|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
EXAMINING RECRUITMENT METHODS FOR COLLEGE WOMEN AT-RISK FOR
EATING DISORDERS AND THE RELATIONSHIP BETWEEN EXERCISE AND
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
This thesis is dedicated to all who have helped me succeed in my academic career.
I thank my committee for their direction, wisdom, and help with this thesis and
guiding me through this master's program. All of our interactions during these past three
years have made an impact on me as a student, a researcher, and a person. All of them are
excellent role models both professionally and personally. Dr. Heather Hausenblas, my
committee chair, has been especially important in providing assistance with my academic
work and research, as well as offering timely words of encouragement that are so very
much needed and appreciated.
I also thank everyone at ProChange Behaviors Systems, Inc., for providing me with
the foundation to pursue my career goals. The opportunities and experiences that I have
gained while working at ProChange have been invaluable to my development as a
researcher. Most of all, I thank them for believing in my ability to succeed.
Finally, I thank all of my friends and family that have guided me throughout my
education. I thank them all for their love and support that have enabled me to pursue my
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iv
LIST OF TABLES ............. ........... ...... ........ .......... vii
LIST OF FIGURES ........................ ........................ viii
ABSTRACT .............. .......................................... ix
1 IN TR O D U C TIO N ......................................................................... .... .. ........
2 LITER A TU R E R EV IEW ............................................................... ...................... 7
G en eral O v erv iew ............ .................... .............................................. ....... ....... .. ..
E tiology .............................................................................. ...............11
Stage O ne ...................................................................................... ..... .. .......11
S ta g e T w o ......................................................................................12
S ta g e T h re e .....................................................................................12
P re v a le n c e .............................................................................................................. 1 4
P re v e n tio n .............................................................................................................. 1 7
R ecru itm ent ............................................................................... 2 0
Exercise and Eating Disorders ................................. .......................... ....24
3 M E T H O D S ............................................................................................................ 2 8
P a rtic ip a n ts ............................................................................................................ 2 8
M e a su re s ................................................................................................................ 2 8
Demographic Questionnaire ................................................... 28
D rive for Thinness Subscale................................ ................... 28
Exercise D ependence Scale .......................................... ............... 29
Body composition....................................... 29
Leisure-tim e Exercise Questionnaire ........................................ ............... 29
P ro c e d u re ................................................................ 3 0
Prim ary Purpose................................................. 32
E eligibility C riteria.................................................. 32
D ata A n a ly sis .................................................................................................. 3 2
Secondary Purpose................................................... 33
4 R E SU L T S ....................................................... 3 5
R espon se R ate.......................................................35
Prim ary Purpose...................... ......... .. ..... ..... .. ............35
S econ dary P u rp o se........... ...... ........................................ .................. .......... ....... 37
5 D ISC U S SIO N ............................................................................... 40
O v e rv iew .........................................................................................4 0
Stu dy F in din g s ................................................................................ 4 0
R e c ru itm e n t ................................................................................................... 4 0
M e d iato r E ffe ct ....................................................................................................4 3
Lim stations .................................. .......................... ..... ..... ........ 46
F utu re D direction s ................................................................4 9
C o n c lu sio n s ........................................................................................................... 5 1
A DEMOGRAPHIC QUESTIONS .............................................. ..................52
B DRIVE FOR THINNESS QUESTIONNAIRE ........................................ ...53
C EXERCISE DEPENDENCE QUESTIONNAIRE ................................ ............. 54
D LEISURE TIME EXERCISE QUESTIONNAIRE ........................55
E STUDY ANNOUNCEMENT ................................................56
F REPLY WITH PIN AND INFORMED CONSENT ................................................57
G E L IG IB L E R E SP O N SE ....................................................................................... 59
H INELIGIBLE RESPONSE ................ ......... ...... ........60
LIST OF REFERENCES ............ .. .......... ................................................ 61
BIOGRAPHICAL SKETCH ................................ ........................................68
LIST OF TABLES
2-1 One-Year-Period Prevalence Rates per 100,000 Young Females............................16
3-1 T hesis T im line. ........................................................................31
4-1 M eans, Standard Deviations, and Alpha Values ............................... ............... .36
4-2 Frequency of Recruitm ent M ethod.................................... ........................... ......... 36
LIST OF FIGURES
2-1 Etiological Model of Eating Disorder Development and Maintenance ...................14
3-1 Study Flow ........................................................................ ... ......... ..... 31
3-2 Relationship of M editor V ariables .................................... ............. ...................34
3-3 Relationship of M oderator Variables .............................................. ............... 34
5-1 M operator M odel ............... ................. ............ ............ ............ 44
5-2 M editor M odel .................................. ... ........ .... .......... 44
5-3 C onceptual M odel of M edition ..................................................................... .. .....45
5-4 Conceptual Model of Mediation with Study Variables.............................................45
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
EXAMINING RECRUITMENT METHODS FOR COLLEGE WOMEN AT-RISK FOR
EATING DISORDERS AND THE RELATIONSHIP BETWEEN EXERCISE AND
Chair: Heather Hausenblas
Major Department: Applied Physiology and Kinesiology
Between 10-30% of college age women are at-risk for eating disorders. This at-risk
population will also be experiencing some of the negative consequences associated with
eating disorders. Prevention interventions are therefore needed to halt development on to
full blown eating pathology. Despite the fact that prevention interventions targeting at-
risk populations, such as college women, are more effective in reducing eating disorder
symptoms than universal program, little research has examined the best methods to
recruit at-risk individuals. The primary purpose of my thesis was to examine the
efficaciousness of recruitment methods for individuals at-risk for developing bulimia
nervosa. Specifically, I examined the following two recruitment pathways: college
campus announcements and announcements made via alternative delivery systems. I
hypothesized that recruitment through the alternative methods, such as the Internet and
email, would produce a larger number of both total respondents and eligible respondents.
Recruiting at-risk individuals allowed for the examination of behaviors believed to
aid the development of eating disorders. Exercise has long been observed in this
population and thought to exacerbate the progression and maintenance of eating
disorders. However, little attention has been given to examining the psychological
motivation for exercise in this population, thus allowing the possibility of a separate
variable mediating or moderating the relationship of exercise and eating disorders. The
secondary purpose of my thesis was to examine the associations among exercise
behavior, drive for thinness scores, and exercise dependence scores. Specifically, I
examined the relationship of exercise behavior and exercise dependence scores on drive
for thinness scores. I hypothesized that drive for thinness scores would be more strongly
influenced by exercise dependence scores than by total amount of exercise alone.
The results of my study indicated that the hypothesis of my primary purpose was
not statistically supported. The results were approaching significance and may have been
improved with a larger sample size. The data also indicated that alternative methods
cover a substantial breadth of potential participants. The results also indicated that the
hypothesis of my secondary purpose was fully supported. These results indicate that the
effect of exercise on eating disorder symptoms is mediated by exercise dependence.
Future research can build upon these results by continuing to use the advantages of
new and evolving alternative technologies in the recruitment of difficult to reach
populations. Through this, future research may also expand on the results of my
secondary purpose by recruiting larger and more diverse samples. Because of the
negative physical and psychological consequences of eating disorder, prevention efforts
are a high research priority.
The most common psychiatric disorders afflicting young women are eating
disorders (i.e., anorexia nervosa, bulimia nervosa, & eating disorder not otherwise
specified; Pritts & Susman, 2003). They are categorized by a persistent and distorted
view of ones own body and subsequent behaviors performed in an attempt to relieve the
psychological turmoil resulting from this faulty self-perception. Although the morbidity
rate of eating disorders in college age women is low (1 3%; Franko et al., 2005), the
mortality rate is high (20% 40%; Harris & Barraclough, 1994, 1998; Keel et al., 2003;
Mehlenbeck, 2002). More alarming is that 10-30% of college age women are at risk for
eating disorders (Cassell, 1994; Franko et al., 2005). Furthermore, serious consequences
of the behaviors associated with eating disorders may not be limited to only those with
the full-syndrome disorders. Some consequences contribute to impaired health even when
they are exclusive of a full-syndrome eating disorder (Pearson, Goldklang, & Striegel-
Moore, 2002). In other words, being at risk for these disorders may not preclude a person
from experiencing their negative health impact. Treating these individuals is also difficult
due to their secretive nature, which allows for many cases to not be recognized
immediately (Fairburn, 1995; Fairburn & Cooper, 1982; Hoek, 1995; Hoek & van
The large number of individuals at risk for eating disorders, and the serious health
consequences experienced throughout the entire spectrum of development, offers an
opportunity and necessity to apply prevention interventions. Eating disorder prevention
has recently made great strides. While early programs tended to be didactic and
psychoeducational in nature, and generally did not produce much effectiveness, the latest
prevention programs are interactive and have shown promising results (Stice & Shaw,
2004). Overall, the most effective prevention programs are delivered to women over the
age of 15 and are interactive, multisession programs not presented as eating disorder
prevention (Stice & Shaw, 2004). Following the traditional modes of universal education
programs as prevention are not effective when applied to this health problem. Perhaps
effects are lost by not selecting those individuals whom are at-risk and therefore the most
in need. In other words, a better understanding and targeting of at risk populations may
be a logical next step in the progression of eating disorder prevention (Stice & Shaw,
2004). Examining recruitment methods for prevention efforts offers an opportunity to
increase the impact of such interventions by providing the knowledge of how to
efficaciously select those most in need.
Recruitment is an immediate concern when preparing to implement an eating
disorder prevention program with women. Ironically, this should not be much of a
challenge considering the previously mentioned large numbers at-risk for eating
disorders. For example, normal to overweight college age women are all at a higher risk
for the development of eating disorder than men (Jacobi et al., 2004; Woodside et al.,
2001). Recruitment efforts need to focus on finding new and more effective ways of
reaching these at-risk people in need of intervention. Examining how efficaciously a
recruitment strategy can contact potential participants can be guided by applying certain
factors of the RE-AIM framework. The RE-AIM framework is a framework that
addresses issues of internal and external validity of health behavior programs (Glasgow,
Vogt, & Boles, 1999; Glasgow et al., 2004). For example, an effective recruitment
strategy must be able to sample large numbers of potentially eligible and willing
participants, while still being implemented in a standard and consistent fashion. Such a
strategy would demonstrate the RE-AIM dimensions of Reach, Effectiveness, Adoption,
and Implementation (Glasgow, Vogt, & Boles, 1999; Glasgow et al., 2004). Specifically,
sampling a large number of potentially eligible participants is a recruitment strategy's
reach. Examining the number of eligible and willing participants a strategy can deliver
represents its effectiveness. A strategy's adoption would be evident by its ability to be
consistently delivered through a variety of settings. Similarly, consistently delivering a
recruitment strategy in each setting is the implementation dimension of the RE-AIM
framework (Glasgow, Vogt, & Boles, 1999; Glasgow et al., 2004).
A review of the literature reveals limited research addressing efficacious methods
of recruiting this population (Lovato et al., 1997; McDermott et al., 2003). New
technologies, such as computers and the internet, enable researchers and practitioners to
reach and recruit populations that were previously difficult to help (Kraut et al., 2004),
and offer both the researcher and participant numerous advantages; such as cost-
effectiveness, accessibility, portability, ease of use, accessibility, and convenience (Kraut
et al., 2004; Myers et al., 2004). Using these advantages may be one way to increase the
effectiveness of recruitment efforts of at-risk individuals.
Both risk factor and prevention research have made substantial advances in their
understanding and effectiveness of intervening on this difficult population. Synthesizing
relevant information from both lines of research with new alternative technologies
clarifies an opportunity to examine new methods of isolating and reaching at-risk
populations. Given this collective understanding, identifying the most efficacious
methods of recruiting this population is the next logical step in the progression of better
Thus, the primary purpose of my thesis was to examine the efficaciousness of
recruitment methods for individuals at risk for developing bulimia nervosa. Specifically, I
examined two recruitment pathways: college campus announcements (in classroom
settings and advertisements posted on a college campus) and announcements made via
alternative delivery systems (placed on an internet website and a mass email sent to
college students). The primary outcome measures of this purpose were the total number
of responses to each advertisement and the total number of respondents who met criteria
indicating being at risk for the development of bulimia nervosa. I hypothesized that
recruitment through the alternative methods, such as the internet and email, would
produce both a larger number of total respondents and a larger number of eligible
respondents (Kraut et al., 2004; McDermott et al., 2003; Myers et al., 2004).
