores with those found in clinical and
subclinical studies due to our use of the 0-5, nontransformed scoring system described in
the methodology chapter. Given our belief that our nontransformed scoring system
provides a more sensitive (i.e., extreme) sample than what might be found using a
transformed (i.e., 0-3) scoring system (see discussion of this in Chapter 3), we believe our
Low Eating Disorder group is comparable to healthy women used in other studies
(Schoemaker et al., 1994), as well as when compared to the women used in the
development of the EDI (Garner et al., 1983). However, results of a clinical study that
did not transform the Bulimia subscale scores suggests that our high scorers on the
Bulimia subscale reflects a "bulimic-like" (Thompson et al., 1987, p. 223) group. Thus,
while it appears our two groups defined as High or Low Eating-Disordered are clearly
distinguishable from each other, the former represents a more mildly subclinical population
than what is typically found in clinical studies that use the EDI.
A series of one-way and two-way ANOVAs were conducted as preliminary tests
to confirm significant differences between the two groups along the measured variables of
interest (Drive for Thinness, Body Dissatisfaction, and Bulimia). These manipulation
checks were followed by a series of ANOVAs and correlations designed to test the
hypotheses of the study
Pretest EDI scores were used to select high and low EDI scoring participants and
these pretest scores determined their assignment to Low or High Eating-Disordered
groups. Selected EDI subscales were administered a second time, at the beginning of the
experimental session itself, in order to confirm the reliable classification of the participants
into high or low eating disorder groups. All of the EDI subscales administered during
protesting were significantly correlated with the same EDI subscales administered during
the experimental session'. Manipulation checks were then performed by testing the group
differences between High and Low Eating-Disordered groups along measures of Drive for
Thinness, Body Dissatisfaction, Bulimia, and Perfectionism.
Specifically, and as predicted, the correlations between the subscales'
preadminstration scores and experimental session scores were r = .82, p<.0001 for the
Drive for Thinness subscale, r = .84, p<.0001 for the Body Dissatisfaction subscale, and r
= .45, p<.01 for the Bulimia subscale.
Drive for thinness As expected, an analysis of the scores for this subscale found a
significant difference between the two groups Specifically, High Eating-Disordered
participants reported a significantly higher mean (M=3.55, SD=.79) on the Drive for
Thinness subscale compared to Low Eating-Disordered participants (M=.74, SD= 54;
F(1,38)=171.88, p<.0001). This confirms the validity of participants' classification along
the Drive for Thinness subscale of the EDI.
Body dissatisfaction. Mean scores revealed a significant difference,
F(1,38)=102.69, p<.0001, between the two groups of women; High Eating-Disordered
participants reported a mean of 3.84 during testing (SD=.88), whereas Low Eating
Disordered participants reported a mean of 1.29 (SD= 70). These (expected) findings
support the validity of the groupings in relation to the Body Dissatisfaction subscale of the
Bulimia. The Bulimia subscale indicated significantly different means for the two
groups of participants. Specifically, High Eating-Disordered women demonstrated a
Bulimia subscale mean of 1.86 (SD=.95), whereas Low Eating-Disordered women
reported a significantly lower mean of.41 (SD=.16), F(1,38)=43.66, p<.0001. Thus, as
predicted, the group composed of individuals highly concerned with weight and body
shape was significantly more likely to be engaging in bulimic-type behaviors than were
individuals in the low eating disorders group.
Perfectionism. The Perfectionism subscale, the final EDI subscale tested, is unlike
the aforementioned EDI subscales because it was not included in protesting (due to
external restrictions regarding number of questions allotted to each researcher for
protesting). It was included in the testing administration along with the Drive for
Thinness, Body Dissatisfaction, and Bulimia subscales, however, because of its role in the
eating disorders literature (see Methodology).
In this study the two groups of women tended to score differently along the
measure of perfectionism, but this tendency failed to reach statistical significance,
F(1,38)=2.98, p<.09 High Eating-Disordered participants reported a mean of 3.28
(SD=.87) for the Perfectionism subscale whereas Low Eating-Disordered participants had
a Perfectionism subscale mean of 2.72 (SD=1.14).
In general, therefore, these manipulation checks supported and clarified the nature
of the two groups. Compared to the Low Eating Disordered group, the High Eating-
Disordered group showed significantly higher drive for thinness, were more dissatisfied
with their bodies, and engaged in more bulimic-type behaviors, and tended toward greater
perfectionism as measured by the EDI's corresponding subscale scores. See Table 4-2 for
relevant means and standard deviations.
High and Low Eating-Disordered Mean Scores (and Standard Deviations) on Selected
EDI Subscale Low Scorers High Scorers p
Drive for Thinness .74 (.54) 3.55 (.79) .0001
Body Dissatisfaction 1.29 (70) 3.84 (.88) .0001
Bulimia .41 (16) 1.86(.95) .0001
Perfectionism 2.72(1.14) 3.28 (.87) .09
Following the analyses concerning the effectiveness of the experimental
manipulations a series of ANOVA's were conducted to test the primary analyses of the
study. These analyses included testing for predicted differences between High and Low
Eating-Disordered groups along measures of cognitive differentiation and integration, as
well as measures of weight-related implications and construct importance.
Cognitive differentiation To test for predicted differences in cognitive
differentiation, Functionally Independent Construct (FIC) scores were first calculated for
the three repertory grids (personal, interpersonal, and vocational; see Appendices C-E; see
Table 4-3) and these FIC scores were used in the 2 (Groups) by 3 (Construct Domain)
mixed factorial ANOVA. Contrary to predictions, results indicated no significant main
effect between the two groups, F(1,38)= 29, p<.59.
Means (and Standard Deviations) for Group and Construct Domain Variables for
Cognitive Differentiation Scores
Eating Disorder Personal Interpersonal Vocational
Low 10.20 (4.63) 995 (3.83) 7.20(3.65)
High 11.10(4.31) 8.10(4.06) 6.60 (3.43)
A significant main effect for Construct Domain was revealed, however,
F(2,76)=13.69, p<.0001. While no predictions were made for direction, the direction of
the effect indicated that constructs in the Personal Domain showed the highest levels of
construct differentiation (M=10.65), followed by constructs in the Interpersonal (M=9.03)
and Vocational (M=6.9) Domains, respectively.
