Citation
An examination of the utility of the health belief model for predicting adult participation in aerobic exercise

Material Information

Title:
An examination of the utility of the health belief model for predicting adult participation in aerobic exercise
Creator:
Gage, Larry, 1953-
Publication Date:
Language:
English
Physical Description:
x, 182 leaves : ; 28 cm.

Subjects

Subjects / Keywords:
Behavior modeling ( jstor )
Diseases ( jstor )
Exercise ( jstor )
Mathematical variables ( jstor )
Modeling ( jstor )
Motivation ( jstor )
Observational research ( jstor )
Psychology ( jstor )
Questionnaires ( jstor )
Research design ( jstor )
Exercise -- Psychological aspects ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1990.
Bibliography:
Includes bibliographical references (leaves 171-180).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Larry Gage.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
001640232 ( ALEPH )
AHR5317 ( NOTIS )
24229042 ( OCLC )

Downloads

This item has the following downloads:


Full Text












AN EXAMINATION OF THE UTILITY OF
THE HEALTH BELIEF MODEL FOR PREDICTING
ADULT PARTICIPATION IN AEROBIC EXERCISE













By

LARRY GAGE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1990


UNI'JRSITY OF tLO"A.IA LI~l..n4










ACKNOWLEDGMENTS


I have been fortunate to have a committee of

scientist-practitioners who have had the time whenever I

have asked for it, and who each by their example have

inspired me. A piece of each now resides within me as I

continue to develop my own identity as a counseling

psychologist. Anchoring this diverse and talented group

has been Dr. Harry Grater, there to support and to

challenge. His guidance has made this project feasible,

with encouragement that has helped to motivate me. His

willingness to step outside his usual purview enabled me

to mount this project; I hope the experience has been

mutually edifying.

Especial thanks go as well to Dr. Jim Archer, whose

mentoring has been invaluable over many spheres, not just

this project. His role as cochair has included prodigious

energy invested in early drafts and discussions of this

work. It was his critical eye that provided for such

meaningful results as occur with this study.

Dr. Peggy Fong provided much of the early impetus and

knowledge that resulted in my choice of this topic, and

her input has been very helpful. In a similar way Dr. Shae

Kosch has provided much expert guidance in my development

as a psychologist in the health area; her matter-of-fact

imparting of substantive guidance and knowledge has been







much appreciated. A special debt of gratitude is owed Dr.

Barbara Probert, whose continuous unconditional positive

regard from my first association with her forward has

helped me to heal past hurts and to believe in myself. Her

support has been a spring whose waters I have gratefully

imbibed, even at a distance.

I must acknowledge the help of Dr. Henry Wang, who

helped advise about the best course to chart in wrestling

with abstract data and the arcane structure of SAS. Also

aiding in the process was Toby Gelman, whose support and

guidance helped me sidestep some deep electronic chasms.

Trenton State College, through the offices of the

Psychological Counseling Services and of Institutional

Computing, has provided critical resources that allowed

the execution of this project.

Finally, I owe deepest thanks to those whose support

has enabled me to approach this threshold. My parents, Joe

and Phyllis Gage, have provided support and encouragement

throughout my life. My wife, Karen Forbes, has supplied

substantial professional advice and consultation, over and

above the many kinds of support I have also gotten from

her as my life partner. She has been an in-house example

of how to set the standard as a new professional, and an

inspiration to me.


iii














TABLE OF CONTENTS


ACKNOWLEDGMENTS ..........................................ii

LIST OF TABLES..........................................vi

ABSTRACT .............................................. viii

CHAPTER

1 INTRODUCTION..... ...................................1

The Costs of Illness and Solutions Offered
by Counseling Psychology......................1
Attrition From Exercise Programs.................6
The Health Belief Model and Its Utility
In Exercise Adoption Research................11
The Role of an Attitude/Beliefs Study
in Exercise Adoption Research................15

2 REVIEW OF THE LITERATURE..........................25

The Benefits of Exercise........................25
Exercise Adoption.. ............................37
Factors in Adherence to Exercise................41
The Health Belief Model and Its Role
in Behavioral Research.......................51
Applying the Health Belief Model to
the Exercise Adoption Problem................67
Hypotheses..... .......................... .......86

3 METHOD.................................................. 90

Subjects...................................... .... 90
Instrumentation............................... 93
Procedure.... ............... ...................100

4 RESULTS............... ..................... ...... 104

Evaluation of the Personal Beliefs
Questionnaire.................................. 104
A Test of the Predictive Validity of the
Personal Beliefs Questionnaire.............. 111













5 DISCUSSION.......................................125

The Comparative Efficacy of the Personal
Beliefs Questionnaire........................125
Assessing the Criterion Validity of
The Personal Beliefs Questionnaire.......... 127
Limitations Affecting the Results of This
Study.................... ......... ...........134
Implications for Future Research on the
Precursors to Aerobic Exercise Behavior..... 138
Conclusion.....................................143

APPENDICES

A CONTACT LETTER FOR QUESTIONNAIRE..................146
B REPLY-CONSENT FORM................................147
C PERSONAL BELIEFS QUESTIONNAIRE....................148
D SECOND ADMINISTRATION OF QUESTIONNAIRE............156
E LIST OF ORIGINAL ITEMS AND ASSIGNED CODES.........165

REFERENCES ......... ....................................171

BIOGRAPHICAL SKETCH....................................181












LIST OF TABLES


TABLE PAGE

1 Summary of Demographic Descriptors For
All Respondents to Questionnaire One.........92

2 Coefficients Alpha for Initial Groupings
of Questionnaire Items......................105

3 Questionnaire Items Screened Due to Poor
Relative Association With Subscale..........106

4 Questionnaire Subscale Test-Retest
Correlation Coefficients....................107

5 Coefficients Alpha for Final Groupings
of Questionnaire Items......................108

6 Questionnaire Subscale Test-Retest
Correlation Coefficients for Final Item
Groupings ....................................109

7 Subscale Correlation Matrix for Final
Item Groupings..............................110

8 Summary of Factor Analysis Results...........111

9 Distribution of Subjects in Exercise
Adoption Categories.........................112

10 Individual Items Correlating Most Highly
With Adoption Stage Criterion............... 113

11 Sources of Variance in Exercise Adoption
Categories..................................114

12 Sources of Variance From Health Belief
Model Belief Subscales in Exercise
Adoption Categories.........................115

13 Sources of Variance in Exercise
Behavior Categories......................... 116









TABLE


14 Sources of Variance From Health Belief
Model Belief Subscales in Exercise
Behavior Categories......................... 117

15 Sources of Variance From Health Belief
Model Belief Subscales in Antipodal
Exercise Behavior Categories................118

16 Change in Total R-square For Criterion
Variables From First to Second
Administrations Using Independent Belief
Variables Obtained in First Administration..119

17 Number of Observations and Percents
Classified Into Exercise Adoption
Categories.................................121

18 Number of Observations and Percents
Classified Into Exercise Behavior
Categories................................... 122

19 Number of Observations and Percents
Classified Into Antipodal Exercise
Behavior Categories.........................123

20 Number of Observations and Percents
Classified Into Self-Reported Change in
Exercise Behavior Categories................124


vii


PAGE









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

AN EXAMINATION OF THE UTILITY OF
THE HEALTH BELIEF MODEL FOR PREDICTING
ADULT PARTICIPATION IN AEROBIC EXERCISE

By

Larry Gage

August 1990

Chairperson: Harry Grater
Major Department: Psychology

Prior research on successful exercise adoption has

focused on programmatic and behavioral aspects of exercise

programs, often omitting personal factors. This study was

designed to focus on individual beliefs contributing to

aerobic exercise adoption, using the Health Belief Model

(HBM) as a foundation for the instrument developed. The

beliefs selected for examination in this study were

perceived benefits of aerobic exercise, perceived barriers

to aerobic exercise, perceived susceptibility to disease,

perceived severity of possible disease conditions, and

general health motivation. The two purposes of the study

were to develop a valid and reliable measure of beliefs

related to aerobic exercise and to test the performance of

the beliefs in predicting aerobic exercise participation.


viii







A survey based on the HBM was developed to assess

respondents' beliefs about exercise and health. The

questionnaire was sent twice to a convenience sample of

334 adult faculty at a mid-sized college in central New

Jersey. The first administration of the questionnaire

yielded 179 usable responses, with 165 received in the

retest administration.

The questionnaire was evaluated by an item analysis

and measures of content validity, internal reliability

(alpha coefficients from 0.832 to 0.935 were found for the

final five scales), and test-retest reliability (r=0.813

for the total scale). Based on these tests it was

concluded that the measure was reliable with indications

of validity. Correlational and factor analysis tests of

the independence of the beliefs constructs as

operationalized in the questionnaire found some

association between the scales.