Examining recruitment methods also provides an opportunity to collect data for
other purposes because responding participants will need to complete assessments once
they have inquired about a program. Adding measures of self-reported exercise and
exercise dependence to risk screening assessments allows the possibility of questioning
the role of exercise and its relationship to weight control in terms of the development of
eating pathology. The view traditionally held is that exercise may exacerbate the effects
of eating pathology based on the observation that eating disordered individuals also
excessively exercising (Brewerton, Stellefson, Hibbs, Hodges, & Cochrane, 1995; Davis,
Katzman, Kaptein et al., 1997; Hoglund, & Normen, 2002; Katz, 1996). More recent
research reveals that exercise may not be exclusively to blame; rather there may be other
moderating variables at play in this relationship (Davis et al., 1997, 1998, 1999a).
Examining the relationship of exercise and eating pathology, and looking at other
possible factors that may or may not be contributing to this relationship, is needed to help
clarify these two very divergent schools of thought that currently exist. Furthermore,
comparing exercise dependence data, total amount of exercise data, and drive for thinness
data may offer insight into the distinction of primary versus secondary dependence of
exercise in the context of eating disorders. In other words, examining the difference
between those whom are exercising as an end in itself (primary dependence) and those
whom are exercising for some other end (secondary dependence; Hausenblas & Fallon,
2002; Hausenblas & Symons Downs, 2002).
Previous research has begun to examine these relationships between exercise
dependence and eating pathology. Hausenblas and Fallon (2002) found that primary
exercise dependence played a marginal role in predicting body image in a sample of
female undergraduate college students. They recommended for future research to
examine mediator and/or moderator effects of body image and exercise behavior.
Similarly, Zmijewski and Howard (2003) found that exercise dependence scores in
female undergraduate students were positively correlated with the Bulimia subscale of
the EAT-26 (Garner & Garfinkel, 1979) and that this subscale was also an important
predictor of exercise dependence scores. These results indicate that many college women
may be exercising in association with a formal or subclinical eating disorder (Zmijewski
& Howard, 2003). Unfortunately, these analyses stopped short of examining a potential
mediating or moderating effect.
Thus, the secondary purpose of my thesis was to examine the associations between
exercise behavior, drive for thinness scores, and exercise dependence scores. Specifically,
I examined the relationship of exercise behavior and exercise dependence scores on drive
for thinness scores. I hypothesized that drive for thinness scores (i.e. cardinal features of
an eating disorder) would be more strongly influenced by exercise dependence scores
than by total amount of exercise alone (Hausenblas & Fallon, 2002; Hausenblas &
Symons Downs, 2002; Zmijewski & Howard, 2003).
The purposes of this chapter are to present a general understanding of eating
disorders and to examine the research that is relevant to this thesis. I will begin by
describing eating disorders and their etiology of acquisition, development, and
maintenance. This will be followed by a brief review of the most widely accepted
prevalence rates for these disorders. I will then provide a basic overview of prevention
and how it is applicable to this thesis, followed by a review of the literature on recruiting
eating disordered individuals. Finally, I will present research relevant to the role of
exercise and eating disorders. The research presented in this literature review will provide
a sound rational for this thesis research.
Anorexia nervosa and bulimia nervosa are similar, yet distinct disorders which are
the two most common eating disorders; accounting for two thirds of the individuals
seeking treatment for eating disorders (Fairburn & Walsh, 1995). Each is characterized
by cognitive distortions regarding body image resulting in behaviors in an attempt to
modify the body.
The criterion for anorexia nervosa is explicitly outlined in the DSM-IV (APA,
1994) and can be generally stated as an intense and unrealistic fear of becoming fat,
engaging in behaviors intended to produce distinct weight loss, and amenorrhea resulting
from the refusal to maintain a health weight (Garfinkel, 1995). The specific criterions are
as follows. The intense fear of becoming fat experienced by anorexics does not cause an
overestimation of body part size; rather the excessive concern with weight and shape of
the body predicates the self-esteem and affect of the person. Furthermore, simply over-
estimating the size of ones body or a body part is not exclusive to eating disorders. It is
the relation of the body or body parts to the individuals' affect which distinguishes
anorexia from other disorders (Garfinkel, 1995). This disturbance of self-evaluation and
consequential denial of the brevity of one's low weight are defined in the DSM-IV (APA,
1994) as maintaining a weight that is less than 85% of what is considered an ideal body
weight for the individual. This denial is physically evident by a physiological criterion of
amenorrhea. Simply stated, this is when at least three consecutively absent menstrual
cycles occur in women. Two specific types, restricting type and binge-eating/purging
type, based on how the extreme low weight is reached and maintained, also define
anorexia. The restricting type is only defined as the absence ofbinging and purging
behaviors. The binge-eating/purging type states that during the current episode of
anorexia, the individual also engages in binges (i.e., eating inappropriately massive
amounts of food in one set period of time) or purging behavior (i.e., self-induced
vomiting, misuse of laxatives, diuretics, or enemas).
The DSM-IV (APA, 1994) criteria for bulimia nervosa are similar to that of
anorexia in that it too outlines an intense fear of becoming fat, but differentiates itself by
including the requirements of powerful urges to overeat and subsequent binges that are
followed by engaging in some sort of purging or compensatory behavior in an attempt to
avoid the fattening effects of excessive caloric intake. Similar to anorexics, the fear
experienced by bulimics is in regards to self-evaluation, thus resulting in compensatory
behaviors to evade weight gain. The paradox is the presence of the uncontrollable urges
to overeat. These binges are defined as occurring within 2 hours and including an amount
of food that is definitely larger than most people would consume in a similar time and
setting and a sense of lack of control during the binge (APA, 1994). Similar to anorexia,
these behaviors are separated into purging type (self-induced vomiting, use of laxative,
diuretics, enemas, or medication abuse.) and non-purging type (other compensatory
behaviors such as fasting or excessively exercising).
These two eating disorders present a variety of serious physical and psychological
health consequence throughout their duration. This is extremely relevant when one
considers that up to half of the mortality rate of eating disordered persons can be
attributed to cardiac complications and failure (Mehlenbeck, 2002) due to the weakening
of the individual's heart muscles and subsequent cardiac irregularities (Sobel, 2004).
Perhaps equally as troublesome is that the consequences of fasting and/or starvation as a
compensatory behavior in either anorexia or bulimia can result in structural abnormalities
in the brain that are nonreversible; even with proper refeeding and nutrition. In other
words, the individual experiences malnutrition and its physiological repercussions. Other
physical symptoms attributed to improper nutrition are kidney dysfunction, electrolyte
disturbances, dehydration, bone mineral and mass loss, and amenorrhea in anorexic
Other compensatory behaviors in both anorexia and bulimia can also manifest
themselves physically. Specifically, vomiting can result in enlarged parotid and
submandibular glands, abdominal pain, dental deterioration and gum disease, and
gastrointestinal problems such as rupture. It is also ironic that some consequences of the
attempts to control one's bodily appearance such as scaring of the hands due to acid burn
marks from self-induced vomiting, easy bruising, a compromised ability to heal wounds,
breast atrophy, dry and cracking skin, lanugo, and yellowing of the skin due to
hypercartotenemia are generally related to ideals of beauty. All of the resulting physical
deficits for all forms of compensatory behaviors listed above are the body's attempt to
shut down nonessential systems in an attempt for survival (Sobel, 2004).
The physical consequences of eating disorders are severe, and in some cases life
threatening, but the psychological consequences present equal gravity. Converse to the
previously mentioned cardiac complications being responsible for up to half of all deaths,
suicide resulting from negative psychological consequences may be responsible for the
other half of the mortality rate (Mehlenbeck, 2002). Specifically, depression and
irrational and labile moods may contribute to this aspect of mortality. Other less extreme
but still serious psychological consequences are anxiety, obsessive thoughts concerning
food and weight, increased isolation, impaired judgment, low self-esteem, guilt, shame,
feelings of imperfection, diminished concentration, and feelings of loss of control
(Mehlenbeck, 2002; Sobel, 2004). A body-image distortion is also a psychological
characteristic that is necessary for an eating disorder diagnosis. Suicide is the most
serious implication, yet self-injurious behaviors such as cutting, hitting, and scratching
may also result from the psychological distress of eating disorders (Paul, Schroeter,
Dahme, & Nutzinger, 2002).
It should be noted that other eating disorders, such as binge eating disorder, do
exist and are categorized in the DSM-IV (APA, 1994) as Eating Disorders Not Otherwise
Specified. This literature review and the intentions of this thesis will only focus on the
two more common and previously described disorders of anorexia and bulimia and there
spectrums of risk, development, and prevention.
The examination of the etiology and prevalence of eating disorders is multifaceted
and points out the methodological difficulties in identifying the development and breadth
of this problem. Cooper (1995) offers an etiological paradigm to consider the course of
eating pathology (Figure 2-1). This etiological model breaks down the process of
development, progression, and maintenance of eating disorders into three distinct stages.
Stage one examines the period from conception to the emergence of behavioral
precursors of the disorder, followed by stage two covering the period from the
development of behavioral precursors through any precipitating factors that may lead to
the onset of the full disorder, and ending with a description of how various maintaining
factors interact with protective factors and determine whether the disorder will take a
transient course or become established or chronic during stage three. Each of the stages is
described in detail below.
Stage one primarily examines risk factors that lead to the behavioral precursors of a
disorder. During this time, the individual may be exposed to predisposing factors for the
disorder that occur prior to onset of the disorder and in turn increase the person's risk for
development of the disorder. Stice (2002) concisely organized the risk literature into a
comprehensive meta-analysis and identified which risk factors are relevant to eating
disorders. His meta-analysis found that several factors believed to place an individual at
risk were not empirically supported, while other lesser know factors did received support.
For example, childhood sexual abuse, the role of stress, control issues, dysfunctional
family systems, and deficits in parental affection are theorized as risk factors, yet
received no empirical support in this meta-analysis. These findings suggest that there is
not an individual, univariate relationship between a single risk factor and the
development of eating disorders, but rather, several factors interact to spark the
development and others may be present to move a person further along the spectrum of
development of pathology to disorder. Interestingly, individuals at risk in the presence of
pathological behaviors and their perception of these messages may encourage further
progression of eating disorders. Stice explains this by proposing a Perfectionism X Body
Dissatisfaction X Low Self-Esteem model of development of eating disorders. This
relates to Cooper's model by explaining that individuals possessing the set of factors
identified by Stice would then represent the initial stage of eating pathology.
Most people will exhibit some known eating disordered risk behavior, such as
fasting or restrictively dieting, and never develop an eating disorder (Tylka, 2004; Tylka
& Subich, 2002). This behavior alone does not qualify someone as having an eating
disorder. The primary goal of stage two is to examine what other factors combine and
interact with various known eating disorder risk factors, which in turn, trigger
progression into an eating disorder. Little research has examined this line of questioning.
The sparse amount of existing research is limited in that any possible contributing factors
have not been clearly documented. Furthermore, it is difficult to separate those with
eating disorder symptoms and those who will go on to develop the actual eating disorder.
The third stage of this model explains the course and maintenance of eating
disorders by offering three views. The cognitive view suggests that characteristic
cognitive distortions of the extreme importance of weight and shape explain most of the
features of eating disorders. Therefore treatment efforts, such as cognitive behavioral
therapy, are targeted at changing these mistaken attitudes. The second view stating that
interpersonal events shape the course of an eating disorder is based on the observation
that anorexics experiencing relationship and social changes during the course of family
therapy and bulimics experiencing the same events while engaged in interpersonal
therapy appear to have beneficial effects. These interpersonal events may enhance or
weaken self-esteem, which then lead to cognitive distortions being changed. The third
view is physiologically based and mainly pertains to anorexia. It is support by the
physiological changes that occur as a result of starvation. Some of these changes may
then perpetuate the anorexic cycle.
Cooper's etiological view offers a general understanding of how risk factors
introduce, progressively facilitate the behavioral components leading to an actual eating
disorder, and maintain the course of a disorder, but fails to consider that protective factors
may be at work in stopping this progression in the majority of individuals exhibiting risk
behaviors. Furthermore, much of the research examining this has used clinical samples of
eating disorder patients. This type of selection bias may offer insights into a
subpopulation of eating disordered individuals but not necessarily capture the true nature
of eating disorders. This bias is even more of a shortcoming in the literature given that
many eating disordered individuals do not seek treatment (Fairburn, 1995; Hoek, 1995)
and therefore are not capturing a sizable sub-population. Other methodological
considerations are that many studies do not include control groups and very few studies
have attempted to examine the occurrence of etiological factors.
