Improving the behavioral treatment of obesity in adults


Material Information

Improving the behavioral treatment of obesity in adults
Physical Description:
ix, 105 leaves : ill. ; 29 cm.
Fuller, Pamela Ruth, 1966-
Publication Date:


Subjects / Keywords:
Obesity -- psychology   ( mesh )
Obesity -- Adult   ( mesh )
Obesity -- therapy   ( mesh )
Behavior Therapy -- Adult   ( mesh )
Weight Loss -- Adult   ( mesh )
Patient Education   ( mesh )
Diabetes Mellitus, Type II -- Adult   ( mesh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1994.
Includes bibliographical references (leaves 94-104).
Statement of Responsibility:
by Pamela Ruth Fuller.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 50529869
System ID:

Full Text








I would like to express my sincere appreciation to my

chairman, Dr. Michael Perri, for the significant amount of

time, expertise, and support that he gave me during this

project. I also want to thank the other members of my

doctoral committee, Dr. James Johnson, Dr. Suzanne Johnson,

Dr. Maria Grant, and Dr. Dan Martin for their individual

contributions. I extend my appreciation to all the

weight-loss therapists, especially Timothy Pearman, Elizabeth

Leermakers, and Cheryl Porter, who put significant time and

effort into the project. I am grateful to Dr. Laura Guyer and

her graduate students for their extensive clinical and

research support. Finally, I am grateful for the continuous

support and encouragement that I have received from my husband

during graduate school and from my parents throughout my

academic career.


ACKNOWLEDGEMENTS....................................... ii

LIST OF TABLES........................................ v

LIST OF FIGURES.................................... .... vi

ABSTRACT.......... ..... ...... .............. ......... vii


I INTRODUCTION .............................. 1

Benefits of Weight-Loss Interventions........ 3
Behavioral Weight-Loss Interventions......... 6
Patient Factors and Treatment Outcome........ 7
Treatment Factors and Treatment Outcome ...... 8
Diabetes Status and Treatment Outcome........ 9
Dietary Regimen and Treatment Outcome........ 11
Exercise and Treatment Outcome............... 13
Adherence and Treatment Outcome.............. 15
A Conceptual Framework for Treatment
Intervention............................. 18
The Role of Skill Acquisition in
Treatment Intervention..................... 21
Mastery Criteria and Skill Acquisition....... 22
Minimal Treatment Intervention for Weight
Loss.................. ...................... 28
Present Study................................. 29
Rationale for the Present Study.............. 30
Hypotheses........... ............... ...... 33

II METHOD............................. .......... 34

Subjects ............... .. ................ 34
Measures.................... .................. 36
Demographic/Medical Information............ .. 36
Adherence-Related Measures................. 36
Calorie and Fat Intake.................... 36
Physical Activity......................... 37
Knowledge......................... ....... ..... 37
Weight-Related Measures..................... 38


II METHOD, continued............................
Quality of Life Measures.................... 38
Sickness Impact Profile.................... 38
Beck Depression Inventory.................. 39
Perceived Stress Scale.................... 40
Satisfaction Rating Scale.................. 41
Satisfaction with Life Scale.............. 41
Procedure................................................ 41

III RESULTS..................................... 51

Data Analytic Strategy....................... 51
Preliminary Analyses and Attrition .......... 52
Analyses of Adherence-Related Measures....... 53
Analyses of Weight-Related Measures........... 60
Analyses of Quality of Life Measures......... 66
Analyses of the Effect of Diabetes Status.... 68
Additional Analyses of the Mastery-based
Condition................................ 71

IV DISCUSSION................................... 75


IN OBESE ADULTS........................... 90

B WEEKLY LESSON TOPICS........................ 91

C MASTERY CRITERIA FOR WOMEN.................. 92

D MASTERY CRITERIA FOR MEN.................... 93

REFERENCES... .................... ..................... 94

BIOGRAPHICAL SKETCH.................................... 105


Table Page

1 Means and Standard Deviations of Adherence
Variables at Pre and Posttreatment for the
Mastery-based, Standard Behavioral, and
Weight-loss Education Conditions............ 54

2 Means and Standard Deviations of Body
Weight at Months 0, 3, and 6 for the
Mastery-based, Standard Behavioral, and
Weight-Loss Education Conditions (in pounds) 61

3 Mean Values of the Quality of Life Variables
for the Mastery-based, Standard Behavioral,
and Weight-Loss Education Conditions at
Pre- and Posttreatment..................... 67

4 Means and Standard Deviations for
Weight-Related Variables for Diabetic and
Nondiabetic Subjects at Pre- and
Posttreatment... ........................... 69


Figure Pag

1 Changes in Average Daily Caloric Intake
Over Time................................. 56

2 Changes in Physical Activity Over Time..... 58

3 Changes in Weight Over Time for the Three
Conditions............................... 62

4 Percentage of Subjects at Each Mastery
Level..................................... 72

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy




August 1994

Chairperson: Dr. Michael G. Perri
Major Department: Clinical and Health Psychology

The efficacy of behavioral treatments of obesity may be

improved by tailoring group interventions to the progress of

individual participants. The implementation of a personalized

system of behavioral acquisition, with specific criteria for

mastery of new skills, may allow each group participant to

progress at an individualized pace, with reinforcement based

on mastery of key changes in eating and exercise behaviors.

The current study examined the effects of behavioral treatment

using mastery criteria, compared to a standard behavioral

intervention. A weight-loss education condition was added to

examine the efficacy of a minimal treatment intervention. One

hundred and six mildly and moderately obese adults were

randomly assigned to one of the three conditions. The


education condition received a series of 6 monthly classes

involving lectures and handouts on proper diet, exercise, and

behavioral strategies for weight loss. The standard treatment

involved 25 weekly sessions of behavioral group treatment for

weight loss. The mastery condition also included 25 weekly

behavioral group sessions, but participants progressed through

nine levels of diet and exercise goals, with personalized

pacing of progress and reinforcement contingent on skill

acquisition at each level.

Because of the additional health benefits of weight loss

for individuals with Type II diabetes, the efficacy of the

weight-loss interventions for individuals with diabetes

compared to nondiabetic subjects was conducted as an

additional component of the study.

Results indicated that both the mastery and standard

conditions demonstrated significantly greater weight losses

than the education condition, but they did not differ

significantly from each other. Both the standard and mastery

interventions demonstrated significant and equivalent

reductions in caloric intake, fat intake, perceived stress,

and depression, and increases in physical activity. These

findings suggest that both standard and mastery approaches

were effective in producing positive changes in diet,

exercise, weight, and emotional well-being. Personalizing

treatment through the use of mastery criteria, however, may


not produce better short-term results than standard treatment

alone. In addition, results indicated that the education

condition was insufficient for producing significant weight

loss. Finally, both the diabetic and nondiabetic groups

demonstrated equivalent, significant changes in weight loss.



Obesity is a prevalent and serious health problem in the

United States. Over 25% of Americans are estimated to be

obese (Kuczmarski, 1992), and, in 1990, approximately 30% of

females and 22% of males were attempting to lose weight

through diet and exercise (National Center for Health

Statistics, 1992). The chronic nature of obesity and its

association with decreased longevity and with greater risk of

such diseases as coronary heart disease, hypertension,

diabetes, and cancer (Manson, Stampfer, Hennekens, & Willett,

1987; Robinson, Hoerr, Strandmark, & Mavis, 1993; Sjostrom,

1992) render it a major medical challenge for many


Weight loss can result in beneficial changes in blood

pressure, cholesterol, triglycerides, and glucose control

(Blackburn & Kanders, 1987; Tremblay et al., 1991). Weight

reduction is, therefore, important for reducing risks for

disease. Weight reduction has particular benefits for

individuals with Type II diabetes and currently is the

treatment of choice for management of the disease. Treatments

of obesity, however, have met with varied success (Caterson,

1990; Foreyt, 1987; Foreyt & Goodrick, 1991; Safer, 1991).

Although behavioral interventions that combine behavior

modification strategies with diet and exercise have proved to

be among the most effective for mildly to moderately obese

individuals, problems remain. The significant variability in

the amount of weight lost by individuals and poor maintenance

subsequent to treatment highlight the importance of improving

the efficacy of the behavioral approach to weight loss.

Additionally, as health care costs have escalated, the

importance of determining the minimum amount of intervention

required to effect change has become increasingly important.

The current project examined several important issues in

weight-loss treatment: the application of a personalized

system of skill acquisition using mastery criteria for

enhancing treatment effects; the efficacy of a minimal

treatment intervention; and the treatment response of

individuals with Type II diabetes compared to nondiabetic

individuals. The introduction will provide a review of the

literature on the behavioral treatment of obesity, describing

the benefits of weight loss and highlighting the key factors

that have been examined in treatment outcome including:

patient variables, treatment variables, diabetes status,

adherence, dietary regimen, and exercise. A model is

presented to assist in conceptualizing the process of

behavioral weight loss interventions. Mastery criteria is

proposed as an additional factor that may improve treatment

efficacy, and a treatment study which examined the effect of


mastery criteria, minimal intervention, and diabetes status on

treatment response is described.

Benefits of Weight-Loss Interventions

Both physical and psychological benefits of weight loss

treatment have been demonstrated. Research has provided

substantial evidence that weight loss treatment is associated

with reductions in risk factors for diseases, such as for

diabetes and coronary heart disease (CHD). For example,

beneficial changes in triglyceride and cholesterol levels,

known cardiovascular disease risk factors, have been

demonstrated as a result of weight reduction (Kaplan, Wilson,

Hartwell, Merino, & Wallace, 1985; Wing, Koeske, Epstein,

Nowalk, Gooding, & Becker, 1987). In individuals with

diabetes, weight reduction has been associated with a decrease

in blood glucose levels (Henry, Scheaffer, & Olefsky, 1985;

Wing, Shoemaker, Marcus, McDermott, & Gooding, 1990) likely

resulting from increases in insulin sensitivity (Hughes,

Gwynne, Switzer, Herbst, & White, 1984; Wing, Epstein, Nowalk,

Koeske, & Hagg, 1985) and insulin secretion (Gumbiner et al.,


Some data suggest that the physical benefits of treatment

may be produced by even modest weight loss (for an extensive

review, see Kanders & Blackburn, 1992). Wing et al. (1987),

for example, examined the effect of a behavioral weight

control program on glycemic control in 114 obese adults with

Type II diabetes. Results indicated that weight losses of 15

pounds or more were associated with significant improvements

in glycosylated hemoglobin levels, fasting blood glucose, and

insulin at one year follow-up. Significant improvements in

HDL-cholesterol and triglycerides were also observed.

More extensive data have been collected demonstrating the

effect of weight loss on risk factors for disease than on

direct reductions in morbidity and mortality. The supportive

data that does exist regarding morbidity and mortality comes

from longitudinal studies examining changes in physical status

over time. For example, the Framingham Heart Study, which

began following 5209 individuals in 1948, reported an inverse

correlation between weight loss and risk for CHD (Hubert,

Feinleib, McNamara, & Castelli, 1983). Specifically, a 10%

decrease in relative weight was correlated with a 20%

reduction in CHD risk. Other studies have demonstrated that

reductions in obesity to within a normal weight range can

significantly reduce mortality ratios (e.g., Society of

Actuaries, 1980). Additional contemporary research is needed

to clarify the extent to which weight loss directly affects

morbidity and mortality.

Psychological benefits of weight reduction have also been

examined. Earlier research indicated that weight reduction

adversely affected individuals emotionally, resulting in

increased feelings of anxiety and depression (Stunkard, 1957;

Stunkard & Rush, 1974). Smoller, Wadden, and Stunkard (1987),

however, noted that such studies typically employed

psychodynamic approaches to treatment and more subjective

measures of psychological status. Studies utilizing

behavioral treatment methods and objective test measures have

demonstrated either no effect or positive effects on

psychological status (O'Neil & Jarrell, 1992; Wing, Epstein,

Marcus, & Kupfer, 1984). In a review of mood changes during

behavioral weight loss programs, Wing, Epstein, Marcus, and

Kupfer (1984) reported that 6 of 10 studies demonstrated

significant reductions in depression or anxiety subsequent to

treatment. Other studies also have demonstrated a decrease in

the self-report of depressive symptomatology as a result of

participation in a behavioral weight loss program (Salata,

Marcus, Nowalk, & Blair, 1986; Wadden & Stunkard, 1986; Wing

et al., 1987). Furthermore, Kaplan, Hartwell, Wilson, and

Wallace (1987) have shown that treatment targeting diet and

exercise resulted in significant improvements in quality of

life in patients with Type II diabetes, independent of weight


In summary, research indicates that behavioral treatments

can have positive physical and psychological benefits for

overweight individuals. These potential benefits underscore

the importance of continuing to modify the behavioral approach

to improve overall efficacy of and individual responsivity to


Behavioral Weight-Loss Interventions

Behavioral programs generally consist of weekly sessions

within a group format that extend for approximately 20 weeks

(Wing, 1992). The focus of treatment is on modifying

behaviors in order to decrease caloric intake and to increase

caloric expenditure to yield a negative energy balance.