No significant interaction was found between levels of Group and Construct
Domain, F(2,76)=1.84, p<.17. (See Table 4-4).
ANOVA for Group and Construct Domain Variables for Cognitive Differentiation Scores
Source df SS MS F p
Eating Disorder 1 8.01 801 .29 .59
Construct Type 2 282.92 141.46 13.69 .0001
Eating Disorder X 2 37.92 37.92 1.84 .17
Integration scores. This second set of structural scores was also analyzed using a
2 (Group) X 3 (Construct Domain) mixed factorial design. Contrary to predictions,
results of the ANOVA indicated no significant main effect between the two groups,
A significant main effect for Construct Domain was revealed F(2,76)=12.62,
p<.0001. Although no predictions were made for direction, the direction of the effect
indicated that the Vocational Domain showed the highest level of cognitive integration
(M=1372.85), followed by constructs in the Interpersonal Domain (M=1032 28), and
Personal Domain (M=948.75), respectively.
No significant interaction was found between levels of Group and Construct
Domain, F(2,76)=2.16, p< 12. See Table 4-5 for relevant means.
Means (and Standard Deviations) and ANOVA for Group and Construct Domain
Variables for Cognitive Integration Scores
Eating Disorder Personal Interpersonal Vocational
Low 911.55 1167.30 1362.30
(320.13) (283.25) (655.70)
High 986.95 897.25 1383.40
(357.77) (261.15) (602.70)
Source df SS MS F p
Eating Disorder 1 100298.68 100398.68 .38 .54
Construct Type 2 4030952.55 2015476.28 12.62 .0001
Eating Disorder 2 690175.05 345087.52 2.16 .12
Implications of body weight change. The dependent variables (Implications of
Body Weight Change within Personal, Interpersonal, and Vocational domains) measured
here refer to the pen and paper form located in Appendix F Participants were asked to
assume that their current weight had suddenly increased by 20%, and then asked to rate
(0=No Change, l=Little Change, 2=Moderate Change, 3=A Lot of Change) how much
they would expect to change along each of 30 dimensions (10 personal dimensions, 10
interpersonal dimensions, and 10 vocational dimensions) given their new weight.
Three measures were calculated for each participant, reflecting the degree of
anticipated change across the personal dimensions (e.g., cleanliness), along interpersonal
dimensions (e.g., competitiveness), and along vocational dimensions (e.g., education).
A 2 (Groups) by 3 (Construct Domain) mixed factorial ANOVA was applied to
the measure of perceived implications for change (following a 20% increase in body
weight). The first factor was a between subjects factor and referred to level of eating
disorder (High or Low), whereas the second factor was a within subjects factor reflecting
the domain of the construct dimensions (personal, interpersonal, and vocational). Results
revealed a significant main effect between the two groups, F(1,38)=6.41, p<.02 (See Table
4-6). As expected, the High Eating-Disordered Group demonstrated significantly higher
anticipated change (M=13.2), in comparison to the Low Eating-Disordered Group
Means (and Standard Deviations) and ANOVA for Group and Construct Domain
Variables for Anticipated Weight Changes
Eating Disorder Personal Interpersonal Vocational
Low 13.85 (4.88) 8.70(5.59) 5.20(4.19)
High 17.30(431) 13.10(5.85) 9.20 (6.64)
Note: Range = 0-30.
Source df SS MS F p
Eating Disorder 1 468 08 468 08 641 02
Construct Type 2 1409.15 704.58 120.16 0001
Eating Disorder X 2 4.55 2.28 .39 68
A significant main effect was also found for Construct Domain, F(2,76)=120.16,
p<.0001. Although no predictions were made for direction, the direction of the difference
revealed that the greatest levels of expected change occurred along Personal constructs
(M=15.58), followed by constructs in the Interpersonal (M=10.9) and Vocational (M=7.2)
However, no significant interaction between Groups and Construct Domain was
indicated, F(2,76)=.39, p<.68. Means and standard deviations also appear in Table 4-6.
Weighted evaluations for women. The 2 (Groups) by 3 (Construct Domain) mixed
factorial ANOVA was applied to the measure of weighted evaluations for women. This
measure reflected the relative construct importance of personal, interpersonal, and
vocational constructs in the groups' construct systems. Contrary to predictions, no
significant effect between the two groups, F(1,38)=203, p< 16, was indicated by the
results, and relevant means are depicted in Table 4-7.
Mean Ratings of Construct Importance (and Standard Deviations) by Group and
Construct Domain for Evaluations for Women
Eating Disorder Personal Interpersonal Vocational
Low 9.65 (6.98) 19.10(4.49) 10.25 (5.38)
High 12.10 (6.69) 20.32 (4.07) 13.60(6.98)
Source df SS MS F p
Eating Disorder 1 163.97 163.97 2.03 16
Construct Type 2 1806 86 903.43 70.91 .0001
Eating Disorder X 2 24.09 12.04 .95 39
A significant main effect for Construct Domain was revealed, however,
F(2,76)=70.91, p<.0001. Contrary to predictions, participants rated the Interpersonal
Domain as being the most important construct domain when making judgments about
other women (M=19.71), followed by the Vocational Domain (M= 11.97), and Personal
No significant interaction was found between levels of Group and Construct
Domain, F(2,76)=.95, p<.39.
Weighted evaluations for men The 2 (Groups) by 3 (Construct Domain) mixed
factorial ANOVA was also applied to the measure of weighted evaluations for men.