SDevelopment f Establishedf
BI b Be hav Preeuro Onset Chronic
Predisposing Precipitating Maintaining
Factors Factors Factors
Stage 1 Stage2 S tage3
Figure 2-1. Etiological Model of Eating Disorder Development and Maintenance
Most people will face socio-cultural pressures concerning weight and exhibit some
risk factors over the course of a lifetime (Tylka, 2004; Tylka & Subich, 2002). Separating
those who experience such factors from those who go on to develop an actual eating
disorder can be difficult. Stringent criteria for diagnosis have yielded the clearest picture
of how many people are suffering from actual eating disorders. Hoek (1995; Hoek & van
Hoeken, 2003) offers a two-stage approach in determining the prevalence rates of eating
disorders. Currently, this two-stage approach is the most widely accepted approach in
determining eating disorder prevalence (Hoek & van Hoeken, 2003). Other approaches,
such as survey methods, have yielded drastically different results.
The first stage of Hoek's screening model involves administering a screening
survey to large populations of those who are considered to be at risk to determine the
number of individuals who could have an eating disorder. The second stage consists of
semi-structured interviews with those individuals who are determined to potentially have
an eating disorder. This type of rigorous screening and evaluation has shown the point
prevalence rate for anorexia nervosa is 280 per 100,000 young females (0.28%) and the
point prevalence of bulimia nervosa is 1,000 per 100,000 young females (1%; Hoek,
1995). Recent reexaminations of prevalence rates have reported similar results; 0.30%
anorexia in young females and 1% bulimia in young females (Hoek & van Hoeken,
2003). Results for males with eating disorders were reported as 5 -10% of clinical
samples of eating disordered patients thus supporting women being at higher risk than
men. The surveys used in stage one of this model also indicate that bulimic symptoms are
present in 19% of female students. This is supported by previous research reports
indicating that medical experts estimate that 16 to 30 percent of all women have practiced
bulimic behaviors (Cassell, 1994).
Closer examination of prevalence rates over a one-year period yields a more
comprehensive view of the overall prevalence of eating disorders. Table 2-1 presents
Hoek's (1995) findings. Level 0 represents the number of eating disordered individuals in
the community. Level 1 represents the number of individuals considered to have an
eating disorder by their primary health care physician. Level 2 represents the number of
eating disordered individuals receiving outpatient or inpatient treatment. In regards to the
claim that eating disorders, particularly bulimia nervosa, are secretive in nature, (Fairburn
& Cooper, 1982; Hoek, 1995) examination of the prevalence rates listed in Table 2-1
shows that only 43% of the cases of anorexia nervosa in the community are recognized
by primary care physicians and of those, 79% are referred for treatment. The
secretiveness of eating disorders becomes more apparent when you consider that only
11% of bulimics in the community were recognized by their primary care physician and
of those, only half (51%) are referred for treatment. In other words, most eating
disordered individuals do not receive adequate treatment (Hoek & van Hoeken, 2003).
Less stringent methods of prevalence evaluation yield drastically different results.
For example, Kjelsas, Bjnmstrom, and Gunnar Gdtestam (2004) administered self-report
surveys to 1987, 14 15 year old Norwegian students (1034 girls and 953 boys) in class.
Prevalence rates reported from this study were 17.9% for anorexia nervosa and 0.7% for
bulimia nervosa in girls and 0.2% for anorexia nervosa and 0.4% for bulimia nervosa in
boys. Interestingly, a 14.6% prevalence rate of eating disorder not other wise specified
was reported for the girls. This may be consistent with previously discussed results of 16
- 30 % rates of bulimic behaviors reported in the same age group (Cassell, 1994).
However, the drastic difference in anorexia nervosa prevalence rates illustrates the
importance of using strict criteria such as Hoek's two-stage model.
Table 2-1. One-Year-Period Prevalence Rates per 100,000 Young Females
Level of health care Anorexia Nervosa Bulimia Nervosa
0. Community 370 1,500
1. Primary Care 160 170
2. Mental Health Care 127 87
A major concern is that these prevalence rates and examination of prevalence of
treatment only account for those individuals whom have developed a diagnosable eating
disorder. Many others may be on their way to developing an eating disorder, thus placing
them at risk for at least some of the physical and psychological risks stated above. Tylka
and Subich (2002) found that high school and college women reported frequently
skipping meals (59%), ate fewer than 1200 calories per day (36.7%), eliminated fat
(30.1%) or carbohydrates from their diet (26.5%), fasted for more than 24 hours (25.9%),
used laxatives (7.2%), used diuretics (6.6%), and vomited after meals (4.8%). Tylka and
Subich concluded that these behaviors are similar to clinical eating disorders but without
the frequency required to be considered serious. This is similar to Thomsen, Weber, and
Brown's (2001, 2002) findings in a sample of high school girls who read women's health
and fitness magazines. They found a positive association between reading such
magazines and frequency of unhealthy weight control behaviors; specifically taking diet
pills and restricting caloric intake. The authors also found that frequent readers of such
magazines scored significantly higher on scales of eating disordered cognition. While
these results do not specifically show a relationship to diagnosable eating disorders, they
do show a relationship to the behavioral precursors to eating disorders.
Eating disorders are a serious threat to many young people's lives that, as
suggested by the differences in Hoek's three level prevalence rates, will go unnoticed and
otherwise ignored until it is much too late. The negative physical and psychological
consequences of these disorders begin to impact the individual from the onset of
development. By the time an individual is diagnosed with an eating disorder, they are
experiencing at least some of these consequences. Simply stated, the quality of these
person's lives is compromised by thoughts and behaviors that must be intervened upon.
Understanding that most of individuals developing an eating disorder will not be
recognized or seek or be referred to treatment raises an immediate need for improved
recruiting practices in this population as well as improved prevention efforts.
Furthermore, prevention of eating disorders becomes increasingly important given the
knowledge of negative behaviors such as those reported by Tylka and Subich (2002) and
their potential for development into more serious pathologies.
If something, such as an eating disorder, can quietly affect the health and lives of so
many individuals, then the most prudent course of action would be to stop the problem
before it begins. Perhaps then the most logical approach to dealing with eating disorders
is to address the issues of primary, secondary, and tertiary prevention efforts (Caplan,
1964). Several eating disorder prevention interventions have been conducted, mostly at
the primary prevention level, producing varying degrees of success. Many early
interventions have yielded more cautious results, while recently reported interventions
have shown the most promising effects. Identifying areas in need of improvement, and
then continuing this progression of more effective prevention programs is needed to
advance the efforts of previous research.
Primary prevention efforts attempt to intervene on the population as a whole and
stop negative health behaviors before any signs of an illness are present. This goal is
accomplished in two varying methods: reactive and proactive efforts (Catalano &
Dooley, 1980). Reactive primary prevention or selective prevention (Levine & Piran,
2004) is defined as strategies that improve coping responses and augment the individual's
resistances to any potentially harmful stressors. It is selecting those who have not shown
signs of an illness or condition, but are considered to be at a higher risk. For example,
administering a vaccine for the measles to children would be considered reactive primary
prevention. On the other hand, proactive primary prevention is defined as strategies that
eliminate causal agents. Proactive primary eating disorder interventions typically target
populations that generally have not yet begun to exhibit any behavioral precursors to
eating pathology, such as elementary school and middle school students. This type of
prevention is also called universal or public health prevention because it is often
accomplished in the form of public policy or community change (Levine & Piran, 2004).
Secondary, or indicated, prevention targets those individuals whom have begun to
show signs of a problem but are not quite experiencing the full-blown problem or illness
(Levine & Piran, 2004). Secondary prevention eating disorder interventions focus on
known precursors, such as negative body image (Stice, 2002), to reduce the risk of
experiencing the full-blown disorder. The tertiary level of prevention is an attempt to
improve the course of a disease or minimize the harmful effects of an illness once it has
been diagnosed. In other words, it is basically a course of therapy or treatment for an
illness or disorder. Cognitive behavioral therapy for diagnosed eating disordered patients
is the most widely employed tertiary approach and has been extensively supported for use
in bulimia nervosa (Fairburn, 1995; Herzog et al., 1992; Vitousek, 1995).
Several studies have addressed preventing eating disorders at each of the previously
described levels. Austin's (2000) literature review and Stice and Shaw's (2004) meta-
analysis summarize a wide range of these studies and offer insight into the future of
eating disorder prevention.
Austin (2000) found that of 20 eating disorder prevention interventions, only four
reported positive behavioral changes, while 14 showed improvements in knowledge
and/or attitudes towards eating disorders and concerns about weight and shape. The main
reason these interventions have, for the most part, failed to elicit behavioral changes is "a
fundamental disjunction in the transition from theories of aetiology to theories of
prevention" (Austin, 2000; p.1256). Specifically, many of the interventions focused on
sociocultural risk factors of eating disorders, yet attempted to intervene by focusing at the
individual level. Therefore, implementing proactive primary prevention programs similar
to those commonly used in other areas of public health is the logical next step in eating
disorder prevention. Interestingly, recommendations for the prevention of public health
issues such as coronary heart disease, diabetes, osteoporosis, obesity, and cancer (CDC,
1996; USDHHS, 1990, 1996; Seidell, 1999) overlap with recommendations in the eating
disorder prevention literature (Battle & Brownell, 1996; Killen et al.,1993; Neumark-
Sztainer, 1996; Smolak & Schermer, 1998) in that they all call for programs focusing on
nutrition and exercise.
Since Austin's (2000) literature review of eating disorder prevention programs,
there has been encouraging progress in the evolution of eating disorder prevention. Stice
and Shaw's (2004) meta-analysis of 51 eating disorder prevention found that
prevention have undergone three generations of development. The preliminary research
attempted to intervene by delivering didactic psycho-educational materials which
informed general populations about the adverse health consequences of eating disorders.
The second generation of interventions was similar in that they were universally
distributed and didactic but also included socio-cultural education. The most recent
prevention programs have moved away from universal didactic approaches and focused
on delivering interactive materials and exercises to population that are at risk. Overall,
the most effective programs reduced attitudinal risk factors and promoted healthy weight
control behaviors. This is similar to Austin's (2000) recommendations as discussed
above. In contrast to Austin's proposed public health model, Stice and Shaw (2004)
recommend that programs target selected populations rather than universally intervene on
general populations. Specifically, prevention programs that were targeted, interactive,
and multi-session programs all out performed universal, didactic, and single-session
programs (Stice & Shaw, 2004).
Implementing a prevention program presents a unique obstacle of combining the
recommendation to target specific populations (Stice & Shaw, 2004) with the knowledge
of which risk factors constitute these populations (Stice, 2002). A general search of the
literature identified articles concerned with recruitment for a variety of other behavioral
health problems. The most relevant of which were included in a literature summary
(Lovato et al., 1997). A much broader search of the topic of recruitment shows that the
internet is a viable means of reaching large populations. A review of the eating disorder
literature found that despite a sizable body of intervention and prevention research,
amazingly, only one study has specifically addressed recruiting as its primary objective.
Recruitment is a basic and important issue for any study involving human
participants. Lovato et al. (1997) found over 4000 citations between 1987 and 1995
related to recruiting individuals for a number of studies. Most provided anecdotal reports
of recruitment efforts and little or no data to support any conclusions (Lovato et al.,
1997). The most common strategies of recruitment reported were: using patient registries,
occupational screening, direct mail, and using the media.
A more broad view of potential recruitment pathways leads one to question if
interactive technologies could be useful in reaching problem populations. The American
Psychological Association's Board of Scientific Affairs has established an advisory group
that has released a report outlining several issues related to internet research and in doing
so have provided encouraging preliminary accounts of the widespread reach of this
technology. In the section of this document pertaining only to recruitment, Kraut et al.
(2004) identify several benefits of using the internet for recruitment including low cost,
the ability to attract a large and diverse sample, and the ability for undergraduates,
graduate students, and researchers at smaller institutions to all contribute original
research. In essence, the internet has democratized data collection (Kraut et al., 2004).