Behavioral approaches have utilized four general strategies in

targeting a negative energy balance: stimulus control,

reinforcement of targeted behaviors, self-monitoring of diet,

and "modification of the topography of eating," such as by

slowing the speed of eating (Perri, Nezu, & Viegener, 1992).

Perri et al. (1992) noted that three additional strategies

have become standard components of comprehensive behavioral

treatment interventions: cognitive restructuring, increased

physical activity, and nutritional training.

By the early 1980s, data indicated that behavior therapy

consistently produced greater weight losses and lower dropout

rates as compared to the traditional low-calorie diet approach

to obesity management (Brownell, 1982). More recent data

indicate that treatment programs that combine diet, exercise,

and behavior modification strategies yield the best long-term

results compared to other non-surgical approaches, such as

nutrition education and pharmacotherapy (Caterson, 1990; Wing,

Epstein, & Marcus, 1990). Both initial effects of treatment

and short-term maintenance of weight change have improved with

behavioral treatments: weight loss now averages approximately

20 pounds post-treatment and 13 pounds at one-year follow-up

(Brownell & Wadden, 1992). Long-term maintenance, however,

remains a significant problem. Brownell and Wadden (1986),

for example, reported that individuals, on average, regain 36%

of the initial weight lost during the year subsequent to

treatment. Four-year follow-ups indicate that only a small

minority maintain the total amount of weight lost during

treatment, and the majority regain a significant amount of the

weight lost (Kramer, Jeffery, Forster, & Snell, 1989;

Stalonas, Perri, & Kerzner, 1984). These studies highlight

the importance of continued efforts to improve maintenance of

treatment changes in order to enhance the long-term effect of

behavioral interventions (Perri et al., 1992).

Patient Factors and Treatment Outcome

In order to improve the short and long-term effects of

behavioral weight loss interventions, research has focused on

delineating factors that contribute to treatment outcome.

Investigations into baseline predictors of weight loss

indicate that initial weight and gender are the only patient

characteristics that predict weight loss, with heavier

individuals losing more weight than lighter individuals and

males losing more weight than females (Wing, 1992). These

variables appear to largely reflect the same factor, as gender

differences are attributable, at least in part, to differences

in initial body weight.

Treatment Factors and Treatment Outcome

Numerous treatment factors also have been investigated

for their contribution to outcome. Bennett (1986), in a

review of 105 studies of behavioral weight loss interventions,

found posttreatment weight loss to be significantly positively

related to treatment duration, therapist experience, amount of

therapist contact, and family involvement. Treatment length

has been found to be positively related to both initial weight

loss and maintenance of change (Brownell & Kramer, 1989;

Perri, Nezu, Patti, & McCann, 1989). Wing (1992) described

three components of treatment as being consistently related to

weight loss: caloric reduction, increased exercise, and

regular self-monitoring of caloric intake. In addition,

Perri, Sears, and Clark (1993) identified relapse prevention

training, continued professional contact, social support, and

exercise as factors that may enhance long-term maintenance of

weight loss. Finally, data indicate that weight-loss

education (e.g., regarding nutrition and exercise) is a

necessary, but not sufficient component, for effecting

behavior change (e.g., Heiby, Gafarian, & McCann, 1989).

Nutritional counseling or training, therefore, has become a

common component of behavioral treatment programs. Research

continues to examine and modify treatment elements to

determine the combination of strategies that will optimize


Diabetes Status and Treatment Outcome

Research indicates that obesity plays a particular role

in both the development and management of Type II or

non-insulin-dependent diabetes mellitus (NIDDM). Of the over

10 million individuals currently affected by NIDDM in the

United States, approximately 80 percent are obese (Cox &

Gonder-Frederick, 1992), and weight loss is considered the

cornerstone of treatment for managing the disease. Obesity

also increases the relative risk of developing diabetes to 2.9

times that of nonobese individuals (VanItallie, 1985). Risk

of developing diabetes has been positively associated with

upper body fat distribution (Ohlson et al., 1985) and duration

of obesity (Everhart, Pettitt, Bennett, & Knowler, 1991).

Weight loss, therefore, is critical both for prevention and

management of NIDDM in the obese population.

Because of the particular role that obesity treatment

plays in development and management of NIDDM, it is important

to treatment planning to determine any individual differences

in treatment response due to diabetes status. Some data, for

example, suggest that individuals with non-insulin-dependent

diabetes may have more difficulty in losing weight than

nondiabetic individuals. Wing, Marcus, Epstein, and Salata

(1987) examined potential differences in weight loss according

to diabetes status by comparing 12 obese patients with

diabetes and their overweight spouses who were participating

in a behavioral weight-loss intervention. Nondiabetic spouses

demonstrated significantly greater weight loss than the

patients with diabetes at the end of the 20-week program (Ms

= 13.4 kg and 7.5 kg, respectively). Poorer weight loss in

the patients with diabetes did not appear to be attributable

to diabetic medications, given that no significant differences

in weight loss were found between patients on insulin, oral

medication, or diet only. Significant differences in caloric

intake suggested that outcome was attributable to differences

in dietary adherence. Specifically, nondiabetic spouses

demonstrated significantly greater reduction in caloric intake

than the subjects with diabetes. Similarly, Henry et al.

(1985) compared 10 inpatients with Type II diabetes to 5

nondiabetic inpatients, with both groups placed on a VLCD for

36 days. Results indicated that the nondiabetic patients lost

significantly more weight than the patients with diabetes.

One possible reason for differences in weight loss may be

the effect of diabetes medications on weight loss. Studies

have suggested that individuals taking insulin lose less

weight than those who are not on insulin therapy (Harris,

Davidson, & Bush, 1988; University Group Diabetes Project,

1971). This may be attributable to both psychological

reasons, such as fear of hypoglycemic responses when

diminishing food intake, and physiological reasons, such as

the increase in feelings of hunger associated with the release

of insulin in the bloodstream. Results from a study by Wing,


Shoemaker, et al. (1990), however, indicated that weight loss

was not related to medication regimen.

To date, only a few studies with limited sample sizes

have investigated differences in response to treatment

according to diabetes status. Because of the potential

clinical implications of these results, further clarification

of this issue is important.

Dietary Regimen and Treatment Outcome

Different diets have been examined for their effect on

treatment outcome. In general, recommended components of a

dietary regimen are caloric restriction and reduction in fat

intake. An increase in fiber and reduction in cholesterol

also generally are encouraged. Guidelines provided by the

National Institutes of Health (1987) placed primary importance

on caloric restriction and secondary emphasis on macronutrient

composition of the diet. Specifically, the committee

suggested a moderate caloric reduction that will produce a

gradual weight loss, and a diet consisting of approximately

30% fat, 20% protein, 50% carbohydrate. Research

investigating the use of calorie-counting diets and exchange

system diets to achieve these goals in persons with Type II

diabetes indicate that the two regimens do not significantly

differ in effectiveness (Wing, Nowalk, Epstein, & Koeske,


Emphasizing a low-fat diet is important for both reducing

body weight and total serum cholesterol level. Research has


shown that fat intake is significantly correlated with body

fatness (Miller, 1991). In addition, total fat intake is

inversely related to carbohydrate intake and directly related

to total calorie intake (Hammer, Barrier, Roundy, Bradford, &

Fisher, 1989). Thus, reduction in fat intake may assist in

achieving weight loss, reducing total serum cholesterol, and

balancing macronutrients in the diet.

Very-low-calorie diets (VLCDs) have shown promise in

producing rapid weight loss and improving glycemic control

(Amatruda, Richeson, Welle, Brodows, & Lockwood, 1988; Genuth,

1979; Henry et al., 1985). However, weight is often quickly

regained following treatment with VLCDs. In an effort to

promote maintenance of weight change, some studies have

incorporated the use of VLCDs into a behavioral treatment

program. Studies have found that combining the use of a VLCD

with behavior therapy improves long-term weight loss compared

to use of a VLCD alone, but does not produce better

maintenance of weight loss than behavior therapy alone (Wadden

et al., 1989; Wadden and Stunkard, 1986).

A similar study by Wing et al. (1991) using subjects with

diabetes corroborated these results. Thirty-six subjects with

Type II diabetes were assigned to either a standard behavioral

treatment (BT) or to a standard behavioral treatment that

included the use of a VLCD to examine differences in changes

in weight and glycemic control. Subjects in the VLCD

condition demonstrated significantly greater reductions in

weight, fasting blood glucose, and glycosylated hemoglobin

after the 20-week program compared to the BT group. However,

there were no significant differences in weight loss between

the two conditions at one-year follow-up due in large part to

a regaining of weight in the VLCD condition. Interestingly,

the VLCD condition continued to demonstrate significantly

better glycemic control than the BT group, which the authors

postulated may be due to increased insulin secretion. The

cost and risks of VLCDs and the evidence that VLCDs do not

significantly enhance weight loss or psychological benefits of

treatment compared to standard behavior therapy (Wing, Marcus,

Blair, & Burton, 1991) suggest that their general use may not

be justified at this time. However, the improvement in

long-term glycemic control warrants further investigation to

determine its utility for the diabetic population.

In summary, adherence to a well-balanced, low-fat,

reduced calorie diet is an important component of a behavioral

weight-loss intervention. Efforts to increase adherence to

dietary regimen, therefore, is a critical target in enhancing

treatment efficacy.

Exercise and Treatment Outcome

Research has indicated that exercise contributes to

weight loss and maintenance in the general obese population

(Perri et al., 1988; Perri, McAdoo, McAllister, Lauer, &

Yancey, 1986; Stalonas, Johnson, & Christ, 1978) as well as in

the population of obese individuals with diabetes (Wing,


1992). Wing et al. (1988) examined the effect of adding

exercise to a behavioral treatment program on weight loss.

Results indicated that the combination of diet and moderate

exercise (walking a 3-mile route four times a week) enhanced

both short and long-term effects on weight loss compared to

subjects in a diet only condition. Additionally, both

physical and psychological benefits of exercise have been

documented. Benefits include favorable changes in

cardiovascular risk factors, such as reductions in blood

pressure, triglyceride levels, and cholesterol levels, and

improvements in mood (Dubbert, 1992). Given that many of the

complications of NIDDM are related to cardiovascular disease,

the changes in lipid levels and blood pressure resulting from

exercise are particularly advantageous to individuals with

Type II diabetes. Additionally, exercise has been shown to

improve insulin action by reducing insulin resistance,

particularly by diminishing the amount of adipose tissue and

by increasing the number of insulin receptors (Sherman &

Albright, 1992). Exercise also may ultimately diminish

mortality risk. Kohl, Gordon, Villegas, and Blair (1992)

followed 8715 men for an average of 8.2 years to examine the

relationship between glycemic control, mortality risk, and

cardiorespiratory fitness. Their data suggested that, across

all levels of glycemic control, men with better

cardiorespiratory fitness had a lower mortality risk.

Although the benefits of exercise are potentially

significant, positive effects are transient unless changes in

activity are maintained. Research indicates that those who

continue to exercise are more likely to maintain weight loss

(Kayman, Bruvold, & Stern, 1990). Adherence to exercise

prescriptions and creating activity regimens that individuals

can incorporate into their lifestyles, therefore, is important

(National Institutes of Health, 1987). In developing an

exercise prescription for the obese population,

recommendations must be tempered by considerations of the

likelihood and feasibility that the goals can be achieved.

The exercise goal in obesity treatment generally approximates

the guidelines provided by the American College of Sports

Medicine (1991) of exercising 20 to 30 minutes, 3 to 5 times

per week (e.g., Monk, Adolphson, Hollander, & Bergenstal,

1988; Wing, Shoemaker, et al., 1990). Like dietary regimen,

exercise is a critical part of an effective behavioral

intervention, and warrants continued efforts to increase

individuals' participation in exercise to produce weight loss.

Adherence and Treatment Outcome

Because the management of obesity is reliant primarily

upon self-care behaviors of the individual, adherence to

regimen is critical for weight control. Adherence to

behavioral treatment regimens for weight loss--such as

recording food intake and exercising regularly--is variable,

however, and generally worsens subsequent to treatment (Miller

& Sims, 1981; Perri, 1987; Van Dale, Saris, & Hoor, 1990).

For patients with diabetes, diet and exercise prescriptions

are identified as the most difficult aspects of diabetes

management (Glasgow, McCaul, & Schafer, 1986). Given that

performance of particular health behaviors is critical for

producing and maintaining weight loss in behavioral treatments

of obesity, adherence is essential for successful weight

management. Although a substantial amount of research has

focused on factors that influence adherence, few predictors

have been identified. Adherence does not appear to be

consistently related to demographic variables, such as race,

gender, or socioeconomic status (Levy, 1987; Meichenbaum &

Turk, 1987). Levy (1987) described 3 general reasons for

nonadherence to regimens: patient has inadequate skills or

knowledge to perform the behaviors; patient has a belief

system that is incompatible with or unsupportive of the

regimen; or the patient's environment interferes with

adherence. In addition to patient variables, Meichenbaum and

Turk (1987) identified 3 other general factors involved in

determining adherence: disease variables (e.g., severity of

symptoms); aspects of treatment, such as regimen complexity;

and relationship variables, specifically between the patient

and health care provider. The importance of self-efficacy

expectations (confidence in ability to perform a behavior) for

predicting adherence also has been highlighted (McCaul,

Glasgow, and Schafer, 1987).