Multiple regressions and discriminant function

analyses were computed, providing predictive validity of

the beliefs measure for the criterion variables of

exercise adoption stage and level of exercise reported by

subjects. The independent belief variables accounted for

35% of overall variance (multiple r square) for categories

of exercise behavior, 40% for exercise adoption stages in

multiple regressions on these variables. The instrument

performed best in distinguishing between sedentary







subjects and aerobic exercisers, accounting for 51% of

overall variance and correctly classifying 82% of the

subjects. The main predictor for the criterion categories

was identified as perceived barriers to aerobic exercise,

corroborating similar research using HBM beliefs to

explain other preventive behavior.













CHAPTER 1
INTRODUCTION

The Costs of Illness and Solutions
Offered by Counseling Psychology

Americans spend eleven and one half percent of their

gross national product for health care in all its forms

("A Solution", 1988). This figure represents a two percent

increase since 1984, when concerted efforts began to limit

the growth of spending, and a five and a half percent

increase since 1965. In 1990, the projected share of the

GNP devoted to health care is expected to be twelve

percent, for a total expenditure of 750 billion dollars

(Kiesler & Morton, 1988).

It has been noted that "health" care actually refers

primarily to interventions with an illness, not to actual

promotion of health; only about two percent of U.S.

expenditures on health are for disease prevention and

control measures (Knowles, 1977). The rest of the budget

primarily services the Medicare and Medicaid programs,

programs oriented primarily toward remediation of existing

illness or injury.

This allocation of budget priorities misses the cause

of many of the illnesses treated, allowing the genesis of

most health problems to go unchanged and unaddressed.

Janis, in an article detailing counselor behavior







effective in supporting client behavior change, notes that

"the most serious medical problems that today plague the

majority of Americans and Europeans are not primarily

medical; they are behavioral problems requiring the

alteration of personal habits, preferences, and

decisions." (Janis, 1983, p. 146). Supporting this

observation, Powell (1988) cited an analysis of the causes

leading to lost years of life, noting that Lifestyle

factors were the largest proportion at 53 percent.

Following Lifestyle, in order, were Environment (22%),

Human Biology (16%), and the Health Care System itself

(10%; note: percentages rounded).

The lack of effort in the area of disease prevention

and control is not for lack of knowledge about what

personal health behavioral changes might be effective.

Belloc and Breslow (1972), in a very large longitudinal

study of over 7000 adults, found seven health practices

significantly related to health and life expectancy: 1)

three meals a day at regular times and no snacking; 2)

breakfast every day; 3) moderate exercise two or three

times a week; 4) adequate sleep; 5) no smoking; 6)

moderate weight; 7) no alcohol or only in moderation. A

45-year-old man who practices 6-7 of these habits has a

life expectancy ten and one-half years longer than one who

practices 0-3 of them (Belloc & Breslow, 1972). Despite








the availability of this study and others like it, few

programs to alter behavior exist.

Part of the problem in adopting a health promotion

focus is with the lack of immediacy of results from

efforts directed at changing health habits. Knowles

(1977), in comparing such an undertaking to traditional

public health practice, found "the knowledge required to

persuade the individual to change his habits is far more

complex, far less dramatic in its results, far more

difficult to organize and convey--in short, far less

appealing and compelling than the need for immunization"

(p. 61). For the person with questionable health habits,

the personal consequences of those habits are not visible

in their daily accretion or change, making programs for

behavior change difficult to target and execute.

In a description of the World Health Organization's

Health for All by the Year 2000 (HFA/2000) program,

Diekstra and Jansen (1988) note a large potential role for

the profession of psychology in accomplishing the

ambitious goals set forth there. They specify the fields

of psychological medicine, behavioral medicine, and health

psychology. As HFA/2000 concerns itself chiefly with

prevention and health-enhancing lifestyles, imagining its

realization without psychological support is indeed

difficult (WHO, 1981).







Many possible contributions of the field of

psychology with the goals of HFA/2000 can be identified.

Kiesler and Morton (1988) describe such opportunities for

"psychology in the public interest", citing facets of

psychology that well suit it to preventive efforts in the

area of health: 1) an empirical orientation; 2) expertise

not traditionally allied with medicine and health care; 3)

emphasis on both prevention and wellness; and 4) knowledge

about the effect of the environment on behavior and

well-being. Such expertise is pivotal to the ultimate aim

of changing behavior to reduce risk factors in the

population at large.

Indeed, the concept of "behavioral health" has been

discussed for most of this century, with some recent

attempts to specify and codify what is meant by the term.

Matarazzo (1980) defines behavioral health as

a new interdisciplinary field dedicated to promoting
a philosophy of health that stresses individual
responsibility in the application of behavioral and
biomedical science knowledge and techniques to the
maintenance of health and the prevention of illness
and dysfunction by a variety of self-initiated
individual or shared activities." (p. 813. Emphasis
in original).

He proposes this field as a way of progressing beyond the

limits of traditional approaches to health.

In a discussion of the potential role of counseling

psychology in public health, Thoresen and Eagleston (1985)

emphasize the traditional educational model that serves as

a foundation for the field. This approach lends itself to







health promotion in two related ways: "(1) the promotion

and maintenance of good health through the development of

healthy habits and lifestyle (2) the prevention and

treatment of disease through changing unhealthy behaviors,

especially those that may be associated with certain

chronic diseases and health problems." (Thoresen and

Eagleston, 1985, p. 70). They argue for consideration of

the cognitive, behavioral, and environmental factors that

have an impact on the issue of health along with

physiological ones.

Harris and Guten (1979) point out that an avenue

little explored in the area of health protective behavior

is how individuals attempt to attain health as producers

of this commodity rather than be consumers of others'

services. This suggestion is made in a discussion of their

findings in a study of how individual beliefs relate to

self-identified health protective behavior. Such a view

represents a challenge to the paradigm of conceptualizing

health; it also argues for investigation of the personal

characteristics that affect involvement in health

protective or enhancing activities.

A principal health protective and enhancing activity

that is receiving more research attention is aerobic

exercise. Regular aerobic exercise has been shown to be

beneficial in a number of different human functions,

including both physiological and psychological realms

(Folkins & Sime, 1981; Morgan, 1981; Paffenbarger and Hyde








(1988); Sinyor, Schwartz, Peronnet, Brisson, & Serganian,

1983; Stern & Cleary, 1981; Taylor, Sallis, & Needle,

1985). The need for exercise can be self-assessed or

prescribed by a health professional who has determined an

appropriate application. Exercise has been used

successfully in the treatment of depression (Greist,

Klein, Eischens, Faris, Gurman, & Morgan, 1979), coronary

heart disease risk modification (Stern & Cleary, 1981),

and obesity (Epstein & Wing, 1980). As has been observed,

the problem of adherence has compromised better assessment

of the efficacy of exercise in treatment regimens (Martin

& Dubbert, 1982).

Even moderate amounts of exercise have been shown to

be beneficial. In a very large study examining the

relationship varying levels of fitness to subsequent

mortality, a group of researchers found a "strong, graded,

and consistent inverse relationship between physical

fitness and mortality in men and women" (Blair, Kohl,

Paffenbarger, Clark, Cooper, & Gibbons, 1989). In

reporting their findings the authors observe that the

prevalent sedentary lifestyle presents an important public

health problem.

Attrition From Exercise Programs

The casual observer might well form the impression

that most people in the population at large exercise.

Exercisers are quite visible, as are advertisements for







health clubs and other "health-oriented" services. Such an

impression would be in error, however, when compared to

figures describing actual exercise behavior. A recent

overview of American participation in exercise found that

only 20% of the population exercise intensely enough and

often enough to meet current guidelines for fitness or

reduce the risk of chronic disease or premature death.

(Dishman, 1988). Another 40% are active to some level, but

not enough to derive a fitness benefit. This shows that

most Americans fail to perform a behavior with proven

efficacy in health promotion and remediation of health

problems.

Despite the growing recognition of the value of

exercise among professionals and laypersons alike, a

continuing pattern is the failure of most beginners in

exercise programs to persist in their efforts. This

pattern is observed even in the face of urgent medical

conditions for which exercise has proven an effective

treatment (Carmody, Senner, Malinow, & Matarazzo, 1980;

Dishman, Ickes, & Morgan, 1980; Martin & Dubbert, 1982).

Martin and Dubbert (1982) summarized findings on

clinical applications and promotion of exercise in

behavioral medicine, contending that measuring the

efficacy of exercise as a clinical treatment regimen was

hampered by poor adherence rates. They maintain that this

pattern introduces doubt even in areas where exercise is

generally agreed to be effective, as in obesity or








cardiovascular risk reduction. Commenting on the lack of

an effective means of keeping patients identified as

at-risk in a program of exercise, they state: "we believe

that far too little attention has been directed toward the

cognitive/psychological/social variables controlling

exercise adherence." (Martin & Dubbert, 1982, p. 1013).

They recommend the tailoring of individual packages to the

needs and profile of the individual undertaking an

exercise program. They document a need for more

information about the impact of different personal

variables on the success or failure of an exercise program

before being able to employ such packages successfully.