Examples given include a sample of over 1.5 million completed responses collected in a
four year span (Nosek, Banaji, & Greenwald, 2002b in Kraut et al., 2004), 40,000
respondents from a link on the National Geographic website (Wellman, Quan Haase,
Witte, & Hampton, 2001 in Kraut et al., 2004), and over 18,500 responses from 3,300
participants recruited from an online video game over the course of 7 months (Kraut et
al., 2004). Clearly the internet provides a wider breadth over a shorter time period than
Recruiting special populations, such as individuals at any point on the continuum of
pathological eating, presents its own specific challenges and therefore the common
recruitment methods listed above may or may not be entirely applicable. McDermott et
al. (2003) examined recruitment as part of a larger study involving anorexia nervosa
relapse prevention and two similar studies treating bulimic women. Advertisements in
newspapers, on radio stations, and posting flyers in various locations at a site in New
York, Minnesota, and California were the main reported methods of recruitment. This
study found that using multiple sites to recruit during shorter periods is a preferable
method of recruiting versus using fewer sites and larger windows of time.
While no studies report on this question specifically in terms of eating disorders,
several recent studies have examined interactive technologies with eating disordered
individuals. Myers et al. (2004) summarized the state of several alternative delivery
systems and new technologies employed in this population, including email, the internet,
CD-ROMS, computer software, portable computers, and virtual reality. Possible
advantages listed specifically for computer based delivery systems (email, internet, CD-
ROM, software, etc.) were: cost-effectiveness, accessibility, portability, the possibility
that they may elicit less resistance compared to face to face contact, and the ability for
simply structured programs. Email and internet delivered programs also were found to be
easier to use, have widespread and pragmatic applications, are relatively accessible, and
are very convenient for the participant. The general consensus in the literature is that
these alternative methods hold a great deal of promise when applied to eating disordered
populations (Myers et al., 2004).These conclusions are supported by a meta-analysis of
similar research finding that web-based interventions are more effective than non-web-
based interventions in producing attitudinal and behavioral changes in a variety of health
behaviors (Wantland et al, 2004).
Continuing to apply new computer based technologies to the issue of recruiting
pathologically eating populations may be a logical step based on the reported preliminary
success (Myers et al., 2004; Wantland et al., 2004). These technologies could very easily
be applied to the McDermott et al. (2003) conclusion that recruitment should focus on
many sites over a short period rather than fewer sites over a long period. Commonly
listed methods of recruitment such as advertisements, flyers, and announcements (Lovato
et al., 1997) are only seen or heard by individuals whom frequent the specific location, or
site, of the advertisement. Sampling captive audiences in this way is the traditional
method of recruiting on college campuses for a variety of studies just as sampling
inpatients is the tradition for eating disorder research (Cooper, 1995). Both the
undergraduates and the inpatients will refresh their respective population pools at the
beginning of a new semester or at new patient intakes. Therefore, recruiting enough of
either group would require longer windows of possible recruitment. Alternatively, a one
time only, web based or email announcement could be sent out to entire populations
would turn each email into its own site of recruitment. In other words, it is not limited to
only those who will be in class or pass by an advertisement or in inpatient care and thus
would not require longer amounts of time for the eligible pool to refresh itself. In
essence, this is analogous to the McDermott et al. (2003) recommendation of multiple
sites over a short period because emails or web-based announcements can reach anyone
anywhere, while the announcements and/or flyers over several semesters or the
recruitment of patients at intake are all analogous to fewer sites over longer periods of
time. Therefore, the next step in improving the recruitment of pathologically eating
populations is to examine the possibility of using new and alternative technologies to
sample populations that are larger than were traditionally accessible.
Exercise and Eating Disorders
A review of the literature reveals that the relationship between exercise and eating
disorders began with the observation of high levels of exercise in clinical samples,
questioned if exercise somehow exacerbates eating pathology, and has progressed to
understand that there are other moderating variables involved in this relationship.
Relevant research and a further explanation of each of these views are presented below.
The traditional view of the relationship between exercise and eating disorders has
primarily been based on clinical observations from the 1960s' and 1970's that 65% to
75% of anorexic patients were excessively physically active at some point during the
course of their disorder (Katz, 1996). This observation has seemed to carry over to a
more universal view of all eating disorders as opposed to staying exclusively aligned with
anorexia. Research has followed this clinical observation and has shown that there may
be some validity to this relationship. The best evidence of this comes from research
finding that strenuous exercise can suppress appetite (Rivest & Richard, 1990). Rivest
and Richard conditioned rats to become accustom to eating 3 meals per day in 8 hour
intervals. Some rats were then injected with a corticotropin-releasing factor antagonist
and other with saline. All of the rats were then forced to exercise on motorized wheels for
40 mins prior to feeding. The rats that did not receive the corticotropin-releasing factor
antagonist then began restricting the amount of food pellets they ate, while the other
groups of rats did not. It was concluded that the physiological effects of exercise were
what was suppressing the food intake of these rats. Research such as this is the
physiological basis to show the anorectic effects of exercise.
Replicating these findings in humans is difficult because forcing humans to
excessively exercise would raise ethical concerns. Still, research has focused on showing
similar effects of exercise and arrived at the same conclusion that excessive exercise may
contribute to eating disordered and pathological weight control behavior. For example,
studies such as Hoglund and Normen (2002) examined total amount of exercise in female
aerobic instructors, a group known to have high levels of total exercise. They
hypothesized that high amounts of exercise, simply defined by total hours per week,
would be associated with a strong focus on weight control and body shape. They also
hypothesized that previous experiences of anorexia or bulimia would be more common in
the high exercising group when compared to the general public of similar age and
demographic characteristics. The findings of this study suggest that the first hypothesis
concerning total amount of exercise was supported by participants reporting exercise
levels 1.5 3 times higher than that of the same population in the general public. The
high exercising group also exhibited more pathological weight control behaviors and less
body satisfaction compared to the low exercising group. The second hypothesis was also
supported by 35% of participants reporting currently or previously having an eating
Findings such as Hoglund and Normen (2002) can only make inferences based on
two variables (exercise and presence of eating pathology). Other studies have examined
the possibility that the high amounts of exercise commonly seen in eating disordered
individuals may be a function of some other moderating influence. Davis and colleagues
have published several articles concluding that this moderating effect is likely to play a
role in exercise and pathological weight control. In a study examining the prevalence of
high exercise in an eating disordered population, Davis et al. (1997, 1999b) found that
anorexia was linked to childhood physical activity levels that were higher compared to
other children of the same age. While this alone is similar to Hoglund and Normen
(2002), these results indicated that excessive exercisers also maintained more obligatory
and pathological attitudes to exercise. This is an important result because it allows for the
possibility of some other factor interacting with the more simplistic idea of high exercise
being directly related to more pathology. Subsequent studies have indeed shown this to
be a likely possibility. Davis et al. (1998, 1999a) compared measures of high exercising
and moderately or not exercising inpatients on obsessive-compulsive symptomatology.
The reported results showed that the high exercising group exhibited greater obsessive-
compulsive symptomatology and obsessive-compulsive personality characteristics than
the group of moderate or non-exercising individuals, and that this underscores a
moderating influence of physical activity. However, it was not concluded that over
exercising is uniquely associated with greater obsessionality (Davis et al., 1999a). While
these findings do show that there is a relationship between obsessionality and eating
pathology, they failed to conclude that it was the amount of exercise that influenced
obsessionality. Future research will advance these findings by further examining the
relationship between exercise and eating pathology. Specifically, further examination of
the mediating and/or moderating influence of exercise on eating pathology is
recommended (Hausenblas & Fallon, 2002).
Participants were 330 female college students (age 17-36, M age = 19.97, SD =
2.14). Women in this age group were chosen because this cohort is at greater risk for
eating disorders compared to men of the same age, and incidences of eating disorders
typically peak during adolescence to early adulthood (Jacobi et al., 2004; Stice, 2002).
Most of this sample was Caucasian (67.3%), followed by Hispanic (13.3%), African
American (10.9%), and Asian (4.5%). About 29% of the participants were in their junior
year of school (28.8%), followed by freshman (25.2%), seniors (23.0%), and sophomores
Participants reported their age, year in school, course of study, ethnicity, family
income level, major in school, height, current weight, ideal weight, contact information,
eating disorder history, weekly exercise level, and where they saw the advertisement for
this study. (see Appendix A)
Drive for Thinness Subscale
The Drive for Thinness subscale of the Eating Disorder Inventory-2 (EDI-2)
measures excessive concerns with dieting and weight preoccupation. This scale are
measured by rating items on a 6-point Likert scale ranging from 1 (never) to 6 (always).
Ratings totaling a higher score indicate greater endorsement of the attitudinal and
behavioral correlates of eating disorders. Extensive research supports the validity and
reliability of the Drive for Thinness subscale (Garner, 1991). (see Appendix B)
Exercise Dependence Scale (EDS)
The EDS is a 21-item measure of exercise dependence symptoms on the following
seven subscales based on the criteria for substance dependence (American Psychiatric
Association [APA], 1994): Tolerance (e.g., I continually increase my exercise frequency
to achieve the desired effects/benefits), Withdrawal Effects (e.g., I exercise to avoid
feeling tense), Continuance (e.g., I exercise despite persistent physical problems), Lack of
Control (e.g., I am unable to reduce how intense I exercise), Reductions in Other
Activities (e.g., I think about exercise when I should be concentrating on school/work),
Time (e.g., I spend a lot of time exercising), and Intention (e.g., I exercise longer than I
expect; Hausenblas & Symons Downs, 2002). Responses to the items are on a 6-point
Likert scale ranging from 1 (never) to 6 (always). A lower score reveals less exercise
dependence symptoms. The psychometric properties of this scale are good (Hausenblas &
Symons Downs, 2002; Symons Downs, Hausenblas, & Nigg, 2004). (see Appendix C)
Body mass index (BMI) was calculated from self-reported height and weight by
dividing weight (kg) by height (m) squared. A BMI score of 18.5-24.9 is normal weight,
25.0 29.9 is overweight, and a BMI score of 30.0 or greater is obese (ACSM, 2000).
There is however, a 5% error rate associated with BMI when used to estimate body fat
percentage compared to 3.5% error rate of skin fold calipers (ACSM, 2000).
Leisure-time Exercise Questionnaire (LTEQ)
The LTEQ is a self-report of the frequency that an individual engages in strenuous,
moderate, and mild bouts of exercise during a typical week (Godin & Shephard, 1985).
Each of the subscale scores are converted into metabolic equivalents (METS; [Mild x 3]
+ [Moderate x 5] + [Strenuous x 9]) and summed to provide an estimate of total METS
expenditure from exercise for an average week. The LTEQ has adequate reliability and
validity (Jacobs, Ainsworth, Hartman, & Leon, 1993). (see Appendix D)
Participants were recruited by responding to an advertisement for larger study
examining an exercise program designed to improve physical appearance and body image
(see appendix E). This advertisement was posted throughout the month of September and
was identical regardless of where or when it was announced in the following four ways:
1. The campus announcements were 45 announcements posted on campus bulletin
boards available for posting such information for students in Turlington Plaza, on
the Reitz Union lawn, and on telephone poles adjacent to bus shelters near sorority
2. The classroom announcements were made in 12 sport and fitness classes and 4
3. The website announcement was placed on my.ufl.edu, which is a secure website,
accessible only to university students.
4. The email announcement was sent once as part of a weekly list serve email which
delivers information only to undergraduate students.
Interested persons were directed to email a request for access to a secure website
containing screening questions for inclusion in this study. A standard email response
containing a unique personal identification number (PIN) and informed consent form was
sent as a reply to email requests for this study (see appendix F). At their own
convenience, participants then accessed a website containing the Leisure-time Exercise
Questionnaire (Godin & Shephard, 1985), Drive for Thinness subscale of the EDI-2
(Garner, 1991), the Exercise Dependence Scale (Hausenblas & Symons Downs, 2002),
and demographic questions. A follow-up email reply was sent to all participants after the
data were analyzed. Individuals who met eligibility criteria for the exercise intervention
were then invited to join the program (see appendix G), while all others were thanked for
their time and participation (see appendix H). Study flow is presented in Figure 3-1. This
study was completed over the course of one academic semester. The timeline for this
study is presented in Table 3-1.
Figure 3-1. Study Flow
Table 3-1. Thesis Timeline.