Historically, poor adherence or noncompliance has implied

that a patient made a conscious decision not to follow a

medical regimen prescribed. In addition, noncompliance

traditionally has been conceptualized as a single construct;

either a patient was "compliant" or noncompliantt." Recent

theoretical and empirical contributions, however, suggest that

multiple factors contribute to adherence. For example, the

Health Beliefs Model emphasizes the role of cognitive

factors--such as perceived severity of illness and potential

barriers to health behavior--in determining adherence (Becker,

1974). Research supports the role of health beliefs in

predicting the performance of health behaviors (Harris & Linn,

1985; Leventhal, Zimmerman, & Gutman, 1984; Wilson et al.,

1986). Johnson (1990) identified knowledge, skills, and

patient/physician communication as factors that contribute to

the likelihood of adherence in patients with diabetes.

Johnson recommended that adherence be approached as a

multidimensional construct, with measurements of adherence

involving the assessment of separate self-care behaviors

rather than of global adherence. Factor analyses of adherence

behaviors required of patients with Type I diabetes provide

support for this multifactorial approach (Johnson, Tomer,

Cunningham, & Henretta, 1990). Performance of self-management

skills is an important aspect of successful weight loss, and

although research has delineated a number of factors that

influence adherence, a conceptualization of the process by

which adherence occurs has been lacking in clinical research.

A Conceptual Framework for Treatment Intervention

The objective of behavioral treatment interventions for

obese individuals is to improve patients' adherence to (i.e.,

performance of) specific self-management behaviors. The

success of treatment is contingent upon this objective being

met. A model of health behavior change has been developed to

conceptualize the process by which this may occur (please see

Appendix A). According to the model, various mediating

variables affect the likelihood of a change in health

behaviors. Changes in health behaviors may then directly

result in improved behavioral outcomes (e.g., quality of life)

or indirectly affect behavioral outcomes through biological

changes. Adherence to health behaviors mediates desired

outcomes. Accordingly, specific research for improving the

efficacy of behavioral interventions has involved identifying

and manipulating factors that affect adherence (i.e., the

mediating variables) and delineating the health behaviors

necessary for producing the desired outcome.

Application of the model to obesity treatment provides a

specific example of this conceptual approach to intervention.

In obesity treatment, the targets of intervention are specific

health behaviors that will increase the likelihood of weight

loss, such as caloric restriction, physical activity, and

diminished intake of fat. Changes in these behaviors may be

influenced by such factors as skill level, knowledge of the

behavior, and health beliefs, such as perceived self-efficacy.

Making changes in these health behaviors may directly enhance

quality of life, such as by improving mood, or indirectly

enhance the quality of life through biological outcomes like

weight loss and diminished CHD risk. For example, for

individuals with diabetes, greater glycemic control may

improve quality of life by diminishing the need for

medications, decreasing the number of diabetic complications,

and improving physical functioning. For nondiabetic

individuals, weight loss may enhance quality of life by

elevating self-esteem or improving capacity for work and other


The focus on behavioral outcomes as the critical measure

of treatment efficacy has recently been emphasized in health

psychology literature. Kaplan (1990), for example, has argued

that behaviors are the most important outcome measures and,

therefore, should be the focus of research in behavioral

medicine. Kaplan asserts that biological measures and disease

parameters are mediators that influence behavioral outcomes

and are meaningful only insofar as they affect the quality or

longevity of an individual's life. For example, the primary

outcome measure in interventions with subjects with Type II

diabetes has been blood glucose levels, and improved glycemic

control has been considered the primary goal of treatment (Cox

& Gonder-Frederick, 1992; National Institutes of Health,

1987). The value of these biological indices, however, is

dependent upon their meaning for behavioral outcomes, such as

the extent to which diabetic symptoms interferes with daily

functioning (physical, social, interpersonal, etc.).

The psychological impact of obesity, including

psychological distress related to social discrimination and

body-image disparagement (Stunkard, 1976; Wadden & Stunkard,

1985), provide an additional reason for assessing behavioral

outcome variables when conducting interventions with obese

individuals. Although no significant differences in

psychopathology between groups of obese and nonobese

individuals not seeking help with weight loss (Wadden &

Stunkard, 1985), a significant minority of overweight

individuals in clinics or weight loss programs do demonstrate

psychological consequences, such as low self-esteem and poor

body image, that is related specifically to their obesity.

Additionally, some research indicates that psychological

well-being improves subsequent to weight loss (Wing et al.,

1984). These data highlight the importance of assessing

changes in psychological variables when evaluating the success

of interventions with this population.

According to the model, methods that increase the

likelihood of adherence to weight-management behaviors can

have a direct effect on treatment outcome. Determining such

methods, therefore, is an important part of improving

individuals' response to treatment. The current study

examined the use of a personalized system of skill acquisition

using mastery criteria for increasing adherence to weight

management behaviors in an effort to improve treatment


The Role of Skill Acquisition in Treatment Intervention

Over the last 20 years, improvements in behavioral

treatments of obesity have resulted in part from extending the

use of behavioral principles for promoting the performance of

behaviors, such as the addition of incentive systems and

specific skill training (Wing, 1992). In particular,

considerable research has focused on factors involved in the

maintenance of behaviors necessary for weight management.

These efforts have increased the typical weight loss achieved

in behavioral treatment to approximately 20 pounds. However,

significant variability in patient response to treatment

exists (Brownell & Wadden, 1986; Perri et al., 1992), and the

majority of individuals who do lose weight gradually regain it

after treatment is completed (Brownell & Jeffery, 1987; Wadden

& Bell, 1990; Wing et al., 1985). These limitations warrant

continued investigation into the components of treatment

necessary to enhance both the short and long-term effects of

behavioral intervention.

One component of treatment that has not been adequately

examined is the acquisition or learning of the behavioral

skills necessary for change. Separate variables may be

responsible for the acquisition versus the maintenance of

behaviors (Bandura, 1969), and behaviors that are not fully

acquired (i.e., not adequately learned) may be poorly

maintained. Currently, behavioral treatments are frequently

conducted in a group treatment format, with new information

and skills introduced at a standardized pace that is

relatively independent of individual response repertoires and

progress (Perri, 1989). One possible explanation, therefore,

for the moderate initial treatment effects and poor

maintenance over time is that the interventions may not have

been tailored enough to individual differences in skill

acquisition to have promoted the learning necessary for

behavior change. Improving the initial learning of desired

health behaviors, therefore, may improve adherence to these

behaviors and overall efficacy of treatment.

Mastery Criteria and Skill Acquisition

The use of mastery criteria for promoting behavior change

may be one method for improving the acquisition of behavioral

skills necessary for weight loss. The mastery-based approach

breaks down behavioral skills into smaller increments, defined

by specific criteria, within a number of levels that

successively approximate the desired behavior. When criteria

for a skill are met, the individual is reinforced, and

progresses to a new level where new criteria for the skill are

delineated that more closely approximate the end goal. For

example, if a dietary goal of 1200 calories per day is

established with a patient, mastery criteria on the first

level may be a broad range that includes the 1200 calorie

target, such as a range from 1000-1800 calories per day. Once

the individual demonstrates an average daily caloric intake

within this range for one week, the individual is reinforced.

The individual would then progress to Level 2, where criteria

for the caloric range would be narrower, such as 1000 to 1500

calories per day. Criteria within successive levels would

continue to become narrower until the target goal is reached

at the final level. The individual thereby gradually acquires

the ability to eat an average 1200 calorie/day diet.

The use of mastery criteria tailors the pace of treatment

and reinforcement to the progress of each individual. In this

way, interventions may be adapted to individual differences in

ability even within a group treatment format. Thus, the use

of mastery criteria may improve the efficacy of group

treatment by addressing individual variability in performance.

The importance of individualizing treatment is highlighted in

recent literature (Brownell & Wadden, 1992; Perri et al.,

1992). Responding to this need--by tailoring interventions

and setting goals according to individual abilities--may

increase the likelihood of learning skills and achieving

goals. This may, in turn, enhance feelings of self-efficacy

in addition to improving maintenance of changes.

The use of mastery criteria for promoting behavior change

is supported by learning theory. A basic premise of learning

theory is that effective learning occurs through shaping,

which is defined as the contingent, differential reinforcement

of successive approximations of the desired behavior (Skinner,

1953). Accordingly, shaping may be used to facilitate

acquisition of a new behavior and to increase the probability

that the behavior will be emitted. The targeted behavior is

separated into a series of progressive responses; when the

initial or simplest response is emitted, it is reinforced. A

modification of the response (the next response in the series)

must then be emitted in order to receive reinforcement. In

this manner, a complex behavior is gradually learned by

dividing it into simpler responses that are more likely to be

performed initially, and therefore are more likely to be

reinforced. Additionally, the gradual, sequential learning in

this procedure may strengthen and refine the skill being

acquired (Skinner, 1953).

The mastery-based approach incorporates several of the

principles described by Keller (1966, 1968). His personalized

system of instruction (PSI) combined programmed instruction

with principles based on operant conditioning theory. In this

model, students progress through educational material at a

pace commensurate with their abilities. Advancement to new

material is contingent upon demonstrating mastery of the

preceding material, according to the "unit perfection

requirement." Understanding of material is continuously

evaluated through quizzes and assignments that accompany each

lesson and immediate feedback is provided. This teaching

model applies a variety of behavior principles to improve the

learning of new material. The critical components of this

approach--the individualized pace, emphasis on mastery of

material, and frequent, contingent reinforcement--are utilized

in the mastery-based model to promote learning of the

information and skills necessary for producing behavior


The utility of the mastery-based approach also is

supported by specific findings pertaining to dietary change

and weight loss. For example, research suggests that gradual

changes in diet and exercise are associated with better

outcomes in obesity treatments (Foreyt & Goodrick, 1991). In

particular, data indicate an inverse relationship between the

rate of initial weight loss and maintenance over time

(Brownell & Wadden, 1986). These data suggest that gradual

changes made using mastery criteria in weight loss treatment

may help to enhance maintenance of behavior changes. Ewart

(1989) described two important effects of a gradual,

successive approach to dietary change: (1) increased frequency

of reinforcement due to smaller, more obtainable goals and (2)

greater mastery of skills for simpler tasks before progressing

to skills requiring a greater level of complexity and

difficulty. These characteristics are important components of

skill acquisition and a fundamental part of mastery-based


Research conducted by Epstein et al. (1994) with obese

children and their overweight parents support the use of

mastery criteria in weight-loss treatment interventions.

Children and their parents were assigned to either a standard

behavioral weight-loss treatment that progressed at a rate

independent of individual skill acquisition or to a standard

treatment that paced interventions according to the

participant's mastery of specified self-management skills. In

the mastery-based condition, children and parents progressed

at their own rate in mastering new skills for diet, exercise,

and parenting through five progressive levels and they were

rewarded each time they mastered skills within a given level.

Subjects in the standard treatment also were rewarded at the

same frequency, but noncontingently. The results indicated

that subjects in the mastery-based condition, parents as well

as children, demonstrated significantly greater weight change

than subjects in the standard behavioral treatment control


Reinforcement is a critical factor in facilitating

acquisition of a desired behavior that is highlighted in the

mastery-based approach. It is generally accepted that

continuous reinforcement schedules are the most appropriate

for acquisition of a new behavior and intermittent schedules

(particularly variable ratio schedules) are the most effective

for long-term maintenance (Catania, 1984). More recent

research, however, suggests that the most effective schedule

depends upon the nature of the targeted behavior. Epstein

(1992) suggests that, for free operant behavior (as opposed to

behaviors that are emitted within specific, structured

environments, like a prison), continuous reinforcement

schedules are the most effective in promoting resistance to

extinction. Weight-loss interventions, then, are more likely

to be effective in maintaining changes if the number of

individual responses that are reinforced is maximized during

the acquisition phase. This may be achieved by providing

opportunities for frequent reinforcement within treatment,

structuring reinforcers within the social environment, and

emphasizing the use of self-reinforcement by the patient.

The mastery-based approach increases the frequency of

reinforcement by breaking skills down into increments that

result in attaining goals easier and more often. For example,

dividing caloric intake goals into ranges that successively

become narrower and closer to the final goal increases both

the likelihood of achieving the goal and the number of goals

that are achieved. Consequently, opportunities for

reinforcement are also increased.

In summary, the use of mastery criteria provides a method

for tailoring interventions to individuals within a group

format and warrants further investigation to evaluate its

ability to enhance treatment efficacy.

Minimal Treatment Intervention for Weight Loss

It is widely accepted that behavioral weight-loss

treatment produces significantly greater weight loss than no

treatment in the general obese population. More recent

research has examined simpler, less intensive approaches to

weight reduction in order to determine the minimal

intervention required to effect significant weight loss (e.g.,

Black & Threlfall, 1986). Stunkard (1992) noted that

evaluations of the effectiveness of obesity treatments must

take into account the amount of time and money necessary to

produce an outcome. For example, a cost-benefit analysis of

the efficacy of standard behavioral treatment compared to a

VLCD alone and to a VLCD combined with standard treatment

indicated that, although there were minimal differences in

weight loss between the three conditions, there was a

significant difference in cost by 1-year follow-up (Stunkard,

1987). This type of research has become increasingly more

important as managed health care organizations place greater

emphasis on utilizing the most cost-effective interventions

that yield clinical benefits.