Attrition, or its parallel phenomenon relapse, is not

a problem unique to exercise (McAlister, Farquhar,

Thoresen, & Maccoby, 1976; Meichenbaum & Turk, 1987).

Representative of a number of writers is the Martin et al.

(1984) observation that the exercise relapse curve appears

similar to the negatively accelerated curve found for the

addictions; the majority of the relapses occur during the

first several months, followed by continued, more gradual

rates, finally leveling off at one year between 55% and

75% dropout. For exercise, relapse refers to the

phenomenon of discontinuing the behavior, as compared to

addiction where reference is to resuming an undesirable

behavior. Dishman, Ickes, & Morgan (1980), discussing a

similar observation, speculate that similar mechanisms may







operate across many behavior change problems. As put in a

related study:

the remarkable correspondence between dropout
patterns in exercise programs and programs of
psychotherapy, as well as drug, alcohol, and smoking
treatment, and hypertension control suggests
that similar influences may operate across many
health care settings. This similarity at least
implicates characteristics of the participant as
potential adherence predictors. (Dishman & Gettman,
1980, p. 296)

Activity in exercise research has begun to address

the problem of attrition from exercise programs, examining

the phenomenon in an attempt to identify factors affecting

the successful adoption of an exercise regimen (Martin &

Dubbert, 1982; Dishman, 1982; Gale, Eckhoff, Mogel, &

Rodnick, 1984). In a comprehensive review of these factors

Dishman (1982) organized them into areas that take into

account all the variables likely to have an impact on the

outcome of an exercise adoption attempt. The main areas

named include 1) psychological features of the exerciser;

2) biological features of the exerciser, including health

status; 3) demographic features of the exerciser;

4) situational characteristics of the exercise setting;

5) strategies/clinical interventions used to encourage

adoption of exercise habits; and 6) aspects of the

person-setting interface, including ways the interaction

between the person and the setting affects subsequent

decision-making and behavior. Such a summary is not unique

to the particular challenges posed in the area of exercise

behavior; similar findings are noted in an analysis of








adherence factors affecting the outcome of treatment

regimens prescribed by health care providers (Meichenbaum

& Turk, 1987).

Most research on exercise has addressed the social,

environmental, and behavioral aspects of the exercise

adoption process. Typical of research in the area of

exercise adherence is the comprehensive series of studies

performed by Martin et al. (1984). In these studies, a

decidedly behavioral focus yielded much valuable data

about the effects of social support, cognitive strategies

to accompany the exercise, relapse prevention training,

attendance lotteries, personalized feedback, and the

structure of the goal-setting for the individual's

exercise program. Using this example in light of the

Dishman categories cited above, one can see the emphasis

on 4) situational characteristics of the exercise setting;

5) strategies/clinical interventions used to encourage

adoption of exercise habits; and 6) aspects of the

person-setting interface.

In an already-quoted review that presents a summary

of much of the research on exercise adoption, Martin and

Dubbert (1982) make the puzzling observation that one's

attitude toward exercise does not appear to predict

participation or later adherence to a program of exercise.

Given the established links between attitudes and behavior

(Ajzen & Fishbein, 1977; Cooper & Croyle, 1984) and







specific studies establishing the relationship in the case

of other behaviors (Becker, Haefner, Kasl, Kirscht,

Maiman, & Rosenstock, 1977), such an observation cannot be

accepted as a given without supporting research. The

research necessary to determine the presence of such a

relationship has only recently commenced (Sonstroem,

1988).

The Health Belief Model and Its Utility
in Exercise Adoption Research

As a tool for comprehensive measurement of

psychological process, the Health Belief Model has shown

consistency and robustness, having value for health

practitioners and researchers addressing many varied kinds

of behavior (Janz & Becker, 1984). It enables a

comprehensive view of many beliefs that influence a

person's predisposition to act in a certain way, with the

exact mechanism of that influence still under study.

A belief is considered the cognitive part of an

attitude toward an object. A belief represents the

information an individual has about an object, as compared

to the attitude, which represents the individual's overall

evaluation of that object. The belief links the object to

a quality in the person's mind. Some have postulated a

belief to have an emotional component modifying the

overall response to an object or action (Sonstroem, 1982).

That is, the person's feeling about the outcome of the

action in question becomes part of their beliefs about







that action, affecting the valence of their specific and

general attitudes, beliefs, and behavioral intention.

The relationship between attitudes and behavior has

been a well-documented phenomenon, especially when many

elements of the attitude-behavior pair correspond (Ajzen &

Fishbein, 1977). Theorists in the area of

attitude-behavior relations note the importance of

comparable levels of specificity in the attitude and the

behavior (Cooper & Croyle, 1984). A general attitude will

predict a multiple-act criterion better than a single-act

criterion, while a specific attitude works conversely,

predicting the single-act criterion better than the

multiple-act criterion. The need, as detailed by Godin and

Shepard (1986), is for appropriate measures of attitude to

be employed in the particular area of exercise behavior.

There has been a long-standing interest in

documenting the relationship between attitudes or beliefs

and behavior. The Health Belief Model has been one of the

most widely utilized of theories that have been formulated

to determine the nature and impact of beliefs in the area

of health protective behavior. What started as an attempt

to identify the factors underlying health behaviors has

evolved into a well-documented model with utility in a

variety of settings.

The Health Belief Model was developed in an attempt

to understand why people did not practice health

protective behaviors, first in regard to asymptomatic








diseases, later in an expanding number of conditions and

venues. Use of the model was later expanded to investigate

factors contributing to compliance with medical regimens,

(Becker, Maiman, Kirscht, Haefner, & Drachman, 1977;

Becker, Drachman, & Kirscht, 1974) with recent

investigations broadening its application to preventive

health behaviors (Riddle, 1980; Slenker, Price, Roberts, &

Jurs, 1984; O'Connell, Price, Roberts, Jurs, & McKinley,

1985).

An early overview of research employing the model

included a statement detailing its utility for health care

providers: "by knowing which Health Belief Model

components are below a level presumed necessary for

behavior to occur, the health worker might be able to

tailor intervention to suit the particular needs of the

target group or population" (Becker, Haefner, et al.,

1977, p. 30). Other investigations into the efficacy of

interventions designed to affect peoples' health beliefs

and thus subsequent behavior have shown them to be

successful (Becker, Maiman, Kirscht, Haefner, & Drachman,

1977; Kirscht, 1974).

The components of the Health Belief Model were

posited to identify cognitions and perceptions making up

the beliefs that early developers of the model found to

have the most impact on health behaviors (Kirscht,

Haefner, Kegeles, & Rosenstock, 1966). They include the







individual's subjective estimate of their susceptibility

to disease states; the perceived severity of the

conditions the individual feels vulnerable to contracting

or suffering; perceived benefits of the health protective

behavior being researched; and the perceived barriers to

performing that behavior. Added later to the Health Belief

Model was a construct addressing health motivation, which

was seen as a necessary part of the model when examining

behaviors with a more preventive or health-enhancing

nature (Becker & Maiman, 1975). Finally, having an impact

within the framework of the model are cues to action which

trigger the individual to act.

The Health Belief Model was described by

theoreticians associated with early work in its

development as measuring specifically elements of two

variables essential to the prediction of behavior: 1) the

value placed by an individual on a particular goal (i.e.

health or the avoidance of a disease); and 2) the

individual's estimate of the likelihood that a given

action will result in that goal (Maiman & Becker, 1974). A

summation offered by the framers of the model who

participated in its refinement includes all the elements:

the Health Belief Model hypothesized that persons
will generally not seek preventive care or health
screening unless they possess minimal levels of
relevant health motivation and knowledge, view
themselves as potentially vulnerable and the
condition as threatening, are convinced of the
efficacy of intervention, and see few difficulties in
undertaking the recommended action. (Becker, Haefner,
et al., 1977, p. 29)








Investigations employing the Health Belief Model have

been successful in their aim of predicting or explaining

the level of non-exercise preventive health behaviors. Two

evaluations of Health Belief Model research found it to

have significance in the degree of association of its

principal components with almost every specific behavior

under study. The appraisals covered 46 different studies

that utilized the Health Belief Model. The pattern of

significance held when the type of behavior being

investigated was sick role behavior as well as with

preventive health behavior, and for earlier as well as

more recent studies (Becker, Haefner, et al., 1977; Janz &

Becker, 1984). The level of significance for the

components was also consistent in both retrospective

(n=28) and prospective (n=18) studies. The model appears

to have potential as a useful tool for the investigation

of factors shaping exercise behavior, one of the most

global forms of health-related behavior.

The Role of an Attitude/Beliefs Study
in Exercise Adoption Research

The relatively new area of research into exercise

adoption has been yielding specific data about the factors

that influence successful adoption and maintenance of an

exercise habit, particularly from the point of view of the

practitioner working with beginning exercisers. Behavior

modification approaches, as detailed by Knapp (1988),

Martin et al. (1984), and Franklin (1978), have shown








effectiveness at increasing adherence rates. Such

approaches now include relapse prevention strategies,

integrating newly gained knowledge about habit change. The

attractiveness of the exercise itself for the beginning

exerciser has been shown to be an important factor

(Franklin, 1978; Gale, Eckhoff, Mogel, & Rodnick, 1984),

along with the approach used by the instructor in an

exercise program.