Although no predictions were made, results indicated no significant effect between the
two groups, F(1,38)=3.03, p<.09, and relevant means are depicted in Table 4-8
Mean Ratings of Construct Importance (and Standard Deviations) by Group and
Construct Domain for Evaluations for Men
Eating Disorder Personal Interpersonal Vocational
Low 13.95 (6.97) 21.05 (3 72) 13.32 (6.88)
High 16.85 (5 54) 22.15 (4.03) 17.60 (5.47)
Source df SS MS F p
Eating Disorder 1 206.23 206.23 3.03 .09
Construct Type 2 1002.82 501.41 38.52 .0001
Eating Disorder X 2 54.62 27.31 2.10 13
As expected, a significant main effect for Construct Domain was found, yet not in
the predicted direction, F(2,76)=38.52, p<.0001. Participants rated the Interpersonal
Domain as being the most important construct domain when making judgments about men
(M=21.68), followed by the Personal Domain (M=15.48), and Vocational Domain
No significant interaction between the levels of Groups and Construct Domain was
revealed, F(2,76)= 2.10, p<. 13.
Correlations Between EDI Subscales and Differentiation Scores
To explore further the possible relationship between levels of disordered-eating
and levels of cognitive differentiation, a series of correlations was computed between the
EDI subscales and cognitive differentiation scores. Contrary to predictions, all
correlations failed to reach statistical significance, and these correlations are depicted in
Correlations for EDI Subscales and Personal, Interpersonal, and Vocational
EDI Subscale Personal Interpersonal Vocational
Drive for Thinness 19 -.16 -.09
Bulimia .01 16 -.08
Body Dissatisfaction 05 -.18 -.09
Perfectionism 12 07 -.12
Note: n = 40. None of the tabled correlations reached levels of statistical significance.
Correlations Between EDI Subscales and Integration Scores
A series of correlations was conducted between the EDI subscales and cognitive
integration scores. In contrast with the insignificant correlational findings as related to
differentiation scores, and as predicted, a number of significant relationships appeared
between these subscales and integration scores.
First, the Drive for Thinness subscale, Body Dissatisfaction subscale, and Bulimia
subscale were all significantly related to the integration scores derived from the
interpersonal constructs. Second, both the Perfectionism subscale and Bulimia subscale
were significantly correlated with the integration scores for the personal grid types See
Table 4-10 for details concerning these correlations and their levels of significance.
Correlations for EDI Subscales and Personal, Interpersonal, and Vocational Integration
EDI Subscale Personal Interpersonal Vocational
Drive for Thinness 05 -.39b .05
Bulimia 33a 32a 09
Body Dissatisfaction 19 -.40b 08
Perfectionism .41b .05 29
Note: n= 40
a = Statistically significant at ps .05.
b = Statistically significant at p<.01.
Correlations Between EDI Subscales and Construct Implications Scores
A series of correlations was conducted to identify any possible significant
relationships existing between EDI subscales and construct implication scores. As
expected, a number of such relationships materialized. Specifically, the Drive for Thinness
subscale was significantly correlated across all types of implications (i.e., personal,
interpersonal, and vocational implications) as were the Body Dissatisfaction and Bulimia
subscales. The Perfectionism subscale failed to demonstrate any statistically significant
relationship with any of the implications domain types. Table 4-11 depicts the specific
correlations and their significance levels.
Correlations for EDI Subscales and Personal, Interpersonal, and Vocational Implication
EDI Subscale Personal Interpersonal Vocational
Drive for Thinness .42b .51b .42b
Bulimia 43b .34a .41b
Body Dissatisfaction 45b .50b .43b
Perfectionism .23 .11 .19
Note: n = 40
a = Statistically significant at p< .05.
b = Statistically significant at ps 01.
Correlations Between Participants' Actual Weights and Differentiation Scores
Correlations were conducted between participants' actual weights and their
differentiation scores across all three grid types (i e., personal, interpersonal, and
vocational).2 Although no predictions were made, results indicated that no significant
relationship existed between the participants' actual weights and their differentiation scores
across the different grid types, and Table 4-12 documents these findings.
Correlations for Participants' Actual Weights and Personal, Interpersonal, and Vocational
Personal Interpersonal Vocational
Participants' Actual -10 -.09 -.07
Note: n = 40. None of the tabled correlations reached levels of statistical significance.
2 Note that the High and Low Eating-Disordered groups did not differ significantly in
their actual reported weights (t =-.31, df=38, p=ns).
Correlations Between Participants' Actual Weights and Integration Scores
Likewise, correlations were also conducted between participants' actual weights
and their integration scores. Although no predictions were made, and in contrast to
correlations between actual weight and differentiation scores, there was one significant
relationship existing between participants' actual weights and integration scores
Specifically, as highlighted in Table 4-13, a significant relationship was found between
participants' actual weights and their integration scores on the personal grid form.
Correlations for Participants' Actual Weights and Personal, Interpersonal, and Vocational
Personal Interpersonal Vocational
Participants' Actual .33* -.09 .16
Note: n= 40
Statistically significant at p< 05.
Correlations Between Participants' Actual Weights and Implications Scores
Lastly, a series of correlations was conducted between participants' actual weights
and their implication scores across the three grid types. Again, while no predictions were
made, several significant relationships were found to exist. In fact, participants' actual
weights were significantly correlated with all three grid types, including personal,
interpersonal, and vocational. See Table 4-14 for the specific correlations and their level
of statistical significance.
Correlations for Participants' Actual Weights and Personal, Interpersonal, and Vocational
Personal Interpersonal Vocational
Participants' Actual 42* .46* .67*
Note: n = 40
Statistically significant at p< 01.
This chapter discusses the relevant findings of this study in relation to the literature
reviewed and hypotheses provided in chapter two. The first section reiterates the study's
predictions concerning structural scores (differentiation and integration), implication
scores, and construct importance ratings, and then considers their relationships with the
results. The latter part of this chapter addresses the study's limitations, and also provides
potential directions for future research in the area of personal construct theory as applied
to eating disorders.
Drawing upon previous findings, the author of this current study anticipated that
the High Eating-Disordered group would demonstrate significantly lower levels of
cognitive differentiation compared to the Low Eating-Disordered group, and that this
would generalize across all grid types (i.e., personal, interpersonal, and vocational).