A few studies have examined the efficacy of education or

information interventions compared to standard behavioral

approaches. Black, Coe, Friesen, and Wurzmann (1984), for

example, compared a minimal intervention condition, in which

subjects were given three simple verbal instructions initially

and returned 7-months later, to a shortened behavioral weight

loss condition on weight loss, which involved weekly

behavioral treatment for 6 weeks. The minimal intervention

condition also was compared to a "complete" behavioral weight

loss condition, consisting of 10 weeks of behavioral

treatment. All groups lost a significant amount of weight

over time and no significant differences in weight loss

between groups were observed. Bibliotherapy programs, which

provide written instructions about weight loss with little or

no therapist contact, also have been explored as a potentially

less intensive and expensive approach to weight loss. Data

indicate these programs produce weight losses of between

approximately 12 and 15 pounds at 6-month follow-up (Marston,

Marston, & Ross, 1977; Pezzot-Pearce, LeBow, & Pearce, 1982).

These data are comparable to those cited for 10-week

behavioral treatments. However, current behavioral

interventions are generally longer than 10 weeks, and it is

important, therefore, to replicate results of these studies

with programs of more typical length (i.e., at least 20


Present Study

The present study examined the effects of behavioral

treatment using mastery criteria, compared to a standard

behavioral intervention and to a weight-loss education group.

Interventions and reinforcement were paced according to

individual mastery of specified self-management skills within

levels that successively approximated desired goals for diet,

activity, and knowledge. Specific mastery criteria included

daily calorie intake, physical activity, macronutrient

distribution, knowledge of nutrition and behavior skills, and

self-monitoring of eating and exercise habits. By defining

personal goals according to the individual's level of

competence and progress, intervention within the group setting

was tailored to the specific needs of individual members.

Given the importance of developing time- and cost-effective

interventions, a second goal of the study was to evaluate the

efficacy of a minimal weight-loss intervention involving

weight-loss education compared to standard behavioral

treatment. Finally, because of the significant number of

obese individuals either diagnosed with or at risk for Type II

diabetes, a third goal of the project was to examine

differences in response to treatment between individuals with

Type II diabetes and nondiabetic individuals.

Rationale for the Present Study

Weight loss improves hyperlipidemia, hypertension, and

glucose tolerance, diminishes mortality risk, and may improve

mood and quality of life. Improving the effectiveness of

obesity treatment, therefore, may be an important step toward

promoting the physical and psychological health of this

population. Although efforts to improve behavioral

interventions have continued, further research is needed on

the acquisition and facilitation of learning for improving

outcome. To date, no study has investigated the effects of

using a mastery-based program to enhance learning of desired

behavioral skills in obese adults, although both theoretical

and empirical evidence suggest that treatment effects may be

enhanced by improving skill competency and performance through

a more individualized approach.

The present study built upon prior research and responded

to recommendations for future investigations in this area.

For example, Brownell and Wadden (1992) emphasized the

importance of theory in the development and evaluation of

treatment. A mastery-based approach to obesity treatment

applies behavioral principles regarding acquisition of

behavior to facilitate performance of skills necessary for

weight management. The present study also responded to other

strategies suggested by Brownell and Wadden (1992) for

improving the effects of weight loss interventions, such as

extending the length of treatment, emphasizing exercise, and

teaching relapse prevention techniques. In addition,

strategies for improving the fit between the individual and

treatment is currently emphasized in obesity treatment

literature (Brownell & Wadden, 1992; Perri et al., 1992), and

the mastery-based approach provides a unique method for

achieving this within a group format by tailoring and pacing

interventions according to individual differences and needs.

The use of the weight-loss education condition addressed the

need to evaluate potentially more cost-effective treatment

alternatives. Finally, the ultimate goal of the study, which

was to improve the quality of life of overweight adults,

reflects an important shift in this area toward behavioral

outcomes in treatment interventions with this population, a

shift that is considered to be broader and ultimately more

meaningful (Epstein, 1992; Kaplan, 1990).


The primary hypothesis was that:

The mastery-based behavioral treatment would

significantly enhance learning of and adherence to

self-management skills related to diet and exercise when

compared to each of the two control conditions, weight-loss

education and standard behavioral treatment, at posttreatment.

Specifically, when compared to the two control conditions at

posttreatment, the mastery-based treatment condition was

predicted to demonstrate:

1. Lower mean daily caloric intake;

2. Lower mean daily fat intake;

3. Higher mean daily physical activity; and

4. Greater mean knowledge of weight-loss strategies.

Secondary hypotheses were that:

1. The mastery-based behavioral treatment would produce

significantly greater reductions in body weight at

posttreatment when compared to the standard behavioral

treatment and to the weight-loss education condition.

2. The mastery-based behavioral treatment would produce

significantly greater improvements in quality of life at

posttreatment when compared to the standard behavioral

treatment and to the weight-loss education condition.

3. Nondiabetic subjects would demonstrate significantly

greater reductions in body weight at posttreatment compared to

subjects with Type II diabetes.



One hundred and six subjects, 78 females (73.6%) and 28

males (26.4%), were selected from over 300 individuals

recruited by physician referral and advertisement through the

local newspaper. Eligibility criteria for participation

included being between 25 and 70 years of age, being willing

and able to post a $100.00 refundable deposit, and being

between 20 and 100 percent over ideal body weight according to

the Metropolitan Life Insurance Tables (1983). Women who were

pregnant or intended to become pregnant during the 12 months

of the study were excluded. In addition, subjects were

required to have clearance from their physician indicating

that there were no medical contraindications to their

participation in the weight loss program, such as

co-morbidities or diabetic complications that would render

exercise a risk. Of 67 individuals determined to be

ineligible for participation through the initial telephone

screening, 21% did not meet the age criteria, 31% the weight

criteria (18% exceeded the weight criteria, 13% were below

it); 21% the time commitment, 13% the exercise requirement, 7%

the financial requirement ($100 deposit), and 4% other reasons

(no transportation, recent weight loss, participation in

another program). Seven percent declined to participate after

the phone screening due to lack of interest. Twenty-seven

additional individuals were excluded subsequent to completing

an application to participate because of weight requirements

(59%), inability to walk for exercise (11%), medical

contraindications (7%), current attendance in another weight

loss program or recent weight loss (7%), time commitment (4%),

age (4%), or prior participation in the program (4%).

Eligible subjects ranged in age from 26 to 70 years, with

a mean age of 48.4 years (SD = 10.91). Ethnic origin of the

sample was predominantly white (92.5%), with only 4 (3.8%)

African-Americans and 4 (3.8%) individuals of Hispanic origin.

Twenty-one (19.4%) subjects in the sample were classified as

having Type II diabetes, 7 (6.5%) of whom were taking insulin.

Average level of education was 14.97 years (SD = 2.51), and

the majority (80.3%) reported an annual household income

between $21,000 and $60,000. Average initial body weight was

213.82 pounds (SD = 34.01), and ranged from 157.8 to 325.0

pounds. The mean percentage over ideal body weight was 51.26

percent (SD = 18.8), with a range from 20 to 99 percent, and

the mean body mass index was 34.41 (SD = 4.24), with a range

from 26.6 to 44.1.

Demographic/Medical Information

The General Information Questionnaire (GIQ) was used to

obtain general demographic information as well as specific

medical information. Medical information included current

symptoms of diabetes and related complications, other current

health problems, and a history of medical illnesses. A

listing of current medications was obtained and diabetic

medications were categorized as no medication, oral

hypoglycemic agents, or insulin injections.

Adherence-Related Measures

Calorie and fat intake. Calorie and fat intake was based

on a 3-day food record analyzed by a dietician using the

Minnesota Nutrient Data System microcomputer dietary analysis

system (Nutrition Coordinating Center, 1990). The database

includes more than 16,000 food items and brand names, with

options allowing more than 150,000 variants of foods. The

system has been used in several major research projects,

including the National Health and Nutrition Examination Survey

(Buzzard & Feskanich, 1987). The Minnesota dietary analysis

system has been described as having excellent standardization,

specificity, and reliability (Mitchell & Shacklock, 1991;

Nieman, Butterworth, Nieman, Lee, & Lee, 1992; Schackel,

Sievert, & Buzzard, 1988). Potential sources of error in the

computer food analysis include misreadings of food entries,

incorrect data entry, and judgment calls regarding

classification of food recordings not listed in the database.

As a second measure, a daily average of reported caloric

intake was derived from the 7-day food diaries maintained by

the subjects.

Physical activity. The Physical Activity Record, a 3-day,

self-report activity record developed by Bouchard et al.

(1983) was used to measure physical activity. The respondent

classified the level of energy cost of activities performed

during each 15-minute interval of the day (total = 96

intervals) according to categories ranging from low energy

cost (1) to high energy cost (9). A list of the categories

and examples of activities were provided to guide responses.

The mean daily energy expenditure and frequencies for

participation in higher energy expending activities may be

derived. Adequate psychometric properties are suggested by a

test-retest reliability of .96, a significant positive

correlation between energy expenditure and physical working

capacity, and a significant negative correlation between

energy expenditure and body fatness (Bouchard et al., 1983).

As a second measure, a weekly average of exercise (in minutes)

calculated from a one-week period was derived from the written

record of exercise recorded by subjects.

Knowledge. Knowledge of nutrition, behavior strategies,

and exercise was measured by the Nutrition and Exercise

Knowledge Questionnaire, a measure recently developed by a

licensed nutritionist (Babroff, 1993). Items were selected to

reflect the key elements in weight-loss involving nutrition,

behavior strategies, and exercise. Items have face validity

and were based on the professional judgment of a licensed

nutritionist. Additional information on the psychometric

properties of the measure are currently being collected.

Weight-Related Measures

Weight loss was measured according to changes in body

weight and body mass index. Weight was measured on a balance

beam scale. Height was taken in order to compare actual

weight compared to ideal weight specified by the Metropolitan

Life Insurance norms (1983) and to evaluate changes in body

mass index (kilograms/meters squared).

Quality of Life Measures

Sickness Impact Profile. This is a 136-item survey that

examines the impact of illness on health status by measuring

the degree of sickness-related behavioral dysfunction (Gilson

et al., 1975). The respondent indicates with a check

behaviors in which he or she engages. Scores on a physical

and a psychosocial dimension are generated in addition to an

overall score. Explicit instructions are included with the


Reliability and validity of the measure are very good.

Gilson et al (1975) reported test-retest reliabilities ranging

from r = .80 to r = .88 for the instrument. Bergner, Bobbitt,

Pollard, Martin, and Gilson (1976) cited criterion validity

measures for the SIP score of r = .54 with self-assessment of

sickness, r = .52 with self-assessment of dysfunction, r = .49

with physicians' assessment of dysfunction, r = .46 with

Activities of Daily Living, and r = .46 with the National

Health Interview Survey Data.

Beck Depression Inventory-Revised. The BDI (Beck, Rush,

Shaw, & Emery, 1979) is a widely-used self-report measure that

assesses neurovegetative, cognitive, and affective symptoms of

depression. The instrument contains 21 groups of statements

reflecting depressive symptoms, such as irritability, crying,

somatic preoccupation, and weight loss. The respondent

selects from 4 statements within each group the description

which best describes his/her behavior during the past week.

Each statement is rated on a 4-point scale and the sum of the

ratings yields the total score. General cut-offs for the

total score indicate the severity of depression (asymptomatic,

mild-moderate, moderate-severe, extremely severe). For this

study, the item pertaining to recent weight loss was omitted.

The BDI has been widely used for both clinical and

research purposes and has demonstrated strong psychometric

properties. For example, the manual for the BDI cites high

internal consistency values ranging from .79 to .90 in both

clinical and nonclinical populations (Beck & Steer, 1987).

Concurrent validity has been demonstrated through studies

correlating the BDI with the Hamilton Psychiatric Rating

Scale, the Zung Self-rating Depression Scale, the MMPI-D

scale, and clinical ratings of depression (Beck & Steer,

1987). Several comprehensive reviews of the BDI provide

detailed information regarding the psychometric properties of

the instrument (Edwards et al., 1984; Lips & Ng, 1985; Steer,

Beck, Riskind, & Brown, 1986).

Perceived Stress Scale. The PSS (Cohen, Kamarck, &

Mermelstein, 1983) is a 14-item measure of perceived stress.

The scale is comprised of 14 statements that describe thoughts

or feelings related to stress and coping. Examples of items

include "During the last month, how often have you felt

nervous and stressed?" and "During the last month, how often

have you felt that things were going your way?" Respondents

rate the frequency with which s/he has experienced the

statements during the last month on the following scale: 0 =

never, 1 = almost never, 2 = sometimes, 3 = fairly often, and

4 = very often.

Psychometric data provided by Cohen et al. (1983)

indicate adequate reliability and validity of the PSS. They

reported an average internal reliability value of r = .85 and

a test-retest reliability of r = .85 for a two-week interval

and r = .55 for a six-week interval. Additionally,

significant correlations between the PSS and life-event

scores, depressive and physical symptomatology, and social

anxiety on self-report measures were cited as evidence of

concurrent validity of the PSS.