The term exercise adoption was intentionally used for

the current study to help provide emphasis on the dynamics

whereby an individual actively seeks to adopt an exercise

program as a part of his/her daily behavior. The adoption

process is illustrated in a stage theory of change

detailed by Prochaska and DiClemente (1983). The process

begins with a precontemplation stage, with movement

proceeding to a contemplation stage. Actual behavior

change begins in the action stage, which becomes the

maintenance stage if the change is long-standing and

durable. Failure to maintain the change moves the person

into a relapse category. Such a schema helps to capture

more of the phenomenon of behavior change than just the

actual presence or absence of behavior.

The thoughts and feelings of the exerciser have been

relatively neglected in research on the exercise adoption

process, as compared to examination of the appropriate

techniques. An example of one such conceptualization of







personal cognition/self perception is self-efficacy theory

(Knapp, 1988; Bandura, 1982). While indications are that

exercise self-efficacy plays a role in the subsequent

outcome of an attempt to begin an exercise program,

results are not conclusive (Sonstroem, 1988). Discussions

of psychological characteristics having an impact on habit

change outcomes usually mention this construct, but it has

not been operationalized for the exercise adoption

problem.

The need for such socio-psychological correlates of

exercise behavior has been asserted by those conducting

cognitive-behavioral programs for change in exercise

adherence studies (Martin et al., 1984). The personal

characteristics and, occasionally, beliefs that influence

decision-making and subsequent behavior have been well

documented in the case of other behaviors, even other

health-related behaviors, but exercise behavior remains

largely unexplored in regard to beliefs.

The Health Belief Model method of assessment has

proven validity and reliability in determining individual

beliefs, and has become established as a tool for

predicting health behavior (Janz & Becker, 1984). The

first applications of the model in the area of exercise

behavior have proven encouraging, with limited

generalizability to all forms of exercise (Riddle, 1980;

Sonstroem, 1987; Slenker, Price, Roberts, & Jurs, 1984).









Dishman (1982) states the need to "determine the

effectiveness of various potential adherence strategies

. Consider the relative effectiveness of manipulating

abstract, conceptual beliefs" (p. 261). To date little

activity has focused on the development of interventions

to effect changed beliefs in those seeking to adopt

exercise behavior as part of their lifestyle. The primary

focus with the Health Belief Model up to now has been an

individual's participation in medical or

disease-prevention strategies. The current study was

proposed to provide a basis for a structured intervention

to increase the success rates for exercise adoption

efforts. With the capability to reliably assess beliefs

about health and exercise, the efficacy of such

interventions may be evaluated.

Aerobic Exercise and Health Beliefs

Extending the Health Belief Model mode of research to

the area of aerobic exercise behavior is a step indicated

by other prior studies that have explored the role of

health beliefs in preventive health behavior, including

specific forms of exercise. The present study was designed

to show that a reliable and valid mechanism of measurement

of beliefs regarding exercise can be developed, including

a demonstration of predictive validity of the instrument

in reference to self-reported aerobic exercise adoption

stages and levels of aerobic exercise.








Individual tailoring of programs to each person

seeking to begin an aerobic exercise program (whether

self-motivated or prescribed) has been recommended for

greater success achieving lasting change and implementing

the goal of an exercise regimen (Gale, Eckhoff, Mogel, &

Rodnick, 1984). Circumstances for a prescribed aerobic

exercise program may be weight reduction or cardiovascular

rehabilitation, while many reasons have been cited by

people undertaking exercise programs voluntarily, from

appearance to preparation for participation in a sport. A

specific, comprehensive beliefs measure would provide for

identification of personal beliefs relevant to entry into

an aerobic exercise program.

Martin and his fellow researchers have identified

many of the program strategies and interpersonal

approaches that enable the fitting of an exercise adoption

approach to the needs of the person starting a program of

exercise (Martin et al., 1984; Martin & Dubbert, 1982).

Missing from the daily practice of health care workers who

address the exercise adoption challenge is a consideration

of the individual's attitudes and beliefs about the

undertaking. Such an omission is, in part, due to lack of

a means with which to assess a person's beliefs about

health and exercise.

An historical drawback in Health Belief Model

research is that measures employed to assess belief







constellations for the behavior under study have been

idiosyncratic to each particular investigation.

Surprisingly, no one procedure of instrumentation has

emerged as a standard in any area covered by the dozens of

studies utilizing the Health Belief Model. In one attempt

to organize and verify the reliability and validity of

constructs within the Health Belief Model, the researchers

comment on the limitations imposed by the lack of such

work:

The absence of reliable and valid measures not only
limits the practical utility of the theoretical
formulation, but also reduces the potential for
developing a reliable body of knowledge on which to
design intervention strategies to change personal
health behavior. (Jette, Cummings, Brock, Phelps, &
Naessens, 1981, p. 83)

No attempt has been made to develop a standard instrument

to be employed as an adjunct for aerobic exercise adoption

situations, whether for research purposes or for use as a

treatment component.

Applying the Health Belief Model to
Aerobic Exercise Behavior

In studies where interventions utilizing Health

Belief Model components have been performed, successful

changes in targeted preventive health behaviors were

effected (Beck & Lund, 1981; Becker, Maiman, Kirscht,

Haefner, & Drachman, 1977). Behaviors included oral

hygiene care and mothers' successful follow-through on

treatment for their children' obesity. The technique of

beliefs modification has favorable potential as a method







for influencing the process of behavior change,

specifically the adoption of aerobic exercise behavior.

No definite relationship has been established between

the Health Belief Model components and aerobic exercise as

a dependent variable. The purpose of the present study was

to test for a link between the belief variables of the

Health Belief Model and aerobic exercise, using a

methodology paralleling prior, successful preventive

health behavior investigations. In these earlier studies,

the use of beliefs assessment instruments has provided for

successful prediction of a variety of targeted preventive

health behaviors (Tirrell & Hart, 1980; Chen & Tatsuoka,

1984; O'Connell, Price, Roberts, Jurs, & McKinley, 1985).

Previous studies have shown the Health Belief Model to

have similar potential in the successful measurement of

relevant beliefs and the prediction of aerobic exercise

behavior (Riddle, 1980; Slenker, Price, Roberts, & Jurs,

1984).

A valid instrument for assessing beliefs in reference

to aerobic exercise will provide a basis for successful

employment of belief component manipulations in public

health, recreation, and counseling settings, toward the

end of positive improvements in the adoption of aerobic

exercise behaviors. Belief manipulations can range from

personalized, one-on-one discussions to mass media

campaigns that involve an entire community. Such regimens

have been proven effective even on a very large scale. An








example of one such intervention often cited is the

Stanford Five-City Project, where significant reductions

in cardiovascular risk factors occurred over a thirty

month period following an ongoing public health

information campaign (Thoresen & Eagleston, 1985).

The method proposed parallels that used in successful

applications of the Health Belief Model to explain and

predict other preventive health behavior, as well as that

used in earlier applications of the model in the example

of illness avoidance/sickness reduction. As has been

noted, correlations of model components with target

behaviors that were obtained in past investigations have

been significant, with some variation in the degree of

significance. Initial exploration of the constellation of

health beliefs for exercisers has yielded encouraging

results. These findings have provided a foundation for an

effort to develop an instrument for determining Health

Belief Model beliefs in reference to aerobic exercise

(Sonstroem, 1987; Slenker, Price, Roberts, & Jurs, 1984;

Riddle, 1980). Such an instrument would provide the basis

for measurement that would enable assessment of any

subsequent manipulation of beliefs, in that change could

then be reliably measured.

Posing the Health Belief Model constructs in the

context of a standard definition of aerobic exercise

(American College of Sports Medicine, 1978) allows for








corroborative research and provides a basis for

comparisons to be made on the basis of a general

consideration of aerobic exercise, not just one particular

form. The College guidelines specify exercise quantity and

quality: 1) Frequency of training 3-5 days per week; 2)

Intensity of training 60%-90% of maximum heart rate

reserve; 3) Duration of training 15-60 minutes of

continuous aerobic activity, with duration dependent on

intensity; and 4) A mode of activity that involves large

muscle groups, that can be maintained continuously, and is

rhythmical and aerobic in nature (American College of

Sports Medicine, 1978, p. vii). The structure of the

current investigation was intended to provide a meaningful

foundation for future intervention studies to be

attempted, as well as other prospective studies that

monitor behavior change given an initial constellation of

beliefs.

In the following chapter, a more detailed examination

of the literature in the areas of exercise and exercise

adoption will be presented, including theory and practice

in exercise adoption, adherence, and compliance contexts.