Contrary to these predictions, no main effect between the two groups was found for this
Relatedly, a significant negative correlation was expected to exist between EDI
scores (Garner et al., 1983) and measures of differentiation, but no such relationship was
Differences did occur in the extent to which women differentiated among the three
construct domains, however. While the direction of the main effect was not predicted, a
discussion of this finding deserves attention The direction of this effect indicated that
personal constructs generated the greatest levels of differentiation, followed by
interpersonal and by vocational constructs, respectively In other words, both groups
(High and Low Eating-Disordered) possessed more dimensions in which to evaluate their
personal (i.e., physical) selves than their interpersonal, or vocational selves.
Given the extent to which sociocultural messages emphasize the importance of a
woman's appearance, the direction of this construct effect for differentiation is not
surprising. In short, women have been acculturated to attend to, and distinguish among,
multiple features of their personal appearance (e.g., size. weight, body shape, physical
attractiveness, etc.), and this difference in the levels of differentiation may be a reflection
of this attention.
The results of this study failed to confirm the hypotheses related to the second
dependent variable of interest, integration Again, drawing upon previous literature in this
area, it was believed that the results would replicate a particular research trend that has
linked eating-disordered persons with more highly integrated cognitive systems than
For the most part, this study's predictions were not substantiated. No main effect
between the two groups for integration scores was found; thus, one is unable to assert that
the High Eating-Disordered group significantly differed from the Low Eating-Disordered
group in the tightness or degree of organization exhibited in their cognitive system.
However, a significant relationship between two of the EDI subscales
(Perfectionism and Bulimia) and integration scores for the personal grid forms did exist,
but in the opposite of the predicted direction. In other words, significant negative
relationships were found between the Drive for Thinness, Body Dissatisfaction, and
Bulimia subscales and the integration scores derived from interpersonal constructs.
These findings are puzzling. We might question whether the choice of
interpersonal constructs used effected this outcome, or whether the High and Low Eating-
Disordered groups can be compared with the disordered-eating women and healthy
women samples used in other studies. Regardless, a reasonable interpretation for these
mixed findings is difficult to provide.
Another unexpected yet significant finding related to integration deserves
attention A significant positive correlation between participants' actual weights and their
personal integration scores was found Thus, as the actual weight of the participant
increased, so did the integration score for that participant (n=20). This finding, while
intriguing, may only muddy our already confusing findings as related to integration scores.
It appears to offer some indirect support to the well-documented finding that eating-
disordered women typically possess highly integrative cognitive systems but it is limited in
that it is correlational in nature and was only evident in the personal construct domain.
As with measures of cognitive differentiation, there was a main effect for
Construct Domain along the measure of integration. The direction of this finding
indicated that both groups (High and Low Eating-Disordered) exhibited the greatest level
of integration, or organization, within their vocational construct domain, followed by their
interpersonal construct domain, and lastly, within their personal construct domain.
One explanation for why both groups experienced a greater degree of organization
or interrelatedness among constructs in the vocational domain may be due in part to the
expanding scope of their identities. In other words, young women of today are more
likely to pursue a "vocational" identity in addition to their interpersonal and personal
identities. Thus, the energy that women may be expending in order to develop and
maintain a vocational or professional identity may be contributing to their greater ability to
predict and make decisions in relation to their vocational functioning.
This explanation may further be supported when we recall that our sample
consisted solely of college students. Thus, our participants, by nature of their selection,
are invested in determining and then pursuing a major field of interest that will ultimately
lead to a specific occupation. Thus, the participants in our sample may be intent on
forging a professional identity.
Another sample selection demographic that might assist us in our interpretation of
this construct domain directional finding is that the average age of our participants was
eighteen. Many, if not most, individuals as this developmental age are experiencing
intrapersonal and interpersonal distress In other words, persons at this stage of
development are likely seeking a better sense of self-understanding They are searching
for greater interpersonal (i.e., relationships with peers, significant others, family members)
and intrapersonal (i e., self) awareness. In all, this is the time for examining and revising
interpersonal and personal identities, which may be reflected by the directional finding
(i.e., vocational constructs exhibiting the greatest degree of integration, followed by
constructs in the interpersonal construct domain, and lastly, by constructs in the personal
domain) of this variable of interest.
It was hypothesized that the High Eating-Disordered group would be significantly
more likely to anticipate substantial self-change should their body weight increase
compared to the Low Eating-Disordered group This main effect between the two groups
was substantiated by the results, as was a main effect for construct type Thus, the High
Eating-Disordered group reported that if their weight suddenly increased by 20%, they
would experience a significant degree of change to occur in their physical sense of self
(e.g., attractiveness, complexion, cleanliness), in their interpersonal characteristics (e.g ,
friendliness, confidence, optimism), and in their vocational (e.g., prestige, work
satisfaction, job security) domains of function, respectively.
Relatedly, it was expected that the EDI subscales would be correlated with the
degree of change implied by weight. More specifically, higher EDI scores were predicted
to be positively correlated with the degree of implied changes in personal, interpersonal,
and vocational domains of experience. Several significant relationships resulted. In fact,
all but one of the EDI subscales (Perfectionism) was significantly correlated across all
types of implications (i.e, personal, interpersonal, and vocational). In other words, high
scores on the Drive for Thinness subscale, the Bulimia subscale, and the Body
Dissatisfaction subscale were shown to be positively related to the degree of change
anticipated for personal, interpersonal, and vocational construct domains should the
person suddenly experience weight gain.
While this finding is not causal it is consistent with the notion that eating-
disordered individuals are implicativelyy bound" by their constructions of experience. In
other words, weight change carries higher levels of anticipated change across a variety of
domains for them (High Eating-Disordered group), implying substantial levels of threat
associated with significant body weight change.
The main effects for group and for construct type findings were anticipated, as
were the significantly positive correlations existing between EDI subscales and degree of
anticipated changes in a person's personal, interpersonal, and vocational domain of
experience should they suddenly gain weight. These significant findings were expected
because of the nature of disordered-eating as well as the sociocultural influences on a
woman's personal, interpersonal, and vocational sense of functioning (see chapter two).