Satisfaction Rating Scale. The Satisfaction Rating Scale

is a face valid measure used to obtain additional information

about quality of life. Respondents are asked to rate their

satisfaction with general areas in their lives such as

physical appearance, relationships, activity level, and

occupation. Areas are rated on a 7-point Likert scale that

ranges from 1 = "Very dissatisfied" to 7 = "Very Satisfied."

Satisfaction with Life Scale. The SWLS (Diener, Emmons,

Larsen, & Griffin, 1985) is a general index of subjective life

satisfaction. It is a brief measure that is comprised of five

statements that the respondent rates on a scale from 1 =

"strongly disagree" to 7 = "strongly agree." A sum total

provides a total score ranging from 5 to 35, with higher

scores indicating more reported satisfaction with life.

Adequate reliability and validity is evidenced by an internal

consistency value of r = .87, test-retest reliability over a

2-month period of r = 82, and significant correlations with

measures of subjective well-being (Deiner, Emmons, Larsen, &

Griffin, 1985).

Body weight measurements were obtained from all

participants at Months 0, 3, and 6 using a balance beam scale.

At Months 0 and 6 only, the following measures were obtained

from all participants: (a) energy intake and nutrient

distribution based on a 3-day, self-report food record; (b)


physical activity level derived from a 3-day activity record;

0(c) general quality of life assessed by the Sickness Impact

Profile, Beck Depression Inventory, Satisfaction Rating

Scales, Satisfaction with Life Scale, and the Perceived Stress

Scales; (d) general medical information; (e) Nutrition and

Exercise Knowledge Questionnaire. General demographic

information and body height were obtained at Month 0 only.

Body height and weight were measured by clinical

psychology graduate students who were supervised by a licensed

clinical psychologist.

Block randomization to condition was conducted according

to diabetes status and gender, because fewer men and fewer

subjects with diabetes were expected to participate in the

study. Subjects were randomly assigned to one of three

conditions: a) weight-loss education condition (14 women; 5

men; 3 individuals with diabetes; M = 216.39 lbs, SD = 32.09);

b) standard behavioral treatment (32 women, 11 men, 8

individuals with diabetes; M = 216.95 lbs, SD = 37.75); or c)

mastery-based behavioral treatment (34 women, 12 men, 10

individuals with diabetes; M = 207.28 Ibs, SD = 32.52). Each

treatment condition was divided into 4 groups (with 12-15

subjects each) and therapists were counterbalanced so that no

one conducted two groups within the same condition. In

addition, lead therapists conducted one group from each


Weight-Loss Education Condition. Subjects in the

Weight-Loss Education Condition received minimal intervention

during the first six months. Intervention involved a group

meeting consisting of a 30- to 45-minute lecture and handouts

provided by a doctoral student or faculty member from the

Department of Clinical and Health Psychology and the

Department of Human Nutrition and Food Sciences. Topics

included information on nutrition and diet, based on

recommendations from the National Research Council (1989);

behavioral modification strategies utilized in standard

behavioral weight loss programs (See Perri, Nezu, & Viegener,

1992); and guidelines for exercise, based on recommendations

by the American College of Sports Medicine (1991). The

information provided was consistent with material presented in

lesson plans for the standard behavioral and mastery-based

treatment conditions and thereby controlled for education on

appropriate ways to manage diet, exercise, and weight.

Subjects in this condition attended six meetings over the six

month period. After the first six months of the study

elapsed, subjects in the weight-loss education condition were

offered the opportunity to participate in a standard

behavioral weight loss intervention. Subjects in this group

received $15 for completing the assessments at 0, 3, and 6

months (total = $45).

Standard Behavioral Treatment Condition. Subjects in the

standard behavioral treatment condition received 26 weekly

treatment sessions. Weekly sessions began with weight

measurement and individual meetings between therapists and

participants to examine self-monitoring records. This was

followed by an educational lecture and group discussion.

Group discussions were structured to enhance understanding of

materials and strategies, problem-solve any difficulties, and

provide social support. Participants received information on

self-monitoring, diet, exercise, and behavior modification

strategies provided through the lectures and a written manual.

Specifically, subjects were instructed in behavior

modification strategies for changing their eating habits,

including procedures for: slowing the pace of eating, reducing

eating cues in the home environment, managing social pressures

to eat, planning for special events, and coping with relapses.

Information regarding nutrition was also included. The manual

predominantly reflects a modification of treatment materials

developed by Perri (1992) for use with weight management

groups. A list of weekly lesson topics may be found in

Appendix B.

Subjects were asked to adhere to an average caloric

intake of 1200 calories per day for women and 1500 calories

per day for men. Subjects gradually worked toward an exercise

goal of walking six days a week, for a minimum total of 180

minutes per week. This goal is consistent with

recommendations by the American College of Sports Medicine

(1991). Subjects also were asked to target a maximal total

fat intake of 30% per day, including a maximum of 10%

saturated fat intake per day, in accordance with guidelines

from the National Cholesterol Education Program (National

Cholesterol Education Program Expert Panel, 1988).

Contingency contracts were established with subjects in

the standard behavioral treatment condition prior to

initiation of treatment. Contracts specified that

participants would be rewarded 1 dollar for attendance and 1

dollar for presentation of complete food records at each

weekly treatment session (excluding the first session), for a

possible total of 50 dollars.

Additional noncontingent reinforcement was provided

through a lottery for prizes that was held during the course

of treatment. Subjects were informed that the lottery was to

be provided to add incentive for involvement in the program.

Each subject in this condition had an equal chance of winning

the lottery.

Mastery-based Behavioral Treatment Condition. Treatment

for subjects in the mastery-based behavioral treatment

condition consisted of the same materials, information,

procedures, type of therapists, and length and amount of

treatment as the standard behavioral treatment condition

described above. The format of each weekly session paralleled

the standard treatment condition. However, individual

meetings between therapists and participants included the

assessment of behavioral skills through a review of eating and

activity records and completion of weekly quizzes. Quizzes

evaluated knowledge of nutrition and behavior modification

strategies, based on information provided in lectures and the

manual. Weekly goals and contingency contracts (i.e., mastery

criteria) were established based on these assessments.

Mastery criteria for weight loss, daily calorie intake,

physical activity, percentage of fats, self-monitoring, and

knowledge of nutrition and behavior strategies were defined

for nine levels that successively approximated desired goals

for diet, exercise, and knowledge. Criteria for Level 1 were

designed to faciliate behavior change easily, with Level 2

though 9 consisting of criteria for mastery that became

increasingly more complex. Subjects moved toward these goals

at individual rates, based upon skill competence and

performance. During the individual meeting at the beginning

of the weekly treatment session, participants' skills were

assessed according to their diet and exercise logs. When all

mastery criteria within a given level were met, the

participant was instructed to progress to the next level

within the program.

Charts of the criteria for each behavioral skill for each

of the nine levels for women and for men are presented in

Appendices C and D, respectively. The categories

selected--weight loss, reduced caloric intake, increased

physical activity, lowered fat intake, knowledge, and

self-monitoring--reflect behavioral skills important for

weight management in prior research. Epstein et al. (1994)

used weight loss, daily calorie intake, physical activity, and

self-monitoring as mastery criteria in his study with obese

children and their overweight parents. Ultimate goals for

calorie intake, fat intake, and exercise were the same as in

the standard condition, but subjects worked toward them in

increments defined by mastery criteria, as described in

Appendices C and D. A minimum number of days on which mastery

criteria for calories, fat, and exercise must be met is

included in order to reduce variability in the performance of

health behaviors.

Mastery criteria for caloric intake consisted of a

maximum average daily intake and the minimum number of days on

which the goal must be met. For example, in Level 3, criteria

for women specified that the daily caloric intake must be no

more than 1400 calories on at least 6 days. The final calorie

goal, specified in Level 9, is a maximal intake of 1200

calories every day of the week. Activity goals changed across

levels in both the amount of minutes per week and the minimum

number of days per week during which exercise must be

conducted, beginning with no exercise and ending with the goal

of 30 minutes per day, six days per week (total = 180

minutes/week). The final goal for total fat intake was to

reduce the percentage of daily fat intake to an maximal total

of 30 percent total fat (or 3 fat exchanges according to


food-exchange lists, American Diabetes Association, 1986) on

6 out of 7 days/week.

Weekly quizzes based on information provided in the

written lessons and lectures were given to assess mastery

criteria for knowledge about nutrition and behavior

modification strategies. Information on the quizzes was

cumulative, with questions retained from earlier tests

combined with new information. Knowledge criteria required

correctly responding to an increasing percentage of questions

across levels, starting at 60% and ending at 90%.

Mastery criteria for weight loss targeted an ultimate

goal of a 14% loss of initial body weight. For the average

person entering the study, this would approximate a loss of 28

pounds. Weight loss criteria were defined by percentage body

weight reduction. Mastery criteria also specified that

participants must continue to self-monitor diet and exercise

behavior 7 days per week throughout the study. To shape

subjects toward more specific monitoring, new elements to

record, such as exercise and fat intake, were added at some


Prior to working on each new level of mastery criteria,

contracts were established with each subject for determining

the contingent reinforcements to be received upon successful

completion of each level of the program. Subjects in the

mastery-based condition were monetarily reinforced 5 dollars

each time a level was successfully completed, for a possible

total of 45 dollars. In addition, subjects also had the

opportunity to participate in the lottery for prizes that was

conducted during treatment. Mastery subjects received one

lottery ticket for each level of mastery criteria completed,

thus allowing them to increase their chances of winning in the

lottery that was conducted during treatment.

Medical Concerns. To minimize the risk of medical

complications during the treatment study, all participants

with diabetes were informed about symptoms of hypoglycemia and

cardiorespiratory problems that they might experience during

participation in the program, particularly during exercise.

Participants were instructed to contact their physician if

they were experiencing any of the symptoms described. In

addition, precautionary measures recommended by the American

College of Sports Medicine (1991) for individuals with

diabetes who exercise, such as conducting proper footcare and

closely monitoring the effect of exercise on blood glucose

levels, were emphasized.

Weight-Loss Therapists. Therapists for both treatment

conditions were Clinical and Health Psychology graduate

students from the University of Florida. Therapists were

supervised on a weekly basis by a licensed clinical

psychologist and were trained in specific skills, such as

behavioral strategies and nutritional information, used for

weight-loss interventions. In addition, therapists were

trained in appropriate assessment skills pertaining to the

study, such as accurate weight measurement and calculation of

caloric intake and fat percentages in diet.

A licensed dietician served as a consultant for the

program throughout the 6 months. In addition, two graduate

students from the Department of Human Nutrition and Food

Sciences, supervised by the licensed dietician, served as

co-leaders for the treatment groups. Their responsibilities

included teaching the lessons pertaining to nutrition,

reviewing food records, and providing consultation on

nutrition-related issues.



Data Analytic Strategy

Repeated measures multivariate analyses of variance

procedures (MANOVAs) were used to analyze differences between

conditions in changes in the dependent variables over time.

Specifically, MANOVA procedures were applied to evaluate the

effects of the mastery-based intervention on performance of

specified self-management skills, weight loss, and quality of

life relative to the weight-loss education group, as well as

differences in treatment effects between the mastery-based

condition and the standard behavioral treatment condition.

Treatment condition was used as a between-group factor and

time (Months 0, 3, and 6) was used as a repeated measure.

MANOVA procedures were used for all repeated measure analyses

because of the lesser risk of a Type I error as compared to

repeated measure ANOVA models (O'Brien & Kaiser, 1985).

Significant MANOVA effects were evaluated further with

analysis of variance procedures (ANOVAs) and

Student-Newman-Keuls tests. MANOVA and ANOVA procedures were

found to yield the same results when examining differences

between conditions on variables that did not involve a time

factor. Therefore, ANOVA procedures were used when comparing


groups on variables at a single time period (e.g., initial

body weight). Correlational analyses were used to evaluate

the strength of association between variables. Finally,

Chi-Square tests were used to examine differences between the

two treatment conditions in the proportion of individuals who

benefitted from treatment, according to differences in the

dependent variables.

Preliminary Analyses and Attrition

ANOVA procedures indicated that there were no significant

pretreatment differences between conditions on initial body

weight (ANOVA F (2,103) = .91, p = .41), percentage overweight

(ANOVA F (2,103) = .58, p = .56), or body mass index (ANOVA F

(2,103) = .68, R = .51). Of the 106 subjects who began the

program, 22 dropped out during treatment, yielding an

attrition rate of 20.8%. Subjects who dropped out did not

differ significantly at pretreatment from subjects who

completed the program in initial body weight (ANOVA F (1,104)

= 2.32, E = .13), percentage overweight (ANOVA F (1,104) =

.13, p = .72), or BMI (ANOVA F (1,104) = .14, E = .71).

Drop-outs also did not differ significantly in initial

self-reported levels of depression or stress, as indicated by

the Beck Depression Inventory (ANOVA F (1,101) = .10, p = .76)

and the Perceived Stress Scales (ANOVA F (1,101) = 1.17, p =

.28), compared to active participants. The number of

drop-outs was not statistically associated with treatment

condition (Xi = 3.44, df = 2, p > .10). Although not

statistically significant, there was a trend toward a higher

drop-out rate in the mastery-based condition (28.9%) compared

to the standard behavioral condition (16.7%) and the weight-

loss education condition (10.5%).