A close inspection of factors affecting the exercise

adoption process will follow, including cognitive,

behavioral, and interpersonal elements. The Health Belief

Model and its application in other studies will then be

presented, along with investigations documenting the

suitability of the Model for aerobic exercise adoption








research and practice. Particular issues affecting the

development and use of an instrument to measure beliefs

about aerobic exercise will also be examined. Chapters 3

and 4 will describe the development, evaluation, and

testing of the beliefs/exercise assessment instrument. In

Chapter 5, specific findings and the implications of the

results for future investigations utilizing Health Belief

Model theory will be discussed, as well as indications for

the area of exercise adoption research in general.













CHAPTER 2
REVIEW OF THE LITERATURE

The following review will examine the literature on

exercise, exercise adoption, adherence to exercise

programs and the factors affecting adherence. Also to be

examined is the Health Belief Model and its use as an

explanatory tool in research on peoples' health-related

decision-making and behavior. Finally, research relevant

to the decision to apply the Health Belief Model in the

exercise adoption context will be reviewed.

The Benefits of Exercise

The benefits of regular exercise seem almost without

question; the findings, however, are not as unqualified as

the casual observer might expect. Many different kinds of

studies have been performed in an effort to ascertain

exercise effects along a continuum ranging from very

narrow, specific questions to very broad, inclusive ones.

An example of the former is a study by Sohn and Micheli

(1984) that investigates the effect of running on the

development of a specific form of arthritis, an example of

the latter is Folkins & Sime's (1981) analysis of exercise

effects on mental health. Their review was a critical

examination of the results of over 39 studies








investigating the impact of exercise on psychological

states.

Research studies on exercise also vary according to

type of exercise, length of exercise, the method of

monitoring adherence to an exercise program, and the

various groups of subjects studied. The following review

will focus on research that utilizes a more general,

inclusive definition of aerobic exercise, that is exercise

that produces an elevated heart rate for extended periods.

Another factor assumed and sought as an element of the

research cited is regular, versus episodic, participation

in programs of exercise. Factors utilized in the selection

follow from American College of Sports Medicine (1978)

recommendations about the quantity and quality of exercise

needed for "developing and maintaining cardiorespiratory

fitness and body composition in the healthy adult"

(American College of Sports Medicine, 1978, p. vii). The

effects "bias" in selecting research on benefits is toward

psychological results.

Stern and Cleary (1981) citing Naughton (1974) list a

summary of the clinically observable physiological

benefits of exercise, illustrating the pattern of changes

that accompany improved fitness. They include

1) significant reduction of heart rate and systolic blood

pressure, at rest and at work; 2) significant increases in

peak oxygen uptake; 3) significant decreases in myocardial







(heart muscle) work at rest and at work; 4) changes in

body composition, including reduced fat and increased

muscle mass; and 5) changes in circulation to patterns

observed in healthy, physically active subjects. This

pattern is noted in many studies as the training effect,

and various of the features are often used to test for

fitness levels, as they were in the Stern and Cleary

(1981) study on changes in a population of exercisers who

had suffered myocardial infarct.

In a chapter examining the effect of exercise on

coronary heart disease (CHD) rate and risk, Paffenbarger

and Hyde (1988) found significant differences in

longitudinal studies of two very different populations.

For a group of 3,686 San Francisco longshoremen, those in

the group who had the highest levels of physical activity

as part of their jobs were found to have significantly

less risk of CHD than their counterparts who did not work

as hard. This effect held for all age groups, and was

strongest for the youngest age group. The groups were

studied over a 21-year period, with the work level of

8,500 kilocalories per week used as the cutoff between

low- and high-energy-output groups. In an impressive

finding, the degree of increased CHD risk was greater for

those who were in a low energy output group (versus a high

output group) than the increased risk that accrued for

being a heavy cigarette smoker (greater than one pack a








day) or for having high blood pressure (greater than or

equal to the mean level for the age cohort)!

A parallel study, involving 16,936 Harvard alumni,

found similar results. Alumni engaging in strenuous

activity plus a minimum of 1000 kilocalories a week of

other light activities had less than half the CHD

incidence of their nonathletic, sedentary classmates

(Paffenbarger & Hyde, 1988). Exercise was found to be the

second most influential factor affecting mortality from

CHD causes, preceded by smoking and followed by

hypertension. Exercise was second only to parental history

of hypertension in its impact on whether the alumni

developed hypertension themselves.

Recent results demonstrate some of the likely

mechanisms by which regular exercise may contribute to

lower CHD mortality. A study investigating the impact of

exercise on blood cholesterol levels found that "good"

cholesterol (HDL or high density lipoprotein) began

increasing while "bad" cholesterol (LDL/VLDL or low/very

low density lipoprotein) began decreasing after a seven

week exercise program, a fairly rapid effect of exercise.

The program consisted of four sessions per week of

treadmill running for twenty minutes per session to 80%

predicted maximum heart rate. Subjects ate a controlled

diet and underwent frequent blood testing to assess

lipoprotein and other blood element levels. A parallel







finding was a significant decrease in postprandial levels

of lipoproteins, showing a faster uptake of cholesterol

(Weintraub, Rosen, Otto, Eisenberg, & Breslow, 1989).

Exercise was demonstrated to have a positive effect

on the coping responses to a stressor when trained and

untrained subjects were compared. Heavy aerobic exercisers

were assigned to the trained group, while non-exercisers

comprised the untrained group. In addition to a standard

"training effect", (Higher maximal oxygen uptake, lower

blood pressure) subjects showed a faster adaptive

biochemical response when subjected to a psychosocial

stressor (Sinyor, Schwartz, Peronnet, Brisson, and

Serganian, 1983). Catecholamine levels peaked earlier, and

at higher levels, for the trained subjects. This effect

was described by the authors as a more efficient coping

arousal. Additionally, heart rate returned to normal more

quickly in the trained group of subjects, reflecting a

recovery from the experience of the stressor that was

faster than in the untrained group.

Stern and Cleary (1981), in the above-noted study on

a male post-infarct population of exercisers, noted an

increased work capacity for the exercisers as measured in

testing on a motorized treadmill. Their program of

low-level exercise consisted of two to three sessions per

week for six weeks for an average of 15 minutes of

exercise per session. This program produced an average ten









percent increase in the work capacity of subjects over the

course of the study.

Wilfey and Kunce (1986) found the magnitude of

training benefit to be greater for subjects with lower

initial fitness levels, in both physiological and

psychological measures. For low fitness/high stress

subjects the overall changes were the greatest, as might

be projected for people of the group with the most to gain

from completing a two month long comprehensive exercise

training program. Heart rate decreases were greatest for

the low fitness/high stress group, followed in order by

the low fitness/low stress group, the high fitness/high

stress group, and the high fitness/low stress group. A

similar pattern obtained in measuring gains on a fitness

test, with the high fitness/low stress and high

fitness/high stress groups exchanging ranks. The finding

that persons of lower fitness gain more from training than

those at higher fitness levels may seem obvious, but it

also illustrates the linear nature of the physiological

gains obtainable through an exercise training program.

Many psychological benefits have been documented as a

product of regular exercise, with findings consistent

through many studies. The positive effects can be

identified with a high degree of certainty as associated

with the exercise itself and not with other variables.

Well-controlled studies do vary in the specific effects








noted, often as a result of their initial focus or

particular changes identified as the target of the

investigation.

In a review of his own and others' work on the short

and long term effects of exercise, Morgan (1981) found

some consistent indications that regular exercise,

performed for a minimum of twelve weeks, results in lower

scores on anxiety and depression scales, while measures of

well-being increased. This pattern held across different

measures and different periods of exercise beyond the

twelve week minimum. Morgan concludes that exercise is

clearly associated with changes in the three areas, with

an additional observation that causality is harder to

prove.

A related study found reduced levels of anxiety in

physically trained subjects following exposure to a

psychosocial stressor when compared to untrained subjects

(Sinyor, Schwartz, Peronnet, Brisson, & Serganian, 1983).

Aerobic points earned and physiological response measures

were used to differentiate trained from untrained

subjects. The anxiety measure was the Spielberger State

Anxiety Inventory, administered prior to and following

different stressor tasks. Following the tasks, trained

subjects had significantly lower SSAI scores, showing

faster recovery from induced anxiety.







One of the largest, most complete studies was

conducted on a population of 651 subjects that had

experienced a myocardial infarct less than three years

prior to their involvement in the research. Stern and

Cleary (1981) investigated the psychosocial effects of a

low level exercise program on participants, finding a

decrease in the percentage of depressed subjects and in

depression scores as measured by the MMPI depression

subscale. On a group of twelve clinical symptoms,

persisters in the exercise program improved on ten of the

twelve as rated by spouses. Differences were found to be

significant between the initial and followup evaluations,

which were administered six weeks apart. Although the

authors see the exercise group as benefiting more than a

non-exercise group, they qualify the finding as not being

wholly attributable to the exercise component.