These findings may be interpreted such that women overly concerned with weight and
body shape define not only their personal, but interpersonal and vocational selves as
greatly influenced by their current weight Thus, should they experience an increase in
weight, they are also likely to experience a significant degree of change in other domains
of function as well.
The last significant finding to be discussed in relation to this "implications" variable
is correlational in nature. After calculating the correlations for participants' actual weights
and personal, interpersonal, and vocational implications scores, significant positive
correlations were found across all three construct domains. In other words, as
participants' weights increased, they anticipated greater degrees of change in their personal
sense of self(i.e., personal construct system), interpersonal sense of self (i.e., interpersonal
construct system), and vocational sense of self (i.e., vocational construct system)
One interpretation of this finding is socioculturally based. It may be that women of
all sizes are cognizant of the weight and body-shape demands placed on them. In fact, in
light of our correlational finding, it may be that those who are farthest away from the
perceived ideal weight (i.e., current self weight differs greatly from ideal self weight) are
more sensitive to the effects that weight may have on a woman's personal, interpersonal,
and vocational areas of function. Thus, the greater their weight already, the more likely
they are to also perceive the potential degree of change in many areas of experience
should their weight increase.
Construct Importance Ratines
Two predictions were made regarding this dependent variable. First, it was
predicted that the High Eating-Disordered group would demonstrate significantly greater
levels of construct importance when evaluating other women compared to the Low
Eating-Disordered group. In other words, it was believed that the former group (because
of their perceived more sensitivity to the number of domains in which a woman can be
judged) would regard personal, interpersonal, and vocational constructs as more
meaningful to their evaluation of other women than would the latter group It was
predicted that the direction of this finding would be in the order of personal, interpersonal,
and vocational, reflecting the degree to which the High Eating-Disordered group was
expected to identify these constructs as critical domains when appraising other women.
Second, it was hypothesized that both groups (High and Low Eating-Disordered)
would evaluate men in the order of their vocational, interpersonal, and personal construct
domains of being as measured by the groups' construct importance ratings of these same
The results of this study were mixed in their support of its hypotheses. First, while
higher means were associated with the Higher Eating-Disordered group for all three
domains (i.e., personal, interpersonal, and vocational), they failed to be significantly higher
than the Low Eating-Disordered group's mean scores. So, contrary to predictions, no
main effect between the two groups was found for evaluating other women. In other
words, High and Low Eating-Disordered groups did not significantly differ in how they
regard the importance of personal, interpersonal, and vocational domains of experience
when evaluating women.
There was a main effect for Construct Domain, yet the direction was different than
expected. Specifically, results indicated that both groups rated the interpersonal domain
as being the most critical in their evaluation of other women, followed by the vocational
domain, and lastly, by the personal domain. This finding could be explained by women's
acculturation to developing and maintaining interpersonal relationships.
However, another interpretation for this finding may lie in examining the opposite
end of the directional finding, the significance of the personal domain being evaluated as
the least important construct domain in their evaluation of other women. Thus, it may be
that women seek to combat the personal (i.e., importance of appearance) demands that
women are subject to by failing to hold other women to them. In other words, it may be
that the groups of women seek to replace the current physically and emotionally taxing
societal messages that target women with more healthy ones; to view interpersonal
functioning as the most important domain in which to be evaluated, vocational functioning
as the second most important domain, and last, personal (i e., physical sense of self)
characteristics as the last area in which to evaluate women.
In regard to the second hypothesis, and as predicted, no main effect was found
between the two groups in how they view the three construct types used when evaluating
men [Note that the High Eating-Disordered group tended to demonstrate higher
construct importance means across all three construct types (i.e., personal, interpersonal,
and vocational), yet these higher scores failed to reach significance levels when compared
to the Low Eating-Disordered group's scores]. Thus, the groups did not significantly
differ in their weighted construct evaluations of men.
Also, while a main effect for Construct Domain was found as predicted, the
direction of this main effect was contrary to expectations. We anticipated that both
groups (High and Low Eating-Disordered) would regard vocational constructs as most
important in their evaluations of men, interpersonal the second most critical area, and
personal constructs the least important domain when judging men. Instead, results
indicated that the groups considered the order of construct importance when evaluating
men to be: (1) interpersonal, (2) personal, and (3) vocational.
An interpretation for this finding may be similar to the one offered above regarding
the main effect for construct domain when women evaluate other women. Specifically, it
may be that women want both men and women to be primarily evaluated by others on
interpersonal constructs. This speaks to a possible desire that acceptance of another
should be based more on a person's core personality characteristics, and less on their
personal (i.e., physical) and vocational selves.
Furthermore, if we recall the high levels of differentiation that both groups
exhibited on the personal grid form it may be more clear as to why all participants
regarded interpersonal constructs as being the most important construct domain when
judging another: The higher differentiation scores for the personal grid type likely reflects
both groups' recognition that there are a multitude of dimensions in which a personal self
is judged (e.g., gracefulness, weight, height). And, our construct importance findings
could be interpreted as an indication that the High and Low Eating-Disordered groups
prefer judgments about men and women to be based on interpersonal factors (e.g.,
friendliness, confidence, intelligence) foremost, and on vocational factors or personal (i.e.,
physical) factors second.
Limitations of the Study
The results of this study need to be interpreted within the contexts of its
limitations. There are several limitations to this study. First, our sample size was small
(n=40). Thus, future work in this area may benefit from a larger sample size which allows
analyses to be performed with greater statistical power
Second, the participants were primarily Caucasian and college-aged thereby
limiting the generalizability of our results to other populations. Relatedly, due to
prescreening constraints and time concerns for the testing administration, only a few of the
eight EDI original subscales (Garner et al., 1983) were used to determine "High" and
"Low Eating-Disordered" groups. Thus, while the three subscales used for prescreening
purposes (Body Dissatisfaction, Drive for Thinness, and Bulimia) and the four subscales
used in the testing administrations (Body Dissatisfaction, Drive for Thinness, Bulimia, and
Perfectionism) have been used for screening purposes in other studies, it must be
recognized that the factors involved in eating disorders and its symptomatology are
multifaceted. If we had included the other EDI subscales (Interoceptive Awareness,
Interpersonal Distrust, Ineffectiveness, Maturity Fears) in both the screening as well as
administrative testing we might have had a more clearly identifiable (i.e., clinical,
subclinical, or moderately subclinical) sample.