Analyses of Adherence-Related Measures

Table 1 presents means and standard deviations for the

four primary adherence variables at pre- and posttreatment,

Weeks 1 and 26, respectively: daily average calorie

consumption, daily average calories from fat, average weekly

exercise, and amount of knowledge about weight-loss strategies

(i.e., nutrition, exercise, and behavior change strategies)

for the mastery-based, standard behavioral, and weight-loss

education conditions. Both calorie and fat data were based on

computer nutritional analysis of self-report food records.

Exercise data were based on self-report information that was

not obtained from the weight-loss education condition.

Analyses of self-report food records using repeated

measures MANOVA procedures indicated that both the

mastery-based and the standard behavioral treatment conditions

significantly decreased caloric intake over time [MANOVA

F(1,55) = 14.4, E = .0001], with no significant interaction

effect for condition over time [MANOVA F(2,53) = .07, E =

.80]. Results of a repeated measures MANOVA of average daily

caloric intake based on computer analyses of food records for

the three conditions indicated (a) a significant main effect

for condition, MANOVA F(2,65) = 7.3, E = .001; (b) a

Table 1
Means and Standard Deviations of Adherence Variables at Pre-
and Posttreatment for the Mastery-based, Standard Behavioral,
and Weight-loss Education Conditions

Variable M (SD) M (SD) M (SD)

Total Calories
Pre-Tx 1989.8 (658.7)x,X 2233.7(595.4)x,X
Post-Tx 1314.1 (415.4)x,Y 1720.0(573.3)x,

Calories From Fat
Pre-Tx 653.4 (260)x,X
Post-Tx 370.8 (140)xy

Calories From Saturated Fats
Pre-Tx 240.3 (101) x,
Post-Tx 132.3 (56),Y

705.6 (292) x,
485.1 (195)xy

253.8 (125) ,X
171.9 (88)x,y

2269.5 (610.4)x,x
2301.1 (974.6)y,

807.3 (281)x,x
869.4 (413)y,

306.9 (141)xX
324.9 (173)y

Exercise (minutes)
Pre-Tx 120.4 (74.3)xX
Post-Tx 212.5 (98.6)x,,


127.5 (107.8)xX
192.8 (99.9)x,y

19.7 (2.5)x,x
22.0 (2.0)x,y

19.5 (3.0)x,X
21.4 (2.7)x,y

18.5 (2.3)x,X
21.0 (2.1)x,y

Note. Dissimilar lower case subscripts indicate significant
differences between conditions within a time period, and
dissimilar upper case subscripts indicate significant
differences within a condition across time periods (p < .05).

significant main effect for time, MANOVA F(1,65) = 17.3, R =

.0001; and (c) a significant interaction effect for Condition x

Time, MANOVA F(2,65) = 4.6, E = .01. Subsequent one-way

analysis of variance and Newman-Keuls post hoc comparisons

indicated that both the mastery and the standard conditions

demonstrated a significantly greater reduction in caloric intake

compared to the weight-loss education condition from pre- to

posttreatment, but did not differ significantly from one another

(ANOVA F(2,65) = 4.6, E < .05). In addition, there were no

significant differences in daily caloric intake between the

mastery and standard conditions during any month of treatment

(Es > .10). Figure 1 illustrates changes in average daily

caloric intake over time for the mastery and standard


Results of a repeated measures MANOVA of average daily fat

intake based on computer analyses of food records for the three

conditions indicated (a) a significant main effect for

condition, MANOVA F(2,63) = 13.5, R = .0001; (b) a significant

main effect for time, MANOVA F(1,63) = 11.8, R = .001; and (c)

a significant interaction effect for Condition x Time, MANOVA

F(2,63) = 6.2, p = .004. Subsequent one-way analysis of

variance and Newman-Keuls post hoc comparisons indicated that

both the mastery and standard conditions demonstrated

significantly greater reductions in total fat intake over time

than the weight-loss education condition (ANOVA F(2,65) = 6.2,

o 0 0 0
o iC 0 It
m rV cV) C













I -




E = .004), but no significant differences were observed between

the two behavioral treatment conditions (R > .05).

Figure 2 demonstrates changes in physical activity over

time for the standard and mastery-based conditions. MANOVA

procedures used to analyze changes in total weekly exercise

yielded a significant main effect for time, MANOVA F(1,54) =

19.7, R = .0001. No significant main effects were observed for

condition (MANOVA F(1,54) = .04, R = .84) or Condition x Time

(MANOVA F(1,54) = 2.4, R = .13). That is, both the mastery and

standard conditions significantly increased the amount of total

weekly exercise during treatment, but there was no significant

difference between them. Examination of the Physical Activity

Record data using MANOVA procedures revealed a significant

change in total energy expenditure over time for the three

conditions (MANOVA F(1,70) = 11.57, E = .001), but no difference

between them (MANOVA F(2,70) = 0.05, R = .95).

Subjects in all three conditions showed statistically

significant increases in knowledge of weight-loss strategies

over time [MANOVA F(1,75) = 49.9, E = .0001], although actual

amount of improvement was small. Specifically, changes in means

reflected correctly answering one or two more questions at

posttreatment than answered at baseline. No differences between

the 3 conditions in changes in knowledge over time were

demonstrated [MANOVA F(2,75) = .28, E = .76].

Additional analyses were conducted to further explore

adherence data. Correlational analyses were calculated to

0 0
2 *0 58
E to

c +

- "----------~ <0r



0 c

0 C1j m
\. g

.. o

\) -


g .o

0.. O

compare the average daily caloric intake derived by subjects and

that derived by a computer nutritional database. Self-report

measurements of average daily calories based on subject

calculation and on computer analysis were found to be

significantly positively correlated both at baseline (r = .44,

R < .001) and at posttreatment (r = .63, E < .001).

Correlational analyses applied to exercise measures

indicated that the Physical Activity Record was not correlated

with self-report of daily exercise in minutes either at baseline

(r = .01, p > .05) or at posttreatment (r = -.08, p > .05). One
potential reason for the lack of association may be that the two

instruments measured different things: the Physical Activity

Record assessed total energy expenditure throughout the day and

the self-report only measured focused exercise time. However,

it is reasonable to assume that total energy expenditure would

be boosted by demonstrated increases in amount of exercise. In

addition, subjects were observed to be inaccurate in their use

of the scale on the Physical Activity Record for estimating the

amount of energy expended in a given activity. The validity and

utility of the Physical Activity Record, therefore, was


MANOVA procedures used to examine changes in daily

saturated fat intake indicated a) a significant main effect for

condition, MANOVA F(2,63) = 12.1, p = .0001; (b) a significant

main effect for time, MANOVA F(1,63) = 8.7, E = .004; and (c) a

significant interaction effect for Condition x Time, MANOVA

F(2,63) = 4.9, p = .01. Subsequent one-way analysis of variance

and Newman-Keuls post hoc comparisons indicated that there were

no significant differences in changes in saturated fat intake

between the two behavioral treatment approaches. Data also

indicated that only the mastery-based behavioral condition

demonstrated a significantly greater reduction in saturated fat

intake over time compared to the weight-loss education condition

(E < .05). Examination of the means reveals that the education

condition actually demonstrated a marginal increase in saturated

fat intake over time.

Analyses of Weight-Related Measures

Table 2 presents means and standard deviations for body

weight and body mass index at Months 0, 3, and 6 for each

condition. Figure 3 demonstrates the changes in weight over

time for the three conditions. Results of a repeated measures

MANOVA comparing changes in weight over time indicated (a) no

main effect for condition, MANOVA F(2,81) = 1.6, E = .22; (b) a

significant main effect for time, MANOVA F(2,80) = 68.5, E =

.0001; and (c) a significant interaction effect for Condition x

Time, MANOVA F(4,162) = 8.1, R = .0001. Use of repeated

measures MANOVAs to analyze changes from pre- to post-treatment

in body mass index and percentage overweight, respectively,

demonstrated the same pattern of nonsignificant main effects for

condition [MANOVA F(2,81) = 2.31, E = .11; F(2,81) = 2.01, p =

.14], a significant effect for time [MANOVA F(1,81) = 123.6, E

Table 2
Means and Standard Deviations of Body Weight and Body Mass Index
at Months 0, 3, and 6 for the Mastery-based, Standard
Behavioral, and Weight-Loss Education Conditions (in pounds)

Month 0

Month 3

Month 6

M (SD)

206.1 (34.0)x,x

193.5 (34.3)x,Y

188.7 (34.7)x,Z

Month 0 34.4 (4.8)x,X

Month 3 32.3 (4.7)x,y

Month 6 31.5 (4.7)x,z

M (SD)

213.9 (34.7)x,X

199.5 (32.3), y

193.0 (32.8)x,Z

34.1 (4.0)x,X

31.8 (3.6)x,

30.8 (3.8)x,Z

M (SD)

215.6 (33.6)x,X

213.8 (33.6)x,X

211.9 (33.2)y,

35.4 (4.4)x,X

35.1 (4.7)y,x

34.8 (4.5)yx

Note. Dissimilar lower case subscripts indicate significant
differences between conditions within a time period, and
dissimilar upper case subscripts indicate significant
differences within a condition across time periods (p < .05).



0 aM



/ U

I0 (


0 / 0 0 LO 0 O
0 0 0) )
I / a a E


C /E -I

S. 0 0-,a'
S/ E0
r ;,/ 0 0

0 0 0 0) a a -
.0 0
au /4 au 0u c ro

= .0001; F(1,81) = 121.6, = .0001], and a significant

interaction effect of Condition x Time, [MANOVA F(2,81) = 13.91,

p = .0001; F(2,83) = 14.1, R = .0001].

Analysis of variance and student Newman-Keuls post hoc

comparisons were used to examine significant MANOVA effects.

Both the standard and the mastery treatment conditions

demonstrated significantly greater changes in weight, BMI, and

percent overweight over time than the weight-loss education

condition at posttreatment (Ps < .05). No significant

difference in weight loss [ANOVA F(1,65) = 1.56, P = .22],

change in BMI [ANOVA F(1,65) = .70, E = .41], or percent

overweight [ANOVA F(1,65) = .95, P = .33] was demonstrated

between the standard and mastery-based treatment conditions.

Additional analyses were conducted to examine the influence

of drop-outs, subjects' sex, monetary reinforcement, and

therapists on weight-related dependent variables. A repeated

measures MANOVA was used to analyze the weight data for all

subjects who began the study. To complete missing weight data,

the most recent clinical weight obtained from drop-outs was

assigned as the posttreatment weight. The same pattern of

significant effects were demonstrated as in the analyses that

excluded the subjects who dropped out of the study.

Specifically, both the mastery and standard conditions made

significantly greater changes in weight, BMI, and percent

overweight during the course of treatment than the weight-loss

education condition (ps < .05), but the behavioral treatment

conditions did not differ from one another.

ANOVA procedures indicated a significant difference in

initial body weight between males and females [E(1,65) = 22.7,

P = .0001], but not in BMI [F(1,65) = .07, p = .78] or initial

percent overweight [F(1,65) = 2.0, p = .16]. No difference was

demonstrated in weight loss [F(1,65) = .18, R = .67] between

males and females. When controlling for initial body weight,

sex differences in changes in weight loss [F(1,64) = .41, R =

.52], BMI [F(1,64) = 1.9, E = .17] and percent overweight

[F(1,64) = 1.5, p = .23] remained nonsignificant. Results

indicated a significant difference in the amount of monetary

reinforcement received by subjects within the two treatment

conditions [ANOVA F(1,65) = 18.8, R = .0001], with subjects in

the standard treatment group receiving more money (M = 41.11

dollars) than subjects in the mastery group (M = 31.59 dollars).

Although there was a significant correlation between weight loss

and monetary reinforcement (r = .57, p < .001), differences in

weight loss between the standard and the mastery conditions

remained nonsignificant when monetary reinforcement was

co-varied in the ANOVA procedure [F(1,64) = 1.8, p = .18].

ANOVA procedures were used to explore potential therapist

effects by comparing differences on weight variables between

nights that subjects received treatment. Significant

differences in weight loss [F(3,63) = 3.92, R = .01] and in

changes in BMI [F(3,63) = 3.24, p = .03] over time were observed

between the nights that subjects received treatment, suggesting

a therapist effect on weight loss. One-way analysis of variance

and student Newman-Keuls post hoc comparisons were used to

examine differences within the four groups of the mastery

condition. Results indicated that one group did significantly

poorer than the other three groups in the amount of weight and

BMI change made during treatment (ps < .05). No significant

differences were found between groups in the standard condition

(Rs > .05). To eliminate potential therapist effects, weight

analyses were calculated excluding the poorest performing

mastery group. Analyses demonstrated nonsignificant differences

between the treatment conditions for changes in weight [MANOVA

F(1,59) = .17, p = .69], BMI [MANOVA F(1,59) = .01, p = .99],

and percent overweight [MANOVA F(1,59) = .03, R = .87]. Average

weight losses for the two conditions became less discrepant,

however, (M = 19.7 pounds for Mastery, M = 20.9 pounds for

Standard) and more consistent with previous behavioral outcome


Chi-square procedures were conducted to compare proportions

of mastery and standard subjects who responded to treatment.