September October November
Month 1 Month 2 Month 3
Study Announcement X
Participant Inquiry X
Email Reply X
Web Based Assessment X
Data Analysis X
Follow-up Reply X X
Preparing and Reporting Results X X
To obtain a targeted population of women at-risk for an eating disorder, I used the
following eligibility criteria.
1. Low Active Only those participants who were exercising 4 or fewer times per
week were selected. The total amount of exercise per week was measured by the
Leisure-time Exercise Questionnaire, as well as an open ended question in the
demographic section of the survey simply asking, "How many times a week do you
exercise?". The screening website allowed for alphanumeric data entry on LTEQ
questions. Through this it was found that 52 participants (15.75%) included
qualitative data stating that they were reporting inappropriate physical activity,
rather than exercise behavior, in the LTEQ and therefore were over reporting total
weekly exercise amount. Examples of these responses are: depends, varies, walking
to class, ideally once or twice, sometimes, sporadically, whenever I a in the mood,
and the days that I am late to class. The single exercise behavior question was
chosen to determine eligibility because it was apparent that LTEQ questions were
not accurately capturing the nature of exercise in this sample. (see chapter 4 results
section for statistical justifications) The total weekly amount of exercising 4 times
or less was chosen because it reflected that these individuals were not meeting
ACSM (2000) criteria for total weekly exercise. Research has shown that single
item measures of exercise behavior have reasonable validity (Schectman et al.,
1991; Prochaska et al., 2001; Weiss et al., 1990).
2. Drive for Thinness Drive for Thinness score cutoffs of 8 or greater were chosen
because this reflects the 75th percentile and therefore indicates a risk for
developing eating disorders (Garner, 1991). A score of 14 or greater, or the 91st
percentile, on the Drive for Thinness would typically be used to indicate the
potential of more serious risk of eating pathology (Garner, 1991).
3. Body Mass Body Mass Index (BMI) scores between 18.5 and 29.9 were used as
an eligibility criterion. This range is considered the normal to overweight range
(ACSM, 2000) and was chosen because normal to overweight women are at a
higher risk for the development of eating disorder (Jacobi et al., 2004).
Measures of internal consistency were performed on the Exercise Dependence scale
and the Drive for Thinness subscale. Descriptive statistics and frequencies were also
performed on all relevant variables.
A 4 (recruitment levels campus announcements, classroom announcements,
website announcement, email announcement) x 2 (eligible or ineligible) ANOVA was
performed to determine if individuals meeting eligibility criteria were more prevalent
from any of the four methods of recruitment examined. A one-way ANOVA was used to
determine differences in self-reported exercise from the two previously discussed
questions measuring exercise behavior. Similarly, a one-way ANOVA was used to
examine how responses to these questions would affect eligibility rates. Finally, an
independent sample t test was performed to determine if new alternative methods of
recruitment gathered significantly more eligible individuals than traditional methods of
The procedure for the secondary purpose of this thesis, examining the hypothesized
relationship of Leisure-time Exercise scores (LTEQ), Exercise Dependence scores (EDS)
and Drive for Thinness (DT) scores, was followed as detailed by Baron and Kenny
(1986). My independent variables were LTEQ scores and EDS scores and my dependent
variable was DT scores. First, correlations were used to determine the relationship
between the independent and dependent variables. A significant correlation indicates a
mediator effect, while a nonsignificant correlation indicates a moderator effect (Baron &
Kenny, 1986). See figures 3-2 and 3-3.
Figue 32 R ad o qufiato ri
Figure 3-2 Relationship of Mediator Variables
Figure 3-3 Relationship of Moderator Variables
Four hundred forty eight individuals responded to the announcement and sent
emails indicating interest in the study. All 448 were sent an email reply including a
personal identification number (PIN) to access the website (see appendix F). Twenty
participants (4.5%) lost or misused there PIN when logging in and emailed a request for a
new PIN. Eight participants (1.8%) wrote back explicitly saying they would not like to
continue with the study. A total of 330 participants (73.7%) accessed and completed the
First, descriptive statistics such as means, standard deviations, frequencies, and
alpha values, were conducted for the primary purpose study variables. (See Tables 4-1 -
4-2.) The eligibility criterion was then calculated resulting in 87 eligible and 241
ineligible individuals. A 4 (recruitment levels) x 2 (eligible or ineligible) ANOVA was
performed to determine if individuals meeting eligibility criteria were more prevalent
from any of the six methods of recruitment examined. Two recruitment groups ("found
out from a friend" and "other") were not included in this analysis because only 2
individuals were recruited from each. No significant influence of method of recruitment
on eligibility was found, F (3, 319) = 1.04, p = .38. A Pearson chi-square test also
revealed that the recruitment groups did not differ significantly for eligibility [x2 (3) =
3.12, p = .37].
Table 4-1 Means, Standard Deviations, and Alpha Values
Variable M SD a
Exercise per week (single question) 2.21 1.71
LTEQ 36.81 23.02
EDS 40.63 13.09 .93
DT 6.68 6.00 .88
BMI 25.16 4.77
Age 19.97 2.14
Table 4-2 Frequency of Recruitment Method
Announcement Frequency Percent
Ad posted on campus 12 3.6
Ad posted on my.ufl.edu 17 5.2
Classroom announcement 67 20.3
From a friend 2 0.6
Wednesday Update email 229 69.4
Other (unspecified) 2 0.6
Missing 1 0.3
Total 330 100.0
A one-way ANOVA was used to determine differences in self-reported exercise
from the two questions measuring exercise behavior (LTEQ and a single question). This
ANOVA compared the amount of exercise reported in the Leisure-Time Exercise
Questionnaire (LTEQ) and the single question, "How many times a week do you
exercise". The dependent variable was frequency of exercise reported and the
independent variable was which question the frequency was measured by. I found that the
self-reported total amount of exercise per week was significantly lower when reported
through a single question than by the total of strenuous and moderate exercise measured
by the LTEQ [F (1, 647) = 207.96, p = .001]. These results also indicated that a
significant positive correlation existed between these two variables [r = 621, p =.001].
This self-report of exercise was important in the determination of eligibility. A one way
ANOVA was used to examine how responses to these questions would affect eligibility
rates. The dependent variable was eligibility and the independent variable was which
question the exercise frequency was used to determine eligibility. I found that using the
single question to measure weekly exercise behavior yielded more eligible persons than
by using the LTEQ [F (1, 647) = 207.96, p = .001]. In other words, the lower amounts of
exercise reported by the single question indicated more inactive, and therefore eligible,
persons than did the LTEQ.
Finally, an independent sample t test was performed to determine if new alternative
methods of recruitment gathered significantly more eligible individuals than traditional
methods of recruitment. Alternative methods were defined as announcements posted on
the internet website and the email announcement. Traditional methods were defined as
announcements posted on campus and announcements made in classrooms. The two
other levels of the recruitment variable ("from a friend" and "other") were excluded due
to the uncertainty of where the announcement was originally seen. The independent
variable was method of recruitment (traditional or alternative), and the dependent
variable was eligibility. The Levene's test for equality of variance was significant [F=
14.344, p = .01], therefore the equal variances not assumed statistic was used to interpret
the t test. Traditional and alternative methods of recruitment were not significantly
different in their ability to reach eligible participants [t (146.761) = -1.823, p = .07].
Correlations were performed between exercise dependence scores (EDS), drive for
thinness scores (DT), and Leisure-Time Exercise Questionnaire scores (LTEQ). I found
that all of the variables were significantly correlated. That is LTEQ was significantly
correlated with EDS scores [r = .440, p =.001], followed by EDS and DT being
significantly correlated [r = .207, p =.001], and LTEQ being significantly correlated with
DT [r = .149, p =.008]. Because LTEQ and DT score were significantly correlated and
LTEQ and EDS scores were also significantly correlated, mediation was assumed and a
three-step mediation model was followed (Baron & Kenny, 1986). Furthermore, exercise
and exercise dependence exhibit a temporal relationship with exercise preceding exercise
dependence (Hausenblas & Symons Downs, 2002; Symons Downs, Hausenblas, & Nigg,
2004). This is also an indication that the relationship may potentially be that of a
mediator and not a moderator (Baron & Kenny, 1986; Kreamer et al., 2001). This model
outlines that three forced entry regressions must be performed. First, a simple linear
regression with forced entry was performed with EDS (dependent variable) regressed on
LTEQ (independent variable). Results of this regression showed that higher LTEQ scores
(P = .440, p = .001) resulted in higher exercise dependence scores [F (1,306) = 73.38, p =
.001]; with 19.3% of the variance in exercise dependence explained by LTEQ. Next, DT
(dependent variable) was regressed on LTEQ (Independent variable). Results of this
regression showed that higher LTEQ scores (P = .149, p = .008) resulted in higher drive
for thinness scores [F (1,312) = 7.12, p = .008]; with 2.2% of the variance in exercise
dependence explained by LTEQ. Finally, a forced entry multiple regression was
performed with DT (dependent variable) regressed on LTEQ and EDS (independent
variables). This regression found that EDS and LTEQ explained 4.6% of the variance in
drive for thinness scores [F (2,296) = 7.06, p = .001], with only EDS scores being a
significant predictor of DT (P = .180, p = .005).
For mediation to be found, the independent variable in the first regression (LTEQ)
must affect the mediator (EDS). The second regression must show that the independent
variable (LTEQ) must affect the dependent, or outcome, variable (DT). Finally, the third
regression must show that mediator (EDS) affects the dependent variable (DT). If the
beta of the independent variable becomes non significant, then there is complete
mediation (Baron & Kenny, 1986). Based on these criteria and the results of these three
regressions, a full mediator effect for EDS on the relationship of LTEQ and DT was
The purpose of this thesis was twofold. First, the ability to effectively recruit a
population of college age women dissatisfied with their body was examined. Specifically,
traditional modes of recruitment, such as verbal announcements and campus postings
were compared to announcements, made on newer, alternative mediums such as the
internet and email. I hypothesized that these alternative channels of recruitment would
produce both more interest and more individuals indicating potential risk for the
development of eating pathology. While this hypothesis was not statistically supported,
the raw data did show that the alternative channels of recruitment do cover a substantial
breadth of potential participants. Second, the relationship between exercise, exercise
dependence, and drive for thinness was examined. I hypothesized that exercise
dependence would affect the relationship between exercise and drive for thinness. This
hypothesis was fully supported. A more detailed description of my study findings,
limitations, and future research directions are discussed below.
The results of this study indicated that there is no significant difference of how or
where participants are recruited on the amount of potentially at risk individuals that are
reachable by each recruitment pathway. While these results are not significant
statistically, they are of note practically because these results were approaching
significance [p = .07]. The relatively small sample size of this study may have
contributed to these results by lowering the power of these analyses.
Researchers concluded that recruitment efforts should be short in duration and
widespread (McDermott et al., 2003); and that alternative technologies may be a viable
means of achieving these recommendations (Kraut et al., 2004; Myers et al., 2004;
Nosek, Banaji, & Greenwald, 2002). This study supports both of these conclusions. The
announcement of this study was short in its duration (one month) and focused on
reaching a vast number of individuals. McDermott et al. (2003) similarly concluded that
recruitment advertisements should be posted at more sites over shorter lengths of time.
Each email acted as its own site, just as the more usual sites of campus billboards and
classroom announcements were there own sites. Specifically, the advertisement posted on
a website acted similar to an ad posted on a campus billboard. Take for example each
advertisement posted on an internet webpage (i.e. my.ufl.edu) or on campus billboards,
both may have the potential for many to be in the presence of the announcement. In other
words, they are more passively there for anyone to see. Accordingly, each produced a
similar number of respondents (12 from ads posted on campus, 17 from ads posted on
my.ufl.edu). The difference is that the website based posting was one ad, whereas the
campus based postings were about 50 ads. Conversely, an in class announcement or
email is proactively forced upon an individual. The difference to consider is the total
population available to be actively advertised to. Classes are limited to the number of
students enrolled or the number of students in as many classes that announcements can be
made in. Emails, however, are limited to only the number of people a email is sent to.
Each classroom, regardless of the number of students in it becomes one singular site with
x number of potential participants. Each email becomes one site with one potentially
potential participant (possibly more if he or she talks to others about the opportunity).
This closely follows McDermott's (2003) suggestion of increasing the number of places
that each advertisement is made. The recruitment data from this study, while not
statistically significant, does practically support McDermott by the number of
participants recruited via emails (n = 229) being almost 3.5 times larger than the number
returned from in-class announcements (n = 67).