Thus, it is important to remember that the group of women identified as "High
Eating-Disordered" were individuals exhibiting symptomatology representative of
disordered-eating, not participants necessarily deserving a clinical diagnosis of an eating
disorder Consequently, we can only assume that they have some attitudes and behaviors
similar to someone with an eating disorder. We do not know the severity of their eating
behavior disturbance. And, while the cognitive structure literature as applied to eating
disorders suggests that eating-disordered individuals are more likely to exhibit lower levels
of differentiation and greater levels of integration or a "tightness" in their decision-making
cognitive style (Butow, Beumont, & Touyz, 1993; Button, 1983; Coish, 1990; Heesacker
& Neimeyer, 1990; Neimeyer & Khouzam, 1985), this finding may not hold true for
individuals such as those included in our study. In fact, another study using subclinical
populations also failed to replicate past cognitive structure findings as related to eating
disorders (Munden, 1982).
Another reason we are limited in our ability to characterize the two groups
according to eating disorders terminology is because of the scoring method we used for
our EDI subscales. Unlike many previous studies, we did not transform the participant's
responses due in part to theoretical reasons (see chapter 3, methodology). Thus, while we
believe we made the instrument more sensitive in distinguishing between more and less
problematic eating behaviors and body shape attitude responses, we are unable to compare
the nature of the groups because there are not enough studies that do not transform their
EDI subscale scores in which to do so.
Another limitation of this study is that the dependent variables were restricted to
pen and paper measures Thus, future research may seek to expand upon this measure
format by including other types of measures, including forms of behavioral assessment.
We also believe that our grid size (a 10 by 10 element grid) may not have been
powerful enough to detect potentially significant cognitive score differences between the
two groups. This may particularly be true because the differences were in the predicted
direction, yet failed to reach significantly different levels. We believe a 12 X 12 grid may
be more beneficial for future related studies.
Lastly, another potential limitation of this study may be that the repertory grid
element choice as well as construct choice for the three domains of interest (i.e., personal,
interpersonal, and vocational) may have contributed to our failure to replicate past
cognitive structural score literature findings in relation to personal construct theory as
applied to eating disorders research. While some of the elements and constructs used in
this present study had been used in other related areas of research, many were not due
much in part to the study's originality Thus, it would be beneficial to more carefully
determine which elements and constructs should be used in light of the areas of concern
(e.g., personal, interpersonal, and vocational domains of function).
While this study sought to replicate some previous findings concerning eating
disorders and personal construct theory, it also sought to determine the extent to which
disordered-eating may interfere with other areas of functioning (i.e., personal,
interpersonal, and vocational) Future research may choose to clarify the results of our
study and may do so by altering our study's design
For example, longitudinal research in this area may provide the most useful
examination of the implications that perceived and actual weight gain have on a person's
personal, interpersonal, and vocational sense of experience. This type of research design
would allow one to explore whether or not structural scores change over time (i.e., the
extent of their durability), and whether or not they continue to be related to a participant's
Also, one might choose to include men in the testing sample. While it is difficult to
obtain a sufficient number of men who exhibit disordered-eating for research such as this,
the findings of such a study (in which the structural and implication scores for men and
women were compared) would yield much information. Specifically, one might learn
more of the relationships between gender role endorsement, disordered-eating, and
cognitive structural scores should such measures be included in a research design.
In all, there are several potential directions for future research in this area which
reflects the need for further exploration of the cognitive structures associated with clinical,
subclinical, and moderately subclinical eating-disordered populations. It is our hope that
this current study provides some impetus for additional work that applies personal
construct theory to the area of eating disorders.
The results of this study provided partial support for the primary hypothesis that
the cognitions associated with higher eating-disordered women differ from those of
women who exhibit fewer eating-disordered characteristics. Specifically, and consistent
with predictions, the group of women identified as "High Eating-Disordered" anticipated
that they would experience greater degrees of change in relation to their personal,
interpersonal, and vocational style if they were to gain twenty percent of their current
weight, compared to the "Low Eating-Disordered" group. However, contrary to
predictions, the greatest number of anticipated changes for both High and Low Eating-
Disordered groups was associated with personal constructs, followed by interpersonal and
vocational constructs, respectively.
In addition, contrary to our expectations, the High Eating-Disordered group did
not produce significantly higher integration scores for the three grid types than the Low
Eating-Disordered group, nor did the Low Eating-Disordered group yield significantly
higher differentiation scores. These predictions were based on previous work
documenting these findings within the interpersonal domain of construing.
Other significant main effects were found in this study, including construct type for
all five dependent variables, including: differentiation, integration, implications, weighted
evaluations (i.e., construct importance) for women, and weighted evaluations (i.e.,
construct importance) for men
First, the differentiation and implication variables demonstrated highest means for
constructs in the Personal Domain, followed by constructs in the Interpersonal Domain,
and lastly, by the Vocational Domain. The highest integration scores were also associated
with constructs in the Personal Domain, however, constructs in the Vocational Domain
reported higher means than constructs in the Interpersonal Domain for this variable.
Lastly, the main effect for Construct Domain for construct importance when
evaluating other women was in the direction of highest means being associated with
constructs in the Interpersonal Domain, followed by the Vocational Domain, and Personal
Domain. Similarly, our construct importance dependent variable when evaluating men
reported highest means for constructs in the Interpersonal Domain, followed by the
Personal Domain, and Vocational Domain.