There was no significant difference between conditions in the

proportion of subjects who lost 20 or more pounds (X2 = 1.25, E

> .10); however, 41% of the mastery subjects lost at least 20

pounds compared to 54% of the subjects in the standard


Analyses of Quality of Life Measures

Table 3 contains mean values for quality of life variables

at baseline and at 6-months for the three conditions. ANOVA

procedures indicated that, at baseline, the weight-loss

education condition evidenced significantly poorer global

functioning [ANOVA F(2,80) = 5.6, E = .01], lower life

satisfaction [ANOVA F(2,80) = 3.6, E = .03], and more depressive

symptomatology [ANOVA F(2,80) = 3.5, p = .04] compared to the

two treatment conditions. Analyses conducted subsequent to

excluding outliers, defined as those records whose SIP score

exceeded 25 (N = 2), resulted in no significant differences

between conditions at baseline (ps > .10). Therefore,

subsequent analyses of quality of life variables were conducted

without the two records containing outliers. MANOVA and

subsequent univariate analyses revealed significant or

marginally significant improvements in global functioning, as

measured by the Sickness Impact Profile, in all three conditions

(Es > .10). In addition, there was significant improvement over

time in general life satisfaction (Es < .10), depression (Es <

.05), and stress (ps < .10) for the mastery and standard

conditions. None of the three conditions demonstrated a

clinically significant elevation in global functioning or

depression at baseline, however. The mastery and standard

conditions showed a greater increase in life satisfaction than

the weight-loss education condition [ANOVA F(2,74) = 4.7, E =

.01]. The three conditions did not significantly differ in

Table 3
Mean Values of the Quality of Life Variables for the
Mastery-based, Standard Behavioral, and Weight-Loss Education
Conditions at Pre- and Posttreatment

M (SD) M (SD) M (SD)
Pre-Tx 4.5 (3.8)x,X 4.1 (4.5)x,x 5.6 (7.6)x,
Post-Tx 0.7 (3.4)x,y 2.6 (3.9)x,X 0.8 (5.4)x,y

Pre-Tx 27.0 (5.4)x,x 25.7 (4.7)x,x 24.0 (7.7)x,X
Post-Tx 32.6 (4.5)x,y 29.7 (7.5)x,y 23.8 (8.5)y,

Pre-Tx 23.7 (5.9)x,x 20.9 (6.7)xx 20.8 (8.5)x,X
Post-Tx 25.9 (6.2)x,x 23.5 (7.3)x,y 21.1 (7.9)x,

Pre-Tx 7.1 (5.7)x,x 8.7 (6.6)x,x 8.3 (6.2)x,
Post-Tx 3.3 (3.5)x,y 6.4 (6.5)x,y 8.0 (6.5)x,

Pre-Tx 22.3 (8.2)x,X 22.7 (8.5)xX 22.5 (8.7), x
Post-Tx 18.7 (6.9)x,y 20.2 (9.0),x 23.0 (10.9)x,X

Note. SIP = Sickness Impact Profile; SATIS = Satisfaction Rating
Scale; SWLS = Satisfaction with Life Scale; BDI = Beck
Depression Inventory; PSS = Perceived Stress Scale. Dissimilar
lower case subscripts indicate significant differences between
conditions within a time period, and dissimilar upper case
subscripts indicate significant differences within a condition
across time periods (p < .05).

changes on any other psychological variables over time (Rs >

.05). Ratings of self-efficacy significantly increased for

both the standard and mastery conditions over time [MANOVA

F(1,59) = 22.4, E = .001], but there was no significant

difference between them [MANOVA F(1,59) = .002, R = .96].

Analyses of the Effect of Diabetes Status

ANOVA procedures indicated that the subjects with

diabetes were significantly heavier at the onset of treatment

than the nondiabetic subjects, as measured by initial body

weight [F(1,104) = 6.3, p = .01]. Therefore, to examine

potential differences in weight loss between diabetic and

nondiabetic subjects, the 13 subjects with diabetes from the

two treatment conditions were matched on gender, initial body

weight, initial percentage over ideal weight, and treatment

condition with 11 nondiabetic subjects and compared for

differences in response to treatment. ANOVAs indicated that

there were no significant differences between the two samples

in initial body weight [(F(1,22) = .002, R = .97], percentage

overweight [E 1,22) = .26, E = .62], initial body mass index

[F(1,22) = .68, R = .42], or age [F(1,22) = .10, E = .75].

Table 4 presents means and standard deviations for

weight-related variables for the matched sample of diabetic

and nondiabetic subjects at pre- and posttreatment.

MANOVA procedures demonstrated no significant difference

in amount of weight loss [F(1,22) = .210, E = .65], BMI change

[F (1,22) = .02, E = .88] or percent overweight change

Table 4
Means and Standard Deviations for Weight-Related Variables for
Diabetic and Nondiabetic Subjects at Pre- and Posttreatment

M (SD}

M (SD)




232.1 (34.3)xX
210.3 (30.6)x,y

34.6 (5.0)x,X
31.3 (4.4)x,y

55.1 (21.1)xX
40.6 (18.8)x,'y

211.9 (40.7)x,y

36.2 (4.8)x,X
33.1 (4.5)x,y

59.5 (21.9)x,X
45.9 (21.3)x,y

Note. Dissimilar lower case subscripts indicate significant
differences between conditions within a time period, and
dissimilar upper case subscripts indicate significant
differences within a condition across time periods (p < .05).


[F(1,22) = .06, E = .82] for the subjects with diabetes

compared to the nondiabetic matched sample. Additional

exploratory analyses were conducted to further examine

relationships between the two groups on adherence-related and

quality of life variables. ANOVA procedures revealed no

significant posttreatment differences in caloric intake

[F(1,19) = .14, p = .71] or total fat intake [F(1,17) = .40,

E = .54] between the two groups. Subjects who did not have

diabetes, however, demonstrated a marginally greater increase

in exercise over time [MANOVA F(1,18) = 3.4, p = .08]. On

quality of life measures, although subjects who did not have

diabetes reported significantly more depressive symptoms at

baseline (ANOVA F(1,23) = 12.4, E = .002), symptoms in this

group diminished during treatment so that no significant

difference between groups was noted at posttreatment (ANOVA

(1,22) = 2.4, E = .14).

Subjects with Type II diabetes were queried at

posttreatment, using a self-report questionnaire, about

changes in their diabetes management during the weight loss

program. Ninety-two percent (12/13) of the subjects in the

two treatment conditions reported an improvement in their

diabetes during the weight loss program, with 77 percent

(10/13) of these describing positive changes in their

medication regimen. Specifically, 1 subject discontinued

insulin, 1 decreased the insulin dosage, 5 discontinued oral

medications, and 3 decreased oral medication dosages. None of


the three subjects with diabetes in the Weight-loss Education

condition reported changes in their diabetes status over time.

Additional Analyses of the Mastery-based Condition

Figure 4 presents percentages of the number of

individuals at each mastery level at Months 3 and 6. The mean

mastery level completed at the end of the 6-months was 7, with

a mode (31.3%) at Level 5. Only three subjects reached Level

10, the highest level obtainable. To explore potential

discriminating factors between subjects who excelled in the

mastery condition and those who did not perform as well, ANOVA

procedures were used to compare the subjects within the

mastery condition who reached Mastery Level 8 or higher to

those who reached Level 7 or lower. Results indicated that

subjects did not differ on initial weight variables,

demographic variables (age, education, and gender), or

diabetes status (ps > .10). On adherence variables, subjects

who reached higher mastery levels reported significantly more

cumulative minutes of exercise (ANOVA F(1,22) = 4.3, E = .05)

and better attendance (ANOVA F(1,22) = 24.3, E = .0001) over

time compared to those who reached Level 7 or lower. There

was not a significant difference in cumulative caloric intake

over time between the two groups (ANOVA F(1,22) = .17, p =


Given that progress through the mastery levels is

sequential and cumulative (that is, a cross-sectional look at

a subjects' adherence at a particular level reflects adherence

I j













J C1)






u- It


iL .-

0 0 LO 0 LC 0 O
C) C() C C I- T-


to criteria and progress through all proceeding levels),

comparisons between the mastery and standard conditions on

mastery criteria at a particular point in time will not

provide a fully accurate picture of possible differences

between them. However, an examination of differences in the

proportions of subjects within each condition that achieved

the calorie and exercise goals for a particular level at a

given time can suggest variability in adherence to skills

between the two conditions. Chi-square analyses were

conducted to examine discrepancies in the number of subjects

who adhered to mastery criteria for exercise and caloric

intake for Level 7 (the mode at month 3 and mean at month 6)

at months 3 and 6. Results indicated that a marginally

greater proportion of female mastery subjects averaged fewer

than or equal to 1200 calories per day compared to standard

condition subjects at month 3 (X2 = 3.18, E = .07), but not

at month 6 (X2 = .16, R = .69). There was no difference

demonstrated in the proportion of male mastery subjects who

averaged fewer than or equal to 1500 calories per day compared

to the standard male subjects at month 3 (X2 = 1.2, R = .28),

but mastery subjects averaged marginally fewer calories at

month 6 (X2 = 2.9, R = .09). There were also no differences

in those exercising at least 180 minutes per week at month 3

(X2 = .02, E = .89) or month 6 (X2 = .83, E = .36).

Univariate analyses indicated that subject satisfaction

ratings with treatment and therapists at posttreatment were

slightly higher for the mastery condition compared to standard

treatment (t = 1.9, df = 6, R < .10). Neither subjects'

expectancies about treatment nor satisfaction ratings with the

program were correlated with weight loss (Expectancy: r = .08,

R > .05; Satisfaction: r = -.08, p > .05).



The primary objective of this study was to determine if

adding mastery criteria to a behavioral weight loss program

would enhance response to treatment. Results indicated that

individuals in the two behavioral treatment conditions lost an

average of 20.3 pounds over the initial six month period,

which is consistent with the literature on outcomes in

behavioral treatment programs (Brownell & Wadden, 1992).

Individuals in both behavioral conditions also demonstrated

significant reductions in total caloric and fat intake,

increases in exercise, increases in general life satisfaction,

and reductions in stress. Although both behavioral

interventions produced significant positive responses in terms

of adherence to self-management skills, amount of weight lost,

and quality of life, the mastery approach did not differ

significantly from the standard behavioral approach in


Based on theoretical and empirical data, hypotheses had

predicted that the addition of mastery criteria would enhance

effects of a behavioral weight loss intervention by

facilitating the acquisition of self-management skills

necessary for effecting changes in lifestyle and,

consequently, in weight. In addition, the mastery approach

entailed pacing goals and reinforcement according to

individual rates of progress, thereby creating a more

individualized approach within a group setting.

Both the effectiveness of the behavioral interventions

and the inability of the mastery approach to improve outcome

can be understood according to the conceptual model of health

behavior change proposed for this project. The model

postulated that manipulation of certain mediating variables,

such as skills and knowledge, would affect the likelihood of

producing changes in eating and exercise behaviors that would

yield desired biological (e.g., weight loss) and behavioral

outcomes (e.g., quality of life). Both the standard and

mastery-based interventions targeted teaching skills and

knowledge and successfully produced changes in eating and

exercise behaviors. As predicted by the model, changes in

these health behaviors yielded the desired biological and

behavioral outcomes. The inability of the mastery approach to

improve outcome over standard treatment suggests that there

was no significant difference in the ability of the two

interventions to manipulate the variables responsible for

effecting behavior change. That is, it appears that the

mastery-based approach was not better than the standard

treatment in manipulating factors that lead to changes in

weight-related behaviors.


Several factors may have contributed to the inability of

the mastery criteria to bolster initial treatment effects.

First, although the purpose of breaking down end goals for

diet and exercise into smaller increments in the mastery

condition was to facilitate acquisition of skills at an

individual rate, the result was that subjects in the mastery

condition were monetarily reinforced for their skill mastery

on an intermittent schedule and subjects in the standard

condition were reinforced for their attendance and homework

completion on a continuous schedule (i.e., each week). As a

result, subjects in the standard condition, regardless of

performance, were reinforced for attending meetings while

subjects in the mastery-based condition often went weeks of

attending meetings without receiving monetary reinforcement.

In addition, subjects in the standard condition ultimately

received significantly more monetary reinforcement than the

mastery condition. Thus, mastery subjects did not receive as

consistent, or as frequent, or as much, of a monetary

reinforcement during treatment. Qualitative observations by

therapists indicated that the result of this difference was

that mastery subjects experienced disappointment and

frustration when they could not achieve mastery levels,

expressed concerns about getting their money back, and had

less incentive to attend meetings when they were not doing

well because they would not be reinforced for coming.


That the modal mastery level achieved at posttreatment

was Level 5, only midway to the highest level, indicates that

the levels were too difficult to foster consistent, frequent

progression. In particular, clinical observation indicated

that the mastery criteria for weight loss was frequently the

"skill" that subjects were unable to achieve and, thus,

prevented them from passing to the next level. This was

particularly problematic because, unlike the other skills,

weight loss was not a behavior that was under the subjects'

direct influence. Weight loss had been included in order to

maintain similarity to the original study conducted by Epstein

(1994). However, future research should consider limiting

skills to those behaviors that the subject can modify

directly. Other methods for increasing the frequency of

reinforcement, such as by making the mastery levels easier to

achieve and by adding other continuous reinforcers, will also

be important modifications for future studies and critical for

maintenance of changes (Epstein, 1992).

A second factor that may have contributed to the lack of

an experimental effect is that, given the consistent finding

of an approximate 20 pound weight loss following behavioral

interventions, results may reflect a possible ceiling effect.

That is, it may be that the maximum initial weight loss (i.e.,

after 6-months) obtainable with conservative approaches has

been reached and can only be enhanced either by introducing

more vigorous strategies, such as incorporating the use of

very-low-calorie-diets (VLCDs) in initial treatment, or

utilizing techniques that will improve maintenance of new

behaviors and initial weight loss after the initial treatment

phase. This conclusion suggests a third reason why no added

effect of mastery criteria was evidenced. That is, the

objective in the mastery approach was to increase adherence of

diet and exercise behaviors through enhanced skill

acquisition, thereby making it less resistant to extinction.

The positive effect of using mastery criteria, therefore, may

not emerge until after the initial treatment phase, when

maintenance of behaviors and weight loss are critical.

Comparison of the two behavioral conditions at one year will

provide evaluation of this hypothesis.

These results differed from those of Epstein et al.

(1994), who found that the addition of mastery criteria

significantly improved treatment outcome compared to a

standard behavioral treatment condition. Several factors may

account for the differences in results. First, the former

study was a family-based treatment for obese children with a

sample that consisted of children and their parents, sixty-one

percent of whom were also overweight. Parenting skills were

a significant additional component used to reinforce

children' behaviors and weight loss. In addition, both

parent and child had to demonstrate mastery of skills and

weight loss to progress to the next level. These differences

may have produced a differential effect because of the

interaction between technique and context. It may be that the

mastery criteria is particularly helpful in enhancing

adherence within a family context and not as effective with

individual adults. Dissimilarities in reinforcement schedules

between the two studies may also have contributed to

differences. In Epstein's study, subjects in the control

group moved through levels yoked to the mastery group and were

provided with equal, but noncontingent, reinforcement. As a

result, mastery subjects in Epstein's study may have received

more frequent reinforcement than mastery subjects in the

present study and may have more closely approximated the

amount of reinforcement received in its control group.

Finally, it has been hypothesized that the mastery condition

in the present study may demonstrate better maintenance than

the standard condition at follow-up, which would be consistent

with follow-up data from the Epstein study. Therefore, it may

be that, over time, data from the present study will become

more consistent with results from earlier research.

Although results did not support the primary hypothesis

that the mastery-based approach would enhance adherence of

self-management skills, subjects in both of the behavioral

treatment conditions significantly reduced caloric intake over

time. In addition, further analysis revealed that both

behavioral conditions demonstrated significant reductions in

total fat and saturated fat during the program. Average daily

total fat intake was reduced from 75.5 grams to 47.6 grams for

the behavioral treatment conditions. The weight-loss

education condition did not reduce caloric intake or

demonstrate positive changes in diet composition.

Computer analyses of diet composition using a nutrient

database such as utilized in this study represents a critical

addition to the evaluation of obesity treatment outcomes. The

nutritional analyses allowed for evaluation of specific

targeted changes in diet composition, including reductions in

fat and saturated fat, in addition to general reductions in

caloric intake. The importance of specific changes in

diet--particularly in reducing fat--for improving health

status as well as effecting weight loss has been well

documented in research (Robinson et al., 1993; Willett,

Stampfer, Colditz, Rosner, Hennekens, & Speizer, 1987).

Reduction in fat is routinely targeted in weight-loss

interventions but infrequently measured systematically in

outcome. Because of the importance of diet composition both

for producing weight loss and improving health status, use of

nutritional analyses in evaluating behavioral treatment

outcomes should be continued.

The fact that subjects statistically, but not

clinically, increased their knowledge of nutrition is likely

attributable to the general information surveyed on the test.

Average scores across conditions were high at baseline,

leaving little room for improvement. These data suggest that

most individuals had general knowledge about eating management

when entering the program, but needed intervention to be able

to effectively apply their knowledge. This is consistent with

the literature indicating that knowledge is necessary but not

adequate for facilitating behavior change. In addition, it

may be that the instrument was not sophisticated enough to

measure important changes in diet-related information over


When data were examined to further evaluate differential

responses to treatment on weight-related variables, it was

observed that the proportion of individuals who successfully

lost weight did not differ between the standard and mastery

approaches. In addition, contrary to prior research (e.g.,

Kramer, Jeffery, Forster, & Snell, 1989; Wing, 1992), the

current study did not find gender effects in response to

treatment. Men and women demonstrated equivalent changes on

weight-related variables, when initial weight was controlled


The self-reports in increases in life satisfaction and

reductions in depressive symptoms and perceived stress

indicate that the behavioral interventions positively affected

quality of life in participants. These positive changes in

mood are consistent with the literature (Wing et al., 1984).

It is notable that average scores for the conditions indicated

that subjects did not evidence clinical elevations in global

functioning or depression at baseline. The lack of change

between the treatment conditions on psychological variables,

therefore, is likely attributable to a floor effect, in that

participants did not evidence significant psychological

distress initially. The increase in self-efficacy experienced

by subjects in both the standard and mastery conditions

indicated that the behavioral treatments were effective in

boosting individuals' belief in their ability to perform the

behaviors necessary for weight loss. Given that self-efficacy

is an important factor in effecting behavior change (Bandura,

1989; Ewart, 1989), these data provide further indication of

treatment effectiveness.

A secondary goal of this research was to examine the

efficacy of a minimal treatment intervention compared to the

behavioral treatment programs. Results indicated that the

mastery-based and standard approaches were more effective than

weight-loss education alone in producing positive changes in

weight-related variables, increasing adherence to diet and

exercise goals, and increasing general life satisfaction.

Minimal treatment studies have typically compared results to

10-week behavioral treatment programs and found no significant

differences in efficacy (Black & Threlfall, 1986; Marston,

Marston, & Ross, 1977). The current study, however, indicated

that a minimal intervention that provided basic information on

diet, exercise, and behavior strategies performed

significantly poorer than behavioral treatment programs.

Subjects in the weight-loss education condition lost an

average of only 3.7 pounds, and--unlike the behavioral


interventions--subjects in the education condition did not

reduce caloric or fat intake and did not increase life

satisfaction or reduce stress over time. It is likely that

previous studies did not demonstrate differences between

minimal interventions and behavioral programs in weight loss

because treatment was not long enough to produce the effects

found in standard behavioral treatments.

Subjects in the weight-loss education condition

demonstrated significantly higher levels of psychological

distress than the two behavioral conditions. One possible

explanation is that baseline measures of psychological

functioning were assessed after randomization to conditions.

Therefore, the subjects knew at the time of the baseline

assessment whether or not they would be receiving treatment

immediately or in six months, and this may have contributed to

higher or lower scores on self-report measures of distress.

However, removal of two outliers from this condition resulted

in nonsignificant differences in baseline psychological

measures between the three conditions.

Special effort was made to recruit subjects with Type II

diabetes for the project because of the additional health

benefit of weight loss for this population. Comparison of

outcome for diabetic and nondiabetic subjects indicated that

diabetes status did not have an effect on response to the

weight loss treatment. Specifically, individuals with

diabetes were equivalent to a matched subsample of nondiabetic


subjects in changes in weight, body mass, and percentage over

ideal body weight. These results contradict two prior studies

of the effect of diabetes status on weight that had suggested

that individuals with diabetes lose less weight than

nondiabetic individuals (Henry et al., 1985; Wing et al.,

1987). In addition, contrary to findings of the Wing et al.

study (1987), results from the current study revealed no

significant differences in reductions in caloric intake

between the two groups. Subjects who did not have diabetes,

however, did demonstrate a marginally greater increase in

exercise over time. Although subjects who did not have

diabetes reported experiencing more depressive symptoms at

baseline than subjects with diabetes, symptoms diminished

during treatment so that no significant difference between

groups was noted at posttreatment. Finally, although

laboratory studies of metabolic control were not obtained,

subjects' reports suggested positive changes in diabetes

management during treatment. Specifically, all but one

subject diagnosed with Type II diabetes from the two

behavioral treatment conditions reported an improvement in

diabetes management at the end of the initial treatment phase,

with the majority reporting beneficial changes in medications,

including five who discontinued medications altogether.

One possible explanation for discrepancies between these

data and that of Wing et al. (1987) is that the latter used

nonindependent spouse pairs, which may have produced a

reactivity effect. For example, spouse pairs in a weight loss

program may provide support or competition to one another that

might influence adherence and outcome, thereby confounding

potential reasons for differences between the diabetic and

nondiabetic groups. The present study made some important

improvements in this area by using independent pairs who were

matched on age, initial weight variables, and gender. When

independent pairs were compared, no differences in weight loss

were found.

Another possible reason for the discrepancies in results

may be attributable to varying distributions of medication use

among subjects between the different studies. For example,

the current study had fewer subjects taking insulin and more

subjects taking oral medications compared to the Wing study

(1987). Studies have yielded conflicting results about the

effect of diabetes medications on weight loss (Harris,

Davidson, & Bush, 1988; University Group Diabetes Project,

1971; Wing, Shoemaker, Marcus, McDermott, & Gooding, 1990);

thus, this remains an important potential confound to explore

when examining the effect of diabetes status in future


Finally, although these data suggest that individuals

with diabetes can demonstrate results equal to nondiabetic

individuals in a behavioral weight loss intervention, the

small sample size used in this study hinders making definitive

conclusions about the effect of diabetes status on treatment

outcome at this time. This remains an important finding to

replicate with a larger sample size, particularly because of

the implications for treatment planning.

The present study yielded results consistent with the

obesity treatment literature, but did not demonstrate

increased efficacy with the addition of mastery criteria to a

standard behavioral intervention. Although quantitative data

did not demonstrate advantages in using mastery criteria,

qualitative observations highlighted some benefits. According

to therapists who led the weight loss groups, advantages of

the mastery approach over the standard intervention were that

it provided helpful means for breaking down long-term goals

into smaller increments, delineated more specific goals, and

worked well for those who were competitive. In addition,

satisfaction ratings with treatment were slightly higher for

subjects in the mastery condition.

Results from the present study have implications for

future work in the area of obesity treatment, both clinically

and empirically. Data indicate that the addition of mastery

criteria alone is not sufficient to enhance treatment outcome.

Future studies may improve upon the use of mastery criteria by

modifying mastery levels to optimize opportunity for

advancement and by developing more frequent reinforcement

schedules. Additional methods for individualizing

interventions and facilitating skill acquisition within a

group format are also needed. Some recent research has

combined different approaches--such as VLCDs and behavioral

treatment--to try to improve initial response to intervention.

The key question continues to be, however, which approach

provides the best long-term results, since methods that

produce the best initial results are not always maintained at

follow-up (Wadden et al., 1989; Wadden & Stunkard, 1986).

Therefore, six-month follow-up data to the present study will

provide useful additional information for further evaluating

the utility of mastery criteria.

The findings in this study also indicated that

individuals with diabetes can accomplish clinically

significant weight loss equivalent to that of nondiabetic

individuals in a behavioral weight-loss program. Replication

of these findings with a larger sample that controls for

medication regimen will be an important expansion of the

current research, and necessary for elucidating the effect of

diabetes status on treatment response. Finally, although data

from the current study indicated that education is not

sufficient to produce significant weight loss or behavior

change, the search for better time and cost effective

approaches to obesity treatment should remain an essential

part of research objectives. The escalating cost of health

care and rapid changes in health care delivery highlight the

need to determine the minimal intervention necessary to

produce clinically significant weight loss. This will

continue to be a critical factor in the development and

evaluation of obesity treatment interventions in the future.


I _

Oj >
(.) 0)
mC o o o c o _
zu 0 c 0)
LLCO f C- > uJ
OW z 0

H. < 5 I C a 1
_z i I -0

C --

O m o --
S0) 0 0





24. Autobiographical Statement
25. Assessing Your Progress and Motivating Yourself for
Lasting Weight Loss
26. Final Assessment (No lesson)
27. Advancing into Phase II and Part 2 of Final Assessment


You and Your Eating
Controlling Your Eating
Exercise and Weight Loss
Making a Personal Decision to Exercise
Nutrition: Increasing Your Awareness
Balancing Your Diet
Using the Exchange Lists
Controlling the Factors that Influence Your Eating
Making Eating a Pure Experience and Rewarding Yourself
Self-reward and Preparing for the Unexpected
Chaining and Time-outs
Feedback on Computer Nutritional Analyses of Baseline
Food Intake
Preplanning and Controlling Holiday Eating LOTTERY
Asserting Yourself in Social Situation
Becoming an Informed Consumer
Eating Your Way to a Healthier You
Relaxation Training
Observing and Developing Positive Thoughts
Using Positive Thoughts to Counteract Underlying
Using Imagery to Help You Reach Your Goal
Impulse Control: Saying No to Yourself
Planning for the Restaurant Experience (No
written lesson)
Evaluating Your Restaurant Experience