One of the most thorough evaluations of the impact of

exercise on mental health was a review of research

conducted by Taylor, Sallis, and Needle (1985). They found

positive effects in many of the studies they examined

where exercise was used as a treatment modality for

clinical conditions, many with a significant degree of

change. Mild to moderate depression was cited as showing

the most promise for change, with many studies documenting

positive effects. The hypothesized mechanisms of change in

exercise programs used to treat depression, while not








certain, were increased neurotransmission of

catecholamines, endogenous opiates, or both; diversion,

social reinforcement, and improved self-efficacy.

Taylor et al. (1985) found both state and trait

anxiety to be effectively reduced by varying levels of

exercise, with a greater effect noted in subjects whose

initial anxiety levels were higher. Effects were more

conclusive for state anxiety than for trait anxiety, with

both types showing a positive benefit in most studies

examined. The primary effective mechanisms here were

described as diversion, social reinforcement, experience

of mastery and an improved response to stress. Taylor et

al. (1985) also noted that improved self-confidence,

cognitive function, and sense of well-being were

associated with exercise and physical activity.

Perhaps the most complete of the major studies

reviewing the impact of exercise on mental health is the

massive compilation by Folkins and Sime (1981). Their

critical review is often cited in other studies as a

significant overall finding of the beneficial effects of

exercise on human psychological function. Often omitted in

a citation of their work is the observation that most of

the research on exercise effects that they reviewed were

not true experimental designs. They note a preponderance

of pre-experimental or quasi-experimental designs, a

drawback that is particularly limiting in longitudinal










studies. Even within the true experimental research,

however, a pattern of positive effects is cited.

Folkins and Sime (1981) found fitness to be related

to the following aspects of mental health: 1) improved

work patterns, including reduced absenteeism, fewer

errors, and improved output; 2) improvement in mood state,

particularly for distressed or initially unfit

individuals; 3) improved self-concept and body image; 4)

reduced levels of depression where exercise is used for

treatment of depression; and 5) improved cognitive

function. The authors caution that in some instances the

results obtained in a study were not significant; overall

they were convinced of the utility of exercise in

producing improved psychological states. They single out

self-concept and depression as areas where the results are

less equivocal.

In a carefully controlled study, Bahrke and Morgan

(1978) compared exercise, meditation, and quiet rest to

determine their relative efficacy at inducing relaxation

for experimental subjects. Each "treatment" was employed

for twenty minute periods, with the exercise condition

being treadmill walking to seventy percent of age-specific

maximal heart rate. Relaxation was measured using a state

anxiety scale that was employed preceding, immediately

following and ten minutes following the experimental







conditions. All three treatments were effective in

reducing anxiety, causing decrements that differed

significantly from the initial measurement. Interestingly,

the three treatments did not differ significantly from

each other in their effects. The authors found the

findings striking in that the mechanism of exercise

(physiological arousal) is so different from that of

either quiet rest or meditation in the quality of its

effects. They also found the exercise effects corroborated

earlier investigations into the impact of exercise on

anxiety levels. It must be noted that "lack of anxiety"

may not be a universal criterion for relaxation; given the

arousal aftereffects on physiological measures it was the

choice for this study.

Morgan and O'Conner (1988) have summarized some

limitations in the studies which document the positive

effects of various forms of exercise on mental health.

They note a unanimity among researchers that the higher

the level of fitness, the more desirable the level of

mental health. Next is noted the elusive nature of

evidence allowing clear statements of causality. Finally,

the lack of controlled studies with true experimental

designs, as noted above in the review by Folkins and Sime

(1981), precludes clear statements about cause and effect.

This point holds even in longitudinal studies that would









support the hypothesis that vigorous exercise leads to

beneficial mental health effects.

Citing a NIMH workshop that used a strictly limited

criterion for selection of statements about exercise

effects, Morgan and O'Conner (1988), list findings about

the influence of exercise on mental health. These findings

use carefully stated points to summarize what can be

stated about the positive effects to be gained from

exercise/fitness training:

1. Physical fitness is positively associated with
mental health and with well-being.
2. Exercise is associated with the reduction of
stress emotions such as state anxiety.
3. Anxiety and depression are common symptoms of
failure to cope with mental stress, and exercise has
been associated with a decreased level of mild to
moderate depression and anxiety.
4. Long-term exercise is usually associated with
reductions in traits such as neuroticism and anxiety.
5. Severe depression usually requires professional
treatment, which may include medication,
electroconvulsive therapy, and/or psychotherapy, with
exercise as an adjunct.
6. Appropriate exercise results in reductions in
various stress indices such as neuromuscular tension,
resting heart rate, and some stress hormones.
7. Current clinical opinion holds that exercise has
beneficial emotional effects across all ages in both
sexes.
8. Physically healthy people who require psychotropic
medication may safely exercise when exercise and
medications are titrated under close medical
supervision. (p. 95)

The exercise effect is thus shown to be dependable,

applicable to a wide variety of persons, and amenable to

use at the same time as other forms of treatment.

Peoples' perceptions about the benefits of a regular

program of exercise are equally as important as the








documented benefits. As will be seen in the discussion of

the components of the Health Belief Model, perceptions are

an important aspect of beliefs. In a study in which people

were polled about various forms of behavior they performed

to "protect, promote, or maintain health", Harris

and Guten, (1979) found that exercise behavior was cited

the third most often--behind nutrition and sleep--as one

of the "three most important things that you do to protect

your health." (p. 21). The important distinction here was

that the researchers recorded what behaviors the

respondents considered important for health protection,

versus a list of behaviors that clinicians perform or

would recommend. The level of utilization of health

protective behaviors held even when health condition of

the respondent varied from good to poor, with the emphasis

shifting from self-initiated lifestyle choices like diet

and exercise to utilization of the health care system.

Exercise Adoption

The phenomenon of exercise adoption may be patterned

after an area of research that provides a basis for

measurement and interpreting observations. Prochaska and

DiClemente (1983) have developed a theory of behavior

change which describes a person's movement through a

series of psychological/behavioral stages, resulting in

several different outcomes. In their work developing this

stage theory, Prochaska and DiClemente (1983) have focused








primarily on the cessation of undesirable behaviors, such

as smoking cigarettes.

As specified in their schema, a person's initial step

in the change process is from an immotive or

precontemplation stage to a contemplation stage. Using

cognitive/affective reevaluation of their situation and

behavior, the person is compelled to move to an action

stage, changing actual behavior. Should the target

behavior change be long term and durable, with

accompanying cognitive and behavioral changes to support

the target behavior, the stage entered into is that of

maintenance, where the behavior is ongoing and maintained

at a stable level. If behavior drops below the minimum

specified level of change, the person drops into the

relapse category, wherein the attempted behavior change is

abandoned for a period of time. The period may be brief or

extended, depending on the determination and resources of

the changer toward renewing the effort at change. As will

be detailed below, the relapse outcome is fairly common in

behavior change projects.

Employing the Prochaska and DiClemente (1983) schema

in the context of exercise underscores the need for

differentiation of adoption from adherence, which, as

generally employed by exercise researchers, includes only

the maintenance stage of the process. This stance excludes

other important and universal features of human attempts

to change behavior that are noted in a broader view of









change. The stage process as used in early studies applies

whether the change is self-initiated or prescribed by

other parties. At present the terms adoption and adherence

are often used interchangeably to refer to success or

failure to adhere to a program of exercise begun for any

reason, with many of the cognitive factors and much

personal history not taken into account.

Dishman (1988) notes the confusions inherent in the

current terms being used in behavioral science to describe

the exercise adoption process. He observes that compliance

and adherence are often used to describe the same

phenomena, yet are not generally recognized in catalogues

of psychological literature by definitions that conform to

their current utilization. He proposes employing the term

adherence as a general descriptor in the area of exercise,

to include the area of compliance. This comes closer to

meeting current needs and recognizing the entirety of the

exercise adoption process, yet the early stages of a

self-initiated change can be omitted by this term as well.

Too, other researchers explicitly link the terms

compliance and adherence (e.g. Oldridge, 1981), serving

their own needs in investigating exercise treatment

programs, but excluding important population groups of

exercisers and parts of their experience in entering into

an exercise program.







The research on compliance to exercise programs

addresses situations that are primarily prescribed

behaviors in response to a stimulus situation like obesity

or high blood pressure. The focus of compliance research

has been on practitioner behavior and characteristics of

the setting, with the interaction between client and

health care provider the main interest (Becker & Maiman,

1980; Janz & Becker, 1984). Research in compliance

assuredly does include individual characteristics,

providing the overlap and confusion of terms alluded to

above.

Examination of the research findings shows by

illustration some of the phenomena to be anticipated in

research on exercise adoption. A finding that has results

useful for the issue of self-change of habits is the

comprehensive study carried out by Perri and Richards

(1980). The study compared people who had carried out

successful self-change programs with those who had

attempted unsuccessful ones, in an attempt to determine

what personal characteristics or strategies differentiated

the two outcomes. Important indicators emerged for those

undertaking self-change programs or working with those who

do. Successful self-changers appear to be intuitively

skilled at behavioral technique, employing

self-reinforcement and persisting far longer in their

attempts than did unsuccessful changers. Of note for the

present study was the recommendation that persons









undertaking self-change programs insure that their program

is comprehensive in nature and should involve a number of

individually tailored procedures.

Some research (Sonstroem, 1987; Ward & Morgan, 1984)

has been performed using the stage schema applied to

exercise. Ward and Morgan appear to be the first to apply

the stage theory of change to exercise, finding that

adherence at 10, 20, and 32 weeks in an exercise program

was predicted by different sets of variables. Using the

Theory of Reasoned Action (Ajzen & Fishbein, 1977) as a

basis for a beliefs investigation, Sonstroem related the

belief constellation to a reduced number of the Prochaska

and DiClemente stages. He termed the process exercise

adoption, although without providing a rationale for the

use of that term versus the others available. The main

body of work cited in his review of the theoretical bases

for his research, he notes, does include subjects who were

seeking to change on their own. As the term "exercise

adoption" is rarely used, the more widely held "adherence"

term will apply through most of this review.

Factors in Adherence to Exercise

Exercise adherence is examined from many different

perspectives, with each providing information to compose a

picture of someone in the exercise adoption process. Some

studies examine the factors associated with dropout, where

others test for personal characteristics associated with a







successful, continued exercise program. Perhaps one of the

most complete recent studies on exercise adherence was the

collection of researches reported by Martin et al. (1984).

The series was designed in an attempt to isolate factors

that affect the outcome of exercise adoption attempts by

sedentary adults. The investigators' main interest was in

determining the efficacy of various cognitive-behavioral

procedures in support of exercise programs begun by people

attending a community exercise program. Dividing 143

subjects into six treatment conditions allowed the

researchers to test for the effectiveness of different

elements of behavioral techniques.

Each treatment group that began a running program

received differing combinations of cognitive-behavioral

strategies for the three months of the program, apart from

a common structure utilized by all groups. For each group

an "ideal" strategy was included, with the ideal based on

the past observed efficacy of each element. In Group 1,

the ideal was personalized feedback, distance goals (for

running), and fixed goals. Each group provided varying

combinations, allowing the researchers to compare group

and personalized feedback; time and distance goals;

flexible and fixed goals; distal (long range) and proximal

(short range) goals; and associative versus dissociative

cognitive strategies.

Analyses of variance and t-tests were used to test

for significant differences between conditions within








groups. Many of the variables investigated proved a

positive influence on the ultimate adherence of the

subjects beginning exercise programs. In particular,

social support, personal feedback/praise, flexibility in

goal setting, and distraction-based cognitive strategies

proved most effective in increasing rates of adherence. No

advantage was seen for time versus distance goals.

Impressively, overall adherence at the end of the three

months, for all subjects in the study, varied from 80-85

percent. This was considerably better than the 55 percent

adherence usually observed for this time period in an

exercise adoption group program (Martin et al., 1984). Of

interest here is the fact that 49% of the subjects in the

study had previously attempted an exercise program. No

differentiation of these "relapsers" from the subjects

beginning exercise for the first time was attempted.

Martin et al. (1984) also found class attendance and

self-report of exercise participation to correlate

significantly with "post" measures of fitness.

Another study illustrating recent methods used to

scrutinize exercise adherence is one by Dishman and

Gettman (1980) that examines many different psychobiologic

influences on ultimate adherence. The study used

physiological measures such as weight and percent body fat

along with psychological measures designed to measure

attitudes toward and valuing of exercise. A prospective







study with 66 subjects, it used subjects that were

beginning exercise programs for the first time, combining

healthy exercisers with coronary heart disease

rehabilitation subjects. The outcome of the entry into an

exercise program was labeled as either adherence or

dropout, with anyone not participating at the end of the

20 week period of the study labeled a dropout.

Using stepwise multiple regression, three variables

of the ones measured were found to significantly enhance

the prediction equation for adherence: 1) body fat

percentage; 2) a measure of self-motivation; and 3) body

weight. Taken together, these three variables accounted

for 45 percent of the overall variance and provided for

predicting adherence/dropout category in 79 percent of all

cases examined. The authors chose only one other

independent measure in addition to their own on

self-motivation, that of the health locus of control

(Wallston, Wallston, Kaplan, & Maides, 1976). The other

scales were subscales of the Attitude Toward Physical

Activity scale or the Physical Estimation and Attraction

scale. The ability of these subscales to contribute to a

prediction of subsequent exercise status may have been

compromised by their application within this study as

stand-alone variables.

In a study examining the long-term rates of adherence

for a population of male and female victims of myocardial

infarction, the investigators found only 18.7% remained at








the end of a forty month period (Carmody, Senner, Malinow,

& Matarazzo, 1980). Two hundred and six subjects were

followed up at four month intervals to see which remained

active in a prescribed exercise program. In the initial

four month period dropout was 30%, with the subsequent

rate of dropout reduced in each followup after the initial

period. This pattern duplicates the typical finding for

relapse/dropout in habit-change studies (e.g. Dishman,

1988), with the dropout rate somewhat greater than that of

a general population. Studies of this duration are rare,

so comparison is difficult. This is nonetheless a

surprising finding, given the explicit urgency attached to

the successful adoption of the exercise prescription for

this group.

Summarizing findings in a chapter reviewing studies

on adherence to post-myocardial infarction exercise

regimens, Oldridge (1981) identified some common

characteristics that tended to be associated with dropout.

The research reviewed focused primarily on the

characteristics of the enrollees, omitting formal measures

of their psychological or physical states. A profile

constructed using characteristics that recurred across

studies shows that a dropout is likely to be a blue collar

worker, to be inactive physically in his leisure time, to

be a smoker, and to have had more than one previous

myocardial infarction. The mechanisms whereby these







characteristics contributed to the outcome of dropping out

were described by Oldridge as being unclear. Some of the

subjects' negative evaluations of the programs in which

they were participating included statements like

"difficult to perform," "dislike physical training," or

"aversion to the hospital".

Franklin (1978) investigated the factors affecting

exercise adherence in fitness programs for both healthy

adults and patients in rehabilitation programs. Although

the analysis focused primarily on individual physical

characteristics, fitness level, and behavioral regimens,

motivation was mentioned as a main personal factor to be

addressed. Franklin's suggested solutions to motivational

deficits focused on the structure and form of the exercise

itself rather than directly on psychological factors.

Program features associated with dropout were lack of

encouragement for clients, client injuries incurred while

participating, individual participation versus group

programs, inadequate instruction, and absence of regular

updating of clients on progress made. Other client

features cited as factors were presence of support from

spouse or peers and a convenient time schedule for

involvement in exercise.

A prospective study that followed 106 subjects for a

total of six months was conducted by Gale, Eckhoff, Mogel,

and Rodnick (1984) to determine factors that could be

associated with exercise adherence. Many measures were








taken of physiological characteristics, self-motivation,

demographic descriptors, health related behaviors, and

previous involvement in exercise. Subjects were healthy

adult men and women; 18% dropped out before attending 10%

of the classes, 40% attended between 10 and 40% of the

classes, and 42% attended more than half the classes.

Factors were examined for correlation to adherence.

Findings were disappointing given the array of

related factors that were examined. Only one of the

factors (marital status) was significantly correlated with

adherence to the program of exercise, and the two factors

with the highest predictive value were the number of years

in present occupation and number of years at present

address. Indicators of less stability of overall life were

associated with early dropout; features such as being

single or being only a couple of years at present address

or job. The other variables that distinguished between

categories were self-motivation and number of children.

Discussing the puzzling results, the authors speculate

that the dropouts in their study may have continued to

exercise away from the program, or that they may have not

found the program to be engaging or challenging enough.

Investigations attempting to identify specific

personal factors affecting adherence to exercise sometimes

yield results in unexpected directions. An example is

research cited above by Dishman, Ickes, and Morgan, (1980)







to verify the concept of self-motivation as an explanatory

mechanism for differential outcomes for people beginning

exercise regimens. Their findings found their construct of

self-motivation to be significantly associated with

adherence to an exercise program; it also found the most

important factor affecting adherence to be percentage of

body fat. The higher the percentage of fat (and the

greater the weight) the more likely the person was to drop

out of the exercise program. Although self-motivation was

one of the three factors that reached significance in

distinguishing between adherers and dropouts, this finding

illustrates the way in which many different--and sometimes

unexpectedly prominent--biopsychosocial variables affect

behavior.

As an aside in surveying factors affecting

adherence/compliance, Dishman (1982) notes the importance

of taking into account those who may drop out of the

formal exercise program under study, but continue to

exercise on their own or in another program. Dishman also

cautions against losing research information by

dichotomizing the population into adherers and dropouts,

to the exclusion of the large class that does succeed in

participating in an alternative pattern of exercise.

In the above pair of studies Dishman, Ickes, and

Morgan, (1980) and Dishman and Gettman (1980) advanced the

concept of self-motivation as an explanation for

differential outcomes in people beginning exercise








regimens. They dismiss the role of attitudes or beliefs,

positing a weak or absent relationship. Godin and Shepard

(1986) have responded with the observation that the

problem may lie with the method. They reply to Dishman and

Gettman (1980) with the assertion that the need is for

theory-driven research to test models, versus the

"shotgun" approach of associating variables with outcomes

and claiming some degree of relationship. For them the

questions to ask are "when" and "how": when is there a

relationship between attitude and behavior, and how is the

effect of an attitude on behavior mediated? Sonstroem

(1982), acknowledges the utility of self motivation, but

notes that its classification as a trait places it outside

the area targeted by research into modifiable

psychological factors. As will be seen below, many

theorists working with the Health Belief Model believe

motivation is determined by the constellation of beliefs

held by persons in relation to certain pivotal areas of

their lives.

Powell (1988) summarizes factors affecting exercise

adoption in table form, collecting together potential

"determinants of physical activity" as an illustration of

the scope of the problem facing those seeking to identify

the specific factors having an impact on adherence to

exercise. The determinants come from a compilation of past

research done to identify the most important and most







salient influences on an exercise adoption effort.

Although research results in the exercise adoption area

are far from conclusive, the list of factors is

instructive:

Personal characteristics--Past program participation;
past extraprogram activity; school athletics, 1
sport; school athletics, >1 sport; blue collar
occupation; smoking; overweight; high risk of
coronary heart disease; type A behavior; health,
exercise knowledge; attitudes; enjoyment of activity;
perceived health; mood disturbance; education; age;
expect personal health benefit; self-efficacy for
exercise; intention to adhere; perceived physical
competence; self-motivation; evaluating costs and
benefit; behavioral skills.
Environmental characteristics--Spouse support;
perceived available time; access to facilities;
disruptions in routine; social reinforcement (staff,
exercise partner); family influence; peer influence;
physical influences; cost; medical screening;
climate; incentives.
Activity characteristics--Activity intensity;
perceived discomfort. (Powell, 1988, p. 27)

Powell's list bears some striking similarities to a

similar group of factors identified as having an impact on

treatment adherence behavior (Meichenbaum & Turk, 1987).

Many of the above findings on adherence to exercise may

also be found in their assemblage of factors affecting

compliance with treatment regimens. Powell's list is a

good tool for guiding research on exercise adoption, in

that it reminds the researcher of potential limitations to

the conclusions drawn from research on any one variable.

Given this caveat, another finding that may be noted is

the number of items on the list that will have








representation in some form in a Health Belief Model

investigation targeted on exercise behavior.

The Health Belief Model and Its Role
in Behavioral Research

The Health Belief Model was developed in the early

1950s in an attempt to understand individual differences

in the practice of preventive health behaviors, first in

regard to asymptomatic diseases, later in an expanding

number of disease states, venues, and situations. The

roots of the Health Belief Model lie in social psychology,

specifically the work of Lewin (1935). Rosenstock (1974a)

identifies the main orientation of the early researchers

as phenomenological, centered on the world of the

perceiver as opposed to the actual physical world. It was

this foundation that led to the formulation of a model

detailing different personal "fields" that were believed

to affect subsequent motivation and thus behavior.

In an explication of the function of personal fields,

Rosenstock (1974a) puts the origin of motivation as toward

positive fields and away from negative ones. Persons

operating within these fields would thus move away from

those areas of their lives that were perceived as negative

and toward those parts that were valued as positive. The

components of the model represent a conceptualization of

the fields likely to be identified as affecting decisions

about behavior having an impact on health.








In one of the earlier theoretical writings detailing

the roots of the Health Belief Model, Suchman (1970)

elaborated on the role of perception and interpretation in

the etiology of illness. He notes, "People do not perceive

the world as it actually is but as they have become

accustomed to perceive it" (p. 107). In his

conceptualization of the motivation to change behavior,

the facts matter less than one's perceptions in making

choices.

Three levels of influence guide the perceptual

process in Suchman's view of individual and social

behavior: the anthropological/cultural, the

sociological/group, and the psychological/individual. Each

level shapes the resultant attitudes and beliefs, thus the

behavior. "Motivation to change one's health practices

depends, to a large extent, upon the individual's feelings

of personal vulnerability and the seriousness with which

he views the health hazard." (Suchman, 1970, p. 109-110).

It is thus possible to see the scope of the challenge if

one thinks in terms of behavior change interventions, as

the focus is not just factual information but the

particular beliefs the individual has about the world.

Becker, Maiman, Kirscht, Haefner, & Drachman (1977)

argue that the Health Belief Model is based on a

"value-expectancy" approach, wherein behavior is

predictable if one knows an individual's valuation of an







outcome and their expectation that a specific action will

result in that outcome. Characteristics the authors deem

relevant to the formulation of a model utilizing health

beliefs include 1) the motivation to avoid illness or to

get well; 2) the amount of desire for a specific level of

health; 3) the belief that specific actions will prevent

or moderate illness or disease conditions. The alignment

of components detailed on pages 13 and 14 of Chapter 1 is

the grouping commonly studied in Health Belief Model

research, and were devised to assess the presence and

strength of the above characteristics.

In Ajzen and Fishbein's theory of reasoned action,

(Ajzen & Fishbein, 1977) beliefs are considered the

cognitive component of the attitude. Kirscht (1974)

defines a belief as "any proposition or hypothesis held by

a person, relating any two or more psychological elements

or objects" (p. 456), while attitudes are posited as

collections of beliefs in which there is an evaluative

component. Highlighted more within Health Belief Model

theory is the evaluative, positive-negative association

that the individual has at the belief level. A parallel

can be discerned between the process described in the

Health Belief Model and attitude-behavior relations, with

the outcome, behavior, dependent on collections of

attitudes, in turn made up of beliefs.

The attitude-behavior link has received much

discussion in social psychology, with lack of consensus








about the role of attitudes in determining behavior

(Cooper & Croyle, 1984). Generally accepted are the

assertions by Ajzen and Fishbein, (1977) that the

specificity of correspondence between attitudes and

behavior has a lot to do with the degree of the

relationship between the two. That is, a general attitude

will predict a multiple-act criterion better than a

single-act one, while a specific attitude will predict a

single-act criterion better than a multiple-act one. An

illustration is the comparison between a general attitude

toward health with its resultant health-protective

behaviors and a specific attitude about tooth-brushing and

subsequent brushing behavior. Recent findings bear out

these assertions, with a study using a measure of general

attitudes and multiple acts showing a high degree of

correspondence between the two (Turk, Rudy, & Salovey,

1984).

Rosenstock (1974b) documents the first application of

the Health Belief Model as a study by Hochbaum in 1952

that attempted to identify factors underlying the decision

to obtain a chest X-ray for the detection of tuberculosis.

This study assessed specific beliefs about the disease of

tuberculosis and related them to the decision to seek an

X-ray as a mechanism for detection of the disease. Two

specific beliefs were measured, one detailing the

respondent's belief that they were susceptible to







tuberculosis, the other their belief in the overall

benefits of early detection. Those having both beliefs

were four times as likely to have received a chest X-ray

during a specified period as those who had neither belief.

These findings spurred much subsequent research and the

delineation of a full model for the organization and

identification of individual beliefs.

It was not until the mid-seventies that the structure

of the Health Belief Model had stabilized into the form

that it has retained to the present. By then it had gained

currency as a tool for explaining health behavior, with a

review of applications of the model finding a durable

pattern to the results of initial investigations (Becker &

Maiman, 1975). Examining the testing of the Health Belief

Model components, the authors found that the relevant

beliefs reliably correlated with behavior. The perceived

level of susceptibility was found to correlate positively

with the performance of health behavior, which is defined

as any action taken by an individual who perceives himself

to be healthy in order to prevent the occurrence of

disease or detect it in and asymptomatic stage (Kasl &

Cobb, 1966). Behaviors ranged from cancer screening to

obtaining immunizations to dental visits.

The perceived severity of the conditions to which the

individual feels vulnerable was found to have a weak

relationship in early reviews of research (Kirscht, 1974;

Becker & Maiman, 1975), with some studies finding that a








higher degree of perceived severity predicts participation

in dental care, responding to disease symptoms, and

working to prevent accidents. No significant correlation

of perceived severity with target behaviors was found in

many of the other studies, but the authors of the main

review (Becker & Maiman, 1975) saw the component as

contributing valuable information to the overall

predictive value of the model.

The elements of the model found to have the greatest

correlation with behavior in these early reviews were the

perceived benefits of the health protective behavior being

researched and the perceived barriers to or costs of

performing that behavior. For the benefits construct,

positive, statistically significant correlations were

usually found between perceived efficacy of a preventive

health action and the performance of that action,

including obtaining immunizations, getting screening for

cancer and tuberculosis, and seeking prophylactic dental

visits (Becker & Maiman, 1975). Similarly, barriers/costs

were found to be inversely