In summary, this study sought to contribute to the personal construct literature as
applied to eating disorders as it anticipated finding significant differences between the two
groups of interest in terms of their cognitive structure. Currently, it appears that this
research is the first of its kind in that it explores the extent to which women who are
overly concerned about their weight, body size and shape are more likely to anticipate
substantial global changes should their body weight increase, and it uniquely examines
how these anticipations compare to those of women less concerned with their physical
selves and eating habits. It remains for future research to replicate the "domino effect" of
anticipated change across various domains of experience, and to continue to further
explore its relationships with disordered-eating.
SELECTED EATING DISORDERS INVENTORY (EDI) SUBSCALES
This is a scale that measures a variety of attitudes, feelings, and behaviors. Some
of the items relate to food and eating. Others ask you about yourself THERE ARE NO
RIGHT OR WRONG ANSWERS, SO TRY VERY HARD TO BE COMPLETELY
HONEST IN YOUR ANSWERS. RESULTS ARE COMPLETELY CONFIDENTIAL.
Read each questions and place an (x) under the column that applies best for you. Please
answer each question very carefully. Thank you
1. I eat sweets and carbohydrates without feeling nervous
2. I eat when I am upset.
3. Only outstanding performance is good enough in my family.
4. 1 stuff myself with food.
5. I think about dietmg.
6. I think that my stomach is too big.
7. As a child. I tned very hard to avoid disappointing my
parents and teachers.
8. 1 feel extremely guilty after overeating.
9 I am terrified of gaining weight.
10. I think that my thighs are too large.
11. I hate being less than best at things.
12. I exaggerate or magnify the importance of weight,
13. I think that my stomach is just the right size.
14. I have gone on eating binges where I have felt that I could
15. 1 am preoccupied with the desire to be thinner.
16. I feel satisfied with the shape of my body.
17. I think about bingeing (overeating).
18 My parents have expected excellence of me.
19. I like the shape of my buttocks.
20. I think my hips are too big.
21. 1 eat moderately m front of others and stuff myselfwhen
22. I feel that I must do things perfectly or not do them at all.
23. 1 think that my thighs are just the right size.
24. If I gain a pound, I worry that I will keep gaining.
25. I think my buttocks are too large.
26. I have had the thought of trying to vomit in order to lose
27. I think that my hips are just the right size.
28. I have extremely high goals.
29 I eat or drink in secrecy.
You will be participating in a study that is designed to look at the relationship
between cognitive structure and other behaviors (including occupation, appearance, and
interpersonal). You will be asked to complete seven pen and paper forms. One is a scale
which asks you questions about a variety of attitudes, feelings, and behaviors. Three other
forms ask you to independently rate various people you know on the basis of bipolar
vocational dimensions (e.g., high vs. low salary), appearance-oriented dimensions (e.g.,
attractive vs. unattractive) and interpersonal dimensions (e.g. introverted vs. extroverted).
Lastly, three other forms will ask you to evaluate the relative importance of these various
dimensions in making judgments about yourself and others
Your responses are anonymous and you are asked not to put your name or other
identifying information on any of the forms to protect your privacy. You do not have to
answer any questions you do not wish to answer and you may decline to participate in or
to withdraw from the research at anytime. Participation will take approximately 50
minutes and you will receive 2 experimental participation credits for compensation of your
time. Are there any questions? If you want to know more about the research study or if
you have questions later, you may call Lori Russ-Eisenschenk, M.S. at 336-8153, or Dr.
Greg Neimeyer, 392-0264, in the Psychology Department. You are reminded that this is a
confidential study and are asked not to discuss the study with anyone. Do you still wish to
participate? If so, let us begin the study.
REPERTORY GRID--PERSONAL CONSTRUCTS
S1 POSITIVE SIDE NEGATIVE SIDE
S < < __< < +3 +2 +1 0 -1 -2 -3
Photogenic Not Photogenic
Good Complexion Poor Co iplexion
REPERTORY GRID--INTERPERSONAL CONSTRUCTS
r ) 0)
< < < < +3 2 +1 0 -1 -2 -3
Stubborn Not Stubborn
Angry Not Angry
Energetic Not Energetic
Intelligent Not Intelligent
REPERTORY GRID--VOCATIONAL CONSTRUCTS
0 | | | |POSITIVE SIDE NEGATIVE SIDE
<_ <__ < < +3 +2 +1 0 -2 -3
Much Education Little Education
High Income Low Income
High Prestige Low Prestige
Influences People Doesn't Influence People
Creative Not Creative
Helps People Doesn't Help People
Much in Demand Seldom in Demand
Interesting Work Dull Work
Works with Thoughts Works with Hands
Offers Much Security Offers Little Serurity
MODIFIED CONSTRUCT IMPLICATIONS GRID
How much do you currently weigh9 Imagine that you were suddenly to weigh 20%
more than that. In other words, you woke up to find yourself weighing pounds. Look
at the chart below to find out how much you would weigh and fill in that number. Then
consider each of the adjective descriptions. If you were to weigh pounds, how much
would you expect to change along each of these dimensions? Indicate your best
assessment by placing a checkmark next to each adjective indicating how much you would
expect to change.
You can use this chart to determine what your imaginary weight would be:
Current weight (Imaginary Weight = Current Weight + 20% More Weight)
For example, if current weight is 85 lbs., then your imaginary weight is 102 lbs.
85 lbs. = 102 120 Ibs
90 = 108 125
95 =114 130
100 = 120 135
105 = 126 140
110 = 132 145
115 =138 150
= 186 190 lbs.
= 192 195
= 198 200
= 222 220
=228 225 lbs. = 255
=228 230 = 276
=240 235 =282
=246 240 =288
= 252 245 = 294
= 258 250 = 300
=264 255 =306
Moderate A Lot of
No Change Little Change Change Change
Moderate A Lot of
No Change Little Change Change Change
Perform Work with Hands
CONSTRUCT IMPORTANCE FOR EVALUATING WOMEN
You will now be asked to rate the following dimensions. Specifically, we want you to
consider how important each of the descriptions are when YOU judge individuals of your
OWN SEX. Indicate your best assessment by placing a checkmark next to each adjective
indicating whether you view it as being not important, a little important, moderately
important, or every important.