A more contemporary view of social evolution is also evident by comparing the
recommendations of the literature to the findings of this study (Kraut et al., 2004; Myers
et al., 2004). In 1985 only 8.2% of the US population had a computer. By 2001 that
number increased to 66% of the US population having access to a computer and 56% of
Americans having access to the internet (Kraut et al., 2004). Clearly this technology is
expanding exponentially and simultaneously changing the delivery of information to
most Americans. As evident by the results of the breadth of alternative methods of
recruitment in this study, this increase in computer and internet usage is a growing
resource to contact and intervene on populations. Simply, remaining with the status quo
of sampling convenient populations such as college classes, may not be keeping with the
Much attention has been given to the fact that eating disorders generally develop
very quietly (Fairburn & Cooper, 1982; Hoek, 1995) and at a young age (Jacobi et al.,
2004; Stice, 2002), thus making intervention difficult. Cooper (1995) outlined this by
elucidating the stages of development, beginning with behavioral precursors,
transitioning into risk factors, and finally progressing onto full syndrome disorder.
Examining new methods of effectively recruiting potentially at-risk populations is
analogous to finding new ways of intervening at Cooper's (1995) first stage of
development. In other words, sampling a population, such as college women, that is
known to show behavioral precursors (Cassell, 1994; Franko et al., 2005; Tylka &
Subich, 2002; Tylka, 2004) allows the opportunity to access and halt the development of
precursors of risk that the individual may or may not be aware of. Alternative
technologies, such as the internet, offer the possibility of reaching far more people than
previously available, yet little attention has focused on exploiting the advantages of them
(Kraut et al., 2004; Myers et al., 2004).
The secondary purpose of my thesis was to examine the relationship between
exercise, exercise dependence, and drive for thinness. Researchers have recommended
that a mediating and/or moderating relationship of such variables be examined
(Hausenblas & Fallon, 2002). Both mediators and moderators are types of variables that
may affect the association between two other variables. Both variables have been
confused, mistakenly referred to, and wrongly identified (Baron & Kenny, 1986;
Kraemer et al., 2001). To help clarify, I will first describe both mediators and moderators,
then continue by using the variables of interest in this thesis as examples to discuss the
difference of the two and why a mediator effect, but not a moderator effect, was
A mediator is any variable that explains the how or why a relationship exists, while
a moderator is any variable that specifies on whom or under which conditions (when) a
relationship exists (Baron & Kenny, 1986; Bennet, 2000; Kraemer et al., 2001). A
potentially mediating variable (exercise dependence) must be correlated to the
independent variable (exercise) and a potentially moderating variable (exercise
dependence) must not be correlated to an independent (exercise) or dependent variable
(drive for thinness; See figures 5-1 and 5-2.). Put another way, a mediator is a variable
that is present and correlated to the independent variable, which helps to explain how or
why the independent variable affects the dependent variable. A moderator is a different
variable, uncorrelated to the independent or dependent variable, that when present
explains to whom or when a relationship occurs.
Variable I Variable
Figure 5-1 Moderator Model
independent -. Mediator .. Outcome
Variable Variable Variable
Figure 5-2 Mediator Model
In my study, the relationship between exercise, as an independent variable, and
drive for thinness, as an outcome or dependent variable, was in question. The relationship
between exercise and exercise dependence in this thesis was correlated. This correlation
is similar to previous research (Zmijewski & Howard, 2003). Because of the correlation
between these two variables, examining a possible mediator effect of exercise
dependence on the relationship between exercise and drive for thinness was the
appropriate analysis. This method specifies that three regressions must meet the
following conditions to show a mediator effect (Baron & Kenny, 1986):
4. Variations in the independent variable significantly account for variations in the
mediator variable (figure 5-3, path a),
5. Variations in the mediator variable significantly account for the dependent variable
(figure5-3, path b),
6. When both path a and b are controlled, the significant relationship observed in
independent and dependent variables disappears. (figure 5-3, path c).
Figure 5-3 illustrates this conceptually, while figure 5-4 illustrates this with my thesis's
variables and results.
^a diator Variable b
Indepndene c________ __ Dependent (Outcome)
Figure 5-3 Conceptual Model of Mediation.
Firstreg ssion irdrel ssian
P = .440 = .001 EDS P= .180,P=.005
Secoil re ge ssion
S= .149,p =.008
Third re re sin
= J59,p= .354
Figure 5-4 Conceptual Model of Mediation with Study Variables
These results build upon the recommendations of previous research by examining a
potential mediating or moderating effect of exercise dependence on eating pathology
(Hausenblas & Fallon, 2002). Furthermore, they improve upon research examining the
relationship of the same variables (Zmijewski & Howard, 2003). These results may also
help clarify the confused relationship of exercise with eating disorders by demonstrating
evidence of a mediating affect of exercise dependence as a means in which exercise
becomes problematic for eating disordered individuals. Specifically, a mediator is a
variable that explains how or why another variable affects the outcome (Baron & Kenny,
1986; Kraemer et al., 2001). The basic question in my thesis then is, how does exercise
affect eating pathology? The potential mediator, exercise dependence, is examined to see
if its presence will affect the outcome of exercise on eating pathology. This is shown
statistically by the three paths described above. Simply stated, there are two causal paths
on the dependent variable (Baron & Kenny, 1986). Path c represents the direct impact of
the exercise on driver for thinness, and path b represents the impact of the exercise
dependence also on drive for thinness. Separately, both of these paths show that exercise
and exercise dependence affect drive for thinness. When the mediating variable paths are
controlled, the significant relationship between exercise and drive for thinness is no
longer significant. Therefore, answer to the question, how does exercise affect eating
pathology?, is answered by these analyses showing that the affect of exercise on drive for
thinness is mediated by exercise dependence.
Several limitations were present in my study. The most obvious limitation is the
capabilities of alternative technologies. For example, many participants took the liberty
of reporting qualitative data in inappropriate areas, specifically when reporting exercise
amounts. While this data was useful in allowing for a better inference of what the
participants were reporting, it points out that computer based technologies are very exact.
In other words, they will only perform in the very specific manner as programmed.
Future research involving the use of websites or computer based interactions with
participants should be very careful to program places for participants to enter data in a
way that does not allow for confused reporting. For example, amount of exercise should
only been allowed to be entered as numeric data, not as alphanumeric data.
A similar limitation of alternative technologies is how many people are actually
using them. In other words, does recruiting via alternative methods, such as email and
internet websites, accurately capture the targeted population? The most recent U.S.
Census report on computer and internet usage (Cheeseman Day, Janus, & Davis, 2005)
states that in 2003, 47.1% of all 15-24 year old Americans had internet access in their
home. Furthermore, 89.4% of all 18-24 year olds enrolled in school were using
computers, while 70.6% of all 18-24 year olds enrolled in school were using the internet.
These data support that most of the 10-30% of college age women whom are at risk of
developing eating disorders (Cassell, 1994; Franko et al., 2005) can be reached by
Several researchers have reported that women overestimate the amount of total
exercise duration (Buchowski et al.,1999; Irwin et al., 2001; & Jakicic et al., 1998).
Specifically, this overestimation has been found in women with higher percentage of
body fat. The BMI of the sample in my thesis was in the overweight range (ACSM,
2000). Therefore, using self-report measures of exercise behavior is another limitation of
my thesis. Similarly, using other self-report measures for eating disordered behavior was
also a limitation of my thesis.
A limitation that was potentially present and somewhat related to the first limitation
is the look and feel of a website. While it was not reported back to me that the drab gray
color and simplistic design and layout of the questions was a problem for participants,
some participants did have trouble initially accessing the website via the PIN login
screen. This may or may not have been because of the onscreen layout. What was a
noticeable limitation of this website's layout was the inability to format instructions and
questions correctly. For example, the instructions for the LTEQ and the first question in
this scale were run together. These instructions are rather lengthy and can be hard on the
eyes when viewed onscreen. Moreover, the length of the website may have been a
problem. Allowing for screen breaks, font changes, deliberate spacing, or even just
placing pictures or changing background colors may have helped with the website's
A fourth limitation was time to respond to the large number of interested
individuals contacting me for access to this study. The announcement made via email to
all University of Florida students resulted in a deluge of inquiries in my inbox
immediately following the transmission of the email announcement. Several interested
individuals even email multiple times a day until they received a PIN. This perhaps
speaks to the previously discussed changing nature of our society and the role of the
internet in it. People are increasingly becoming accustomed to instant responses and
interactions. Allowing a day or two to respond to an email may have been acceptable
etiquette in the past, but the increasingly fast paced and quick response nature of
interactive technologies, such as the internet, is slowly conditioning people to expect
more sooner. Potential solutions to this problem will be discussed in the future directions
section of this thesis.
A fifth limitation pertains to the findings of the mediator effect of exercise
dependence. These data came from participants whom were recruited via advertisements
for an exercise program for women dissatisfied with their bodies. This population of
college females may not be truly representative of a general population in regards to
exercise behavior, exercise dependence, and/or drive for thinness scores. Specifically,
past research has identified the onset of development of eating disorder pathology at
earlier ages than of this sample (Jacobi et al., 2004; Stice, 2002; Thomsen, Weber &
Brown, 2001; 2002). Any interpretation of these data should be mindful and considerate
of the population from which this data was collected.
Several adjustments are needed in any future research continuing to build upon
these data. First, any future research using a website based method of data collection
should be more aesthetically pleasing and user friendly than the one used in this thesis.
Future studies measuring participants on more scales or questionnaires than the ones use
in this thesis must be particularly concerned with this and consider how such
improvements in usability may ease the burden of participation. Standards for protecting
and securely storing computerized data must be kept, but not at the expense of usability
and convenience for the participant.
Time was also discussed as a limitation of this study. Any future directions of
similar research should use expert systems to help ease the management of data
collection and standardized responses sent to participants. An expert system is defined as
any software based program that can mimic the deductive or inductive reasoning of a
human expert (Negotia, 1985). These systems can use algorithms to make decisions
based on entered data and be quite sophisticated. Something as simple as the ability to
review data and send and automated reply to participants would be of extreme
convenience to both the researcher and the participant. For example, in this thesis having
a simple expert system that could have reviewed data collected from the website and
made a decision as to which email to send a participant (either informing them of there
potential eligible status for future studies or not) would have been paramount in time and
A third future direction involves the email method of announcing the study.
Everyday our email inboxes are bombarded with unsolicited advertisements and junk
mail commonly referred to as spam. Not only are these unwanted emails annoying, but
many may carry the risk of infecting and crippling your computer from attached viruses.
The email for this study was sent through a listserve, or list of email addresses
encompassing all University of Florida undergraduates, designed to inform the student
body of a variety of events on campus. The response rate was substantially greater than
any other method of recruitment measure, but was still low considering that it was sent to
thousands of students. Emails such as these, even though they are sent from familiar and
reputable sources, may be discarded along with any other junk mail of the day. A future
direction to help alleviate this problem is to examine what factors would lead to students
taking better advantage of emails, such as the one announcing this study, that inform
them of opportunities at the university rather than deleting them without reading them.
Creating a listserve solely announcing research opportunities may be a way to reach only
those students whom are willing to participate in future studies.
Future research is also needed to further examine the mediator relationship that was
found in this thesis. Efforts should focus on recruiting a more representative cross section
of the population. Future research should also examine these relationships in samples of
men. Similarly, announcements should not focus solely on recruiting those whom are
currently dissatisfied with their body. While this was acceptable for the purposes of this
thesis, future research should focus on recruiting a sample that is more representative of
the general population.
Finally, a limitation discussed above was that this study used self-report measures
of exercise behavior and eating disorder risk. Future studies should use more objective
methods of collecting similar data. For example, accelerometers could be used to collect
more accurate data on exercise behavior and clinical interviews could be used to more
accurately determine eating disorder risk status.
This study illustrated the importance of using new technologies and alternative
methods to recruit a large number of individuals in a short time, thus supporting
recommendations researchers (McDermont et al., 2003) and emphasizing advantages of
computer based research (Kraut et al., 2004; Myers et al., 2004). Using such methods was
not shown to be statistically different than more traditional methods of announcing
research, but did show that there is potential for reaching large numbers of people while
being convenient for both the researcher and the participant. This study also improved
upon previous research finding that exercise dependence is correlated with bulimia
subscale scores in female undergraduate college students (Zmijewski & Howard, 2003)
by further examining the relationship between exercise and eating pathology based on the
recommendation to examine moderator and mediator effects of these variables
(Hausenblas & Fallon, 2002). Eating disorders are serious and secretive both in their
development and their course, (Cassell, 1994; Fairburn, 1995; Hoek, 1995; & Hoek &
van Hoeken, 2003) thus necessitating the need for further research to identify better ways
of recruiting this special population.
What is your age?
Please type how many feet tall you are.
What is your current weight?
What is your ideal weight? (The weight you would like to be at right now.)
What year in school are you?
What is your major?
What is your family's annual income?
What is your ethnic background?
If you selected "other" ethnic background, please type your ethnic background here.
What is your first name?
What is your email address? (Please type an email account that you check regularly.)
What is your phone number?
How many times a week do you exercise?
Have you ever been diagnosed with an eating disorder?
Do you currently have an eating disorder?
What is your gender?
Where did you hear about this study?
DRIVE FOR THINNESS QUESTIONNAIRE
Instructions: This is a scale which measures a variety of attitudes, feelings, and
behaviors. Read each question carefully and circle the number to the right of the
question which applied best to you.
Please use the following scale:
1. I eat sweets and carbohydrates without feeling nervous. 1 2 3 4 5 6
2. If I gain a pound, I worry that I will keep gaining. 1 2 3 4 5 6
3. I think about dieting. 1 2 3 4 5 6
4. I am terrified of gaining weight. 1 2 3 4 5 6
5. I exaggerate or magnify the importance of weight. 1 2 3 4 5 6
6. I am preoccupied with the desire to be thinner. 1 2 3 4 5 6
7. I feel extremely guilty after overeating. 1 2 3 4 5 6
EXERCISE DEPENDENCE QUESTIONNAIRE
Instructions. Using the scale provided below, please complete the following
questions as honestly as possible. The questions refer to current exercise beliefs and
behaviors that have occurred in the past 3 months. Please place your answers in the
blank space provided after each statement.
1 2 3 4 5 6
1. I exercise to avoid feeling irritable.
2. I exercise despite recurring physical problems.
3. I continually increase my exercise intensity to achieve the desired effects/benefits.
4. I am unable to reduce how long I exercise.
5. I would rather exercise than spend time with my family/friends.
6. I spend a lot of time exercising.
7. I exercise longer than I intend.
8. I exercise to avoid feeling anxious.
9. I exercise when injured.
10. I continually increase my exercise frequency to achieve the desired
11. I am unable to reduce how often I exercise.
12. I think about exercise when I should be concentrating on school/work.
13. I spend most of my free time exercising.
14. I exercise longer than I expect.
15. I exercise to avoid feeling tense.
16. I exercise despite persistent physical problems.
17. I continually increase my exercise duration to achieve the desired
18. I am unable to reduce how intense I exercise.
19. I choose to exercise so I can get out of spending time with family/friends.
20. A great deal of my time is spent exercising.
21. I exercise longer than I plan.
LEISURE TIME EXERCISE QUESTIONNAIRE
Instructions. This is a scale which measures your leisure-time exercise (i.e.,
exercise that was done during your free time such as intramural sports-NOT your
physical education class). Considering a typical week, please indicate how often
(on average) you have engaged in strenuous, moderate, and mild exercise more that
20 minutes during your free time?
1. Strenuous exercise: heart beats rapidly (running, basketball, jogging, hockey, squash,
judo, roller skating, vigorous swimming, vigorous long distance bicycling, vigorous
aerobic dance classes, heavy weight training)
How many times per typical week do you perform strenuous exercise for 20 minutes or
2. Moderate exercise: not exhausting, light sweating (fast walking, baseball, tennis, easy
bicycling, volleyball, badminton, easy swimming, popular and folk dancing)
How many times per typical week do you perform moderate exercise for 20 minutes or
3. Mild exercise: minimal effort, no sweating (easy walking, yoga, archery, fishing,
bowling, lawn bowling, shuffleboard, horseshoes, golf)
How many times per typical week do you perform mild exercise for 20 minutes or
Are you dissatisfied with your
body? Would you like to get into
Are you dissatisfied with your body? Would you like to get into better shape? The
exercise psychology lab in the Department of Applied Physiology and Kinesiology
is looking for female students who are interested in participating in an exercise
program designed to improve your physical appearance and body image. Please
contact the exercise psych lab at email@example.com to complete a short
web based questionnaire to determine your study eligibility.
REPLY WITH PIN AND INFORMED CONSENT
Thank you very much for your interest in this study! Please visit the following
website to complete our eligibility questionnaire,
http://survey.hhp.ufl.edu/index.php?id=65 Your PIN login in for this study is
XXXXXX. Enter this PIN when prompted and click on the "Begin the Survey" box.
Simply pressing enter after entering your PIN may not work. Please complete all
of the survey questions. This survey is only accessible through the unique
password that has been emailed to you and cannot be used more than once. Do
not give your PIN to anyone else to use. Please complete the entire survey in
one session. If you need to stop or are disconnected before you have completed
the survey, email the exercise psychology lab at firstname.lastname@example.org and
we will respond with a new PIN for you to use.
A consent form explaining this study and whom to contact if you have any
questions is provided below. Please review this form. Please reply to this email if
you decide not to participate after reading this form.
Thank you again for your interest in this study,
The exercise psychology lab.
PLEASE READ THIS ENTIRE DOCUMENT CAREFULLY BEFORE YOU DECIDE TO
PARTICIPATE IN THIS STUDY.
TO: All Research Participants
FROM: Brian Cook
RE: Informed Consent
PURPOSE OF THIS STATEMENT: The purpose of this statement is to summarize the study I
am conducting, explain what I am asking you to do, and to assure you that the information you
and other participants share will be kept confidential to the extent permitted by law. Specifically,
nobody besides the principal investigator and the research assistants will be able to identify you
in this study and your name will not be used in any research reports that result from this project.
WHAT YOU WILL BE ASKED TO DO: If you agree to participate in this study, you will be
asked to complete a few questionnaires via the web. Based on your responses to these measures,
you may be contacted by a research assistant to participate in an exercise intervention.
TIME RQUIRED: Completion of the measures will take no more than 10 minutes.
RISKS AND BENEFITS: There are no risks expected from participating in this study. There are
no direct benefits to you for participating in this study.
COMPENSATION: You will not be compensated monetarily for your participation in this
study. You may be given extra credit by your instructor for completing this study.
CONFIDENTIALITY: Your identity will be kept confidential to the extent provided by law.
Your information will be assigned a code number. The list connecting your name to this number
will be kept in my laboratory (Room 145 Florida Gym) in a locked file cabinet. When the study is
complete and the data have been analyzed all materials will be destroyed. Your name will not be
used in any report.
VOLUNTARY PARTICIPATION: Your participation in this study is completely voluntary.
There is no penalty for not participating.
RIGHT TO WITHDRAW: You have the right to withdraw from the study at anytime without
WHOM TO CONTACT IF YOU HAVE QUESTIONS ABOUT THIS STUDY: Brian Cook,
Department of Applied Physiology and Kinesiology. Email: email@example.com. Heather
Hausenblas, Department of Applied Physiology and Kinesiology. Email: firstname.lastname@example.org.
WHOM TO CONTACT ABOUT YOUR RIGHTS AS A RESEARCH PARTICIPANT IN
THE STUDY: UFIRB Office, Box 112250, University of Florida, Gainesville, FL 32611-2250;
AGREEMENT: I have read the procedure described above. I voluntarily agree to participate in
the procedure. By clicking here I am giving my consent to participate in this study.
Dear (insert first name here) ,
Recently you completed a web-based screening questionnaire for an upcoming
exercise program designed to improve your physical appearance and body image. Based
on your responses, you have met the preliminary criteria for inclusion in this program.
The program is scheduled to begin at the start of the spring 2006 semester. Please
reply to this email and let us know:
1.) If you are still interested in this program.
2.) If you will be able to complete the 6 month exercise program, as well as a 6
month follow up.
3.) What times) will be the best for you to participate in this exercise program?
Specify if mornings, days, and/or evenings will work best for you. Please
consider your class and academic responsibilities, work responsibilities,
family obligations, and any other time restraints that you will have next
Prior to the beginning of this program, everyone will be randomized into either the
intervention or the control group. Everyone has an equal chance of being in either group.
You will receive notification of which group you are a part of. Thank you for your
interest and participation in this study.
The Exercise Psychology Laboratory
Dear (insert first name here) ,
Recently you completed a web-based screening questionnaire for an upcoming
exercise program designed to improve your physical appearance and body image.
Unfortunately, based on your responses, you have not met the preliminary criteria for
inclusion in this program. We may contact you in the future if other similar programs
begin that may be better suited for you. Please reply to this email if you would not like to
be contacted for any future studies. Thank you for your interest and participation in this
The Exercise Psychology Laboratory
LIST OF REFERENCES
American College of Sports Medicine. (2000). ACSM's guidelinesfor exercise testing
and prescription (6th ed.) Baltimore: Williams & Wilkins.
American Psychological Association. (1994). Diagnostic and statistical manual of mental
disorders (4th ed.). Washington, DC.: Author
Austin, S.B. (2000). Prevention research in eating disorders: Theory and new directions.
Psychological Medicine, 30, 1249-1262.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in
social psychology research: Conceptual, strategic, and statistical considerations.
Journal ofPersonality and Social Psychology, 51(6), 1173-1182.
Battle, E.K., & Brownell, K.D. (1996). Confronting a rising tide of eating disorders and
obesity: treatment versus prevention and policy. Addictive Behaviors, 21, 755-765.
Bennett, J.A. (2000). Mediator and moderator variables in nursing research: Conceptual
and statistical differences. Research in Nursing and Health, 23, 415-420.
Brewerton, T., Stellefson, E., Hibbs, N., Hodges, E., & Cochrane, C. (1995). Comparison
of eating disorder patients with and without compulsive exercising. International
Journal ofEating Disorders, 17(4), 413-416.
Buchowski, M.S., Townsend, K.Y., Chen., Acra, S.A., & Sun, M. (1999). Energy
expenditure determined by self-reported physical activity is related to body fatness.
Obesity Research, 7, 23-33.
Caplan, G. (1964). Principles ofpreventive psychiatry. New York : Basic Books
Cassell, Dana K. (1994). Encyclopedia of obesity and eating disorders. New York New
York: Facts on File, Inc.
Catalano, J.A. & Dooley, D. (1980). Economic change in primary prevention. In R.H.
Price, R.F. Ketter, B.C. Bader, and J. Monahan (Eds.), Prevention in mental health:
Research, policy, and practice. (pp. 21-40). Beverly Hills, CA : Sage Publications
Center for Disease Control, (1996). Guidelines for school health programs to promote
lifelong healthy eating. Morbidity and Mortality Weekly Report, 45(rr9), 1-41.
Cheeseman Day, Janus, & Davis, (2005). Computer and internet use in the United States:
2003. Current Population Reports, U.S. Department of Commerce, U.S. Census
Bureau, October 2005.
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, New York: Guilford Publications, Inc.
Davis, C., Kaptein, S., Allan, M., Olmsted, M., Woodside, D. (1998). Obsessionality in
anorexia nervosa: The moderating influence of exercise. Psychosomatic Medicine,
Davis, C., Katzman, D., & Kirsh, C. (1999b). Compulsive physical activity in adolescents
with anorexia nervosa. A psychobehavioral spiral of pathology. Journal of Nervous
and Mental Disorders, 187, 336-342.
Davis, C., Katzman, D., Kaptein, S., Brewer, H., Kalmbach, K., Olmsted, M., Woodside,
D., & Kaplan, A. (1997). The prevalence of high-level exercise in the eating
disorders: Etiological implications. Comprehensive Psychiatry, 38(6), 321-326.
Davis, C., Woodside, D.B., Olmsted, M.P., & Kaptein, S. (1999a). Psychopathology in
the eating disorders: the influence of physical activity. Journal ofApplied
Biobehavioral Research, 4(2), 139-156.
Fairbum, C.G. (1995). The prevention of eating disorders. In K.D. Brownell & C.G.
Fairbum (Eds.), Eating disorders and obesity: A comprehensive handbook (pp.
289-293). New York, New York: Guilford Publications, Inc.
Fairbum, C.G. & Cooper, P.J. (1982). Self-induced vomiting and bulimia nervosa: An
undetected problem. British Medical Journal, 284, 1153-1155.
Fairbum, C.G., & Walsh, B.T. (1995). Atypical eating disorders. In K.D. Brownell &
C.G. Fairbum (Eds.), Eating disorders and obesity: A comprehensive handbook
(pp. 135-140). New York, New York: Guilford Publications, Inc.
Franko, D.L., Mintz, L.B., Villapiano, M., Green, T.C., Mainelli, D., Folensbee, L.,
Butler, S., Davidson, M.M., Hamilton, E., Little, D., Kearns, M., & Budman, S.
(2005). Food, mood, and attitude: Reducing risk for eating disorders in college
women. Health Psychology, 24(6), 567-578.
Garfinkel, P.E. (1995). Classification and diagnosis of eating disorders. In K.D. Brownell
& C.G. Fairburn (Eds.), Eating disorders and obesity: A comprehensive handbook
(pp. 125-134). New York, New York: Guilford Publications, Inc.
Garner, D.M. (1991). The Eating Disorder Inventory-2 professional manual. Odessa, FL:
Psychological Assessment Resources.
Garner, D.M. & Garfinkel, P.E. (1979). The eating attitudes test: An index of the
symptoms of anorexia nerviosa. Psychological Medicine, 9, 273-279.
Glasgow, R.E., Klesges, L.M ., Dzaltowski, D.A., Bull, S.S., & Eastabrooks, P. (2004).
The future of health behavior change research: What is needed to improve
translation of research into health promotion practice? Annals of Behavioral
Medicine, 27(1), 3-12.
Glasgow, R.E., Vogt, T.M., & Boles, S.M. (1999). Evaluating the public health impact of
health promotion interventions: The RE-AIM framework. American Journal of
Public Health, 89, 1322-1327.
Godin, G., & Shephard, R.J. (1985). A simple method to assess exercise behavior in the
community. Canadian Journal ofApplied Sports Science, 10, 141-146.
Harris, E.C., & Barraclough, B. (1994). Suicide as an outcome predictor for mental
disorders: A meta-analysis. British Journal ofPsychiatry, 170, 205-228.
Harris, E.C., & Barraclough, B. (1998). Excess mortality of mental disorders. British
Journal of Psychiatry, 173, 11-53.
Hausenbals, H.A. & Fallon, E. A. (2002). Relationship among body image, exercise
behavior, and exercise dependence symptoms. International Journal ofEating
Disorders, 32, 179-185.
Hausenblas, H.A., & SymonsDowns, D. (2002). How much is too much? The
development and validation of the exercise dependence scale. Psychology &
Health, 17(4), 387-404.
Herzog, D. B., Keller, M.B., Strober, M., Yeh, C., & Pai, S-Y. (1992) The current status
of treatment for anorexia nervosa and bulimia nervosa. International Journal of
Eating Disorders, 12, 215-220.
Hoek, H.W. (1995) The distribution of eating disorders. In K.D. Brownell & C.G.
Fairbum (Eds.), Eating disorders and obesity: A comprehensive handbook (pp.
207-211). New York, New York: Guilford Publications, Inc.
Hoek, H.W., & van Hoeken, D. (2003). Review of the prevalence and incidence of eating
disorders. International Journal ofEating Disorders, 34, 383-396.
Hoglund, K., & Normen, L. (2002). A high exercise load is linked to pathological weight
control behavior and eating disorders in female fitness instructors. Scandinavian
Journal of Medicine & Science in Sports, 12, 261-275.
Irwin, M.L., Ainsworth, B.E., & Conway, J.M. (2001). Estimation of energy expenditure
from physical activity measures: Determinants of accuracy. Obesity Research, 9(9),
Jacobi, C., Hayward, C., de Zwaan, M., Kraemer, H.C., & Agras, W.S. (2004). Coming
to terms with risk factors for eating disorders: Application of risk terminology and
suggestions for a general taxonomy. Psychological Bulletin, 130(1), 19-65.
Jacobs, D.R., Ainsworth, B.E., Hartman, T.J., & Leon, A.S. (1993). A simultaneous
evaluation of ten commonly used physical activity questionnaires. Medicine &
Science in Sports & Exercise, 25, 81-91.
Jakicic, J.M, Poley, B.A., & Wing, R.R. (1998). Accuracy of self-reported exercise and
the relationship with weight loss in overweight women. Medicine and Science in
Sports and Exercise, 30 (4), 634-638.
Katz, J.L. (1996). Clinical observations on the physical activity of anorexia nervosa. In
Epling, W.F. & Pierce, W.D. (Eds.), Activity anorexia: Theory, research, and
treatment. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
Keel, P.K., Dorer, D.J., Eddy, K.T., Franko, D.F., Charatan, D.L., & Herzog, D.B.
(2003). Predictors of mortality in eating disorders. Archives of General Psychiatry,
Killen, J.D., Taylor, C.B., Hammer, L.D., Litt, I., Wilson, D.M., Rich, T., Hayward, C.,
Simmonds, B., Kraemer, H., & Varady, A. (1993). An attempt to modify unhealthy
eating attitudes and weight regulation practices of young adolescent girls.
International Journal of Eating Disorders, 13, 369-384.
Kjelsis, E., Bjmrnstrom, C., & Gunnar Gotestam, K. (2004). Prevalence of eating
disorders in female and male adolescents (14-15 years). Eating Behaviors, 5, 13-25.
Kraemer, H.C., Stice, E., Kazdin, A., Offord, D., & Kupfer, D. (2001). How do risk
factors work together? Mediators, moderators, and independent, overlapping, and
proxy risk factors. American Journal ofPsychiatry, 158, 848-856.
Kraut, R., Olson, J., Banaji, M., Bruckman, A., Cohen, J., Couper, M. (2004).
Psychological research online. Report of board of scientific affairs' advisory group
on the conduct of research on the internet. American Psychologist, 59(2), 105-117.
Levine, M., & Piran, N. (2004). The role of body image in the prevention of eating
disorders. Body Image, 1, 57-70.
Lovato, L., Hill, K., Hertert. S., Hunninghake, D., & Probstfield, J. (1997). Recruitment
for controlled clinical trials: Literature summary and annotated bibliography.
Controlled Clinical Trials, 18, 328-357.
McDermott, C., Agras, W., Crow, S., Halmi, K., Mitchell, J., & Bryson, S. (2003).
Participant recruitment for an anorexia nervosa treatment study. International
Journal ofEating Disorders, 35, 33-41.
Mehlenbeck, R. (2002). Eating disorders: An overview. Symposium conducted at the
meeting of the New England Chapter of Society for Public Health Education,
Myers, T., Swan-Kremeier, L., Wonderlich, S., Lancaster, K., & Mitchell, J. (2004). The
use of alternative delivery systems and new technologies in the treatment of
patients with eating disorders. International Journal ofEating Disorders, 36, 123-
Negotia, U.N. (1985). Expert systems and fuzzy systems. Menlo Park, CA:
Neumark-Sztainer, D. (1996). School-based program for preventing eating disturbances.
Journal of School Health, 66, 64-71.
Nosek, B.A., Banaji, M.R., & Greenwald, A.G. (2002). Harvesting implicit group
attitudes and benefits from a demonstration website. Group Dynamics, 6(1), 101-
Paul, T., Schroeter, K., Dahme, B., & Nutzinger, D.O. (2002). Self-injurious behavior in
women with eating disorders. American Journal ofPsychiatry, 157, 886-895.
Pearson, J., Goldklang, D., & Streigel-Moore, R. (2002). Prevention of eating disorders:
Challenges and opportunities. International Journal ofEating Disorders, 31, 233-
Pritts, S.D., & Susman, J. (2003). Diagnosis of eating disorders in primary care.
American Family Physician, 67, 297-304.
Prochaska, J.J., Salis, J.F., & Long, B. (2001). A physical activity screening measure for
use with adolescents in primary care. Archives ofPediatric and Adolescence
Medicine, 155, 554-559.
Rivest, S., & Richard, D. (1990). Involvement of corticotropin-releasing factor in the
anorexia induced by exercise. Brain Research Bulletin, 25(1), 169-172.
Schechtman, K.B., Barzilai, B., Rost, K., & Fisher, E.B. (1991). Measuring physical
activity with a single question. American Journal of Public Health, 81, 771-773.
Seidell, J.C. (1999). Obesity: a growing problem. ActaPaediatrica, 88(suppl. 428), 46-
Smolak, L., Levine, M.P., & Schermer, F. (1998). A controlled evaluation of an
elementary school primary prevention program for eating problems. Journal of
Psychosomatic Research, 44, 339-353.
Sobel, S.V. (2004) Eating disorders. Continuing Medical Education Resource, 118(10),
Stice, E. (2002). Risk and maintenance factors for eating pathology: A meta-analytic
review. Psychological Bulletin, 128(5), 825-848.
Stice, E., & Shaw, H. (2004). Eating Disorder Prevention Programs: A Meta-Analytic
Review. Psychological Bulletin, 130(2), 206-227.
Symons Downs, D., Hausenblas, H., & Nigg, C. (2004). Factorial validity and
psychometric examination of the exercise dependence scale- revised. Measurement
in Physical Education and Exercise Science, 84(4), 183-201.
Thomsen, S.R., Weber, M.M., & Brown, L.B. (2001). The relationship between health
and fitness magazine reading and eating-disordered weight-loss methods among
high school girls. American Journal of Health Education, 32(3), 133-38.
Thomsen, S.R., Weber, M.M., & Brown, L.B. (2002). The relationship between reading
beauty and fashion magazines and the use of pathogenic dieting methods among
adolescent females. Adolescence, 37(145), 1-18.
Tylka, T.L. (2004). The relationship between body dissatisfaction and eating disorder
symptomatology: an analysis of moderating variables. Journal of Counseling
Psychology, 51(2), 178-191.
Tylka, T.L., & Subich, L.M. (2002). Exploring young women's perceptions of the
effectiveness and safety of maladaptive weight control techniques. Journal of
Counseling and Development, 80(1), 101-111.
US Department of Health and Human Services (1990). Healthy people 2000: National
health promotion and disease prevention objectives. Washington, D.C.: USDHHS
US Department of Health and Human Services (1996). Physical activity and health: A
report of the surgeon general. Atlanta, Georgia: USDHHS Center for Chronic
Disease Prevention and Health Promotion
Vitousek, K.B., 1995 Cognitive-behavioral therapy for anorexia nervosa. In K.D.
Brownell & C.G. Fairbum (Eds.), Eating disorders and obesity: A comprehensive
handbook (pp. 324-329). New York, New York: Guilford Publications, Inc.
Wantland, D., Portillo, C., Holzemer, W., Slaughter, R., & McGhee, E. (2004). The
effectiveness of web-based vs. non-web-based interventions: A meta-analysis of
behavioral change outcomes. Journal of Medical Internet Research, 6(4), e40.
Weiss, T.W., Slater, C.H., Green, L.W., Kennedy, V.C., Albright, D.L., & Wun, C.C.
(1990). The validity of single-item self-assessment questions as measures of adult
physical activity. Journal of Clinical Epidemiology, 43, 1123-1129.
Wellman, B., Quan Haase, A., Witte, J., & Hampton, K. (2001). Does the Internet
increase, decrease, or supplement social capital? Social networks, participation, and
community commitment. American Behavioral Scientist, 45(3), 436-455.
Woodside, D.B., Garfinkel, P.E., Lin, E., Goering, P., Kaplan, A.S., Goldbloom, D.S., &
Kennedy, S.H. (2001). Comparison of men with full or partial eating disorders,
men with out eating disorders, and women with eating disorders in the community.
American Journal ofPsychiatry, 158, 570-574.
Zmijewski, C.F. & Howard, M.O. (2003). Exercise dependence and attitudes toward
eating among young adults. Eating Behaviors, 4, 181-195.
Brian was raised in Cumberland, Rhode Island, and attended high school at Mount
Saint Charles Academy in Woonsocket, Rhode Island. He began his undergraduate
education at Johnson State College in Johnson, Vermont, where he majored in
psychology and competed on the men's varsity cross country and tennis teams. At the
conclusion of his freshman year, he transferred to the University of Rhode Island to focus
on his psychology major. After graduating from the University of Rhode Island, he
worked as a project assistant at ProChange Behavior Systems, Inc., in West Kingston,
Rhode Island. Brian enrolled at the University of Florida in the fall semester of 2003.