Not A Little Moderately Very
Important Important Important Important
Perform Work with Hands
CONSTRUCT IMPORTANCE FOR EVALUATING MEN
You will now be asked to rate the following dimensions. Specially, we want you to
consider how important each of the descriptions are when YOU judge individuals of THE
OPPOSITE SEX. Indicate your best assessment by placing a checkmark next to each
adjective indicating when you view it as being not important, a little important, moderately
important, or very important.
Not A Little Moderately Very
Important Important Important Important
Perform Work with Hands
American Psychiatric Association. (1952). Diagnostic and statistical manual of
mental disorders. Washington, DC: Author.
American Psychiatric Association. (1968). Diagnostic and statistical manual of
mental disorders (2nd ed ). Washington, DC: Author.
American Psychiatric Association. (1980). Diagnostic and statistical manual of
mental disorders (3rd ed.) Washington, DC. Author.
American Psychiatric Association. (1987). Diagnostic and statistical manual of
mental disorders (3rd ed. revised). Washington, DC. Author.
American Psychiatric Association. (1994). Diagnostic and statistical manual of
mental disorders (4th ed.). Washington, DC: Author.
Andersen, A. E., & DiDomenico, L. (1992). Diet vs shape content of popular
male and female magazines: A dose-response relationship to the incidence of eating
disorders? International Journal of Eating Disorders, 11(3), 283-287
Andersen, A E., Woodward, P J., Spalder, A., Koss, M (1993). Body size and
shape characteristics of personal ("in search of') ads. International Journal of Eating
Disorders, 14(1), 111-116.
Basow, S., & Kobrynowicz, D. (1993). What is she eating9 The effects of meal
size on impressions of a female eater. Sex Roles, 28(5/6). 335-344.
Beebe, D. W., Hombeck, G. N, Schober, A., Lane, M., & Rosa, K. (1996). Is
body focus restricted to self-evaluation? Body focus in evaluation of self and others.
International Journal of Eating Disorders, 20(4). 415-422.
Beumont, P (1995) The clinical presentation of anorexia and bulimia nervosa. In
K D. Brownell and C. G Fairburn (Eds.), Eating disorders and obesity: A comprehensive
handbook (pp. 151-158). New York: Guilford Press.
Bodden, J. (1970). Cognitive complexity and appropriate vocational choice.
Journal of Counseling Psychology. 19, 364-368.
Broverman, I. K., Broverman, D. M., Clarkson, F. E., Rosenkrantz, P. S., &
Vogel, S. R. (1970). Sex-role stereotypes and clinical judgments of mental health. Journal
of Counseling and Clinical Psychology. 34, 1-7
Brownell, K. D., & Fairburn, C. G. (1995). Eating disorders and obesity: A
comprehensive handbook. New York: Guilford Publications.
Bruch, H. (1962). Perceptual and conceptual disturbances in anorexia nervosa.
Psychosomatic Medicine. 24. 187-194.
Bruch, H. (1973). Eating disorders. New York: Basic Books.
Bruch, H. (1978). The golden cage: The enigma of anorexia nervosa. London:
Routledge & Kegan Paul.
Bruch, H. (1988). Conversations with anorexics. New York: Basic Books.
Brumberg, J. J. (1988). Fasting girls: The emergence of anorexia nervosa as a
modern disease. Cambridge, MA: Harvard University Press.
Bushnell, D. E. (1995). "Nagging" questions: Feminist ethics in everyday life.
Lanham, MD: Rowman & Littlefield Publishers.
Butow, P., Beumont, P., & Touyz, S. (1993). Cognitive processes in dieting
disorders. International Journal of Eating Disorders, 14(3). 319-329.
Button, E. (1983). Construing the anorexic. In J. R. Adams-Webber, & J. C.
Mancuso (Eds.), Applications of personal construct theory (pp. 305-316) Toronto:
Button, E. (1993). Eating disorders. Personal construct therapy and change.
Chichester, NY: Wiley.
Button, E. (1996). A prospective study of self-esteem in the prediction of eating
problems in adolescent school girls: Questionnaire findings. British Journal of Clinical
Psychology. 35(2). 193-203.
Clark, C. H, Hershgold, E. J., & Rigdon, M. A. (1984). Personal construct
relationships and therapeutic change Predicting weight loss. Unpublished manuscript.
Cochran, L. R. (1977). Differences between supplied and elicited constructs:
Consideration in career evaluation. Social Behavior and Personality. 5, 241-247.
Coish, B. J. (1990). A personal construct theory of bulimia. Unpublished doctoral
dissertation, La Trobe University, Australia.
Cooper, Z. (1995). The development and maintenance of eating disorders. In K.D.
Brownell & C. G. Fairbum (Eds.), Eating disorders and obesity A comprehensive
handbook (pp. 199-206). New York: Guilford Press.
Crisp, A. H. (1967). The possible significance of some behavioral correlates of
weight and carbohydrates intake. Journal of Psychosomatic Research. 11. 117-131.
Crisp, A. H., & Fransella, F. (1972). Conceptual changes during recovery from
anorexia nervosa. British Journal of Medical Psycholoye. 45. 395-405.
Downs, D. L. (1997). An application of the sociometer model to women with and
without eating disordered behavior. Unpublished doctoral dissertation, The Ohio State
Drewnowski, A., Yee, D. K., Kurth, C. L., & Krahn, D. D. (1994). Eating
pathology and DSM-IIIR bulimia nervosa: A continuum of behavior. American Journal of
Psychiatry 151, 1217-1219.
Fairburn, C. G. (1981) A cognitive behavioral approach to the treatment of
bulimia. Psychological Medicine 11, 707-711.
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EZ76GMX1B_1BPTSR INGEST_TIME 2013-03-25T12:38:51Z PACKAGE AA00013586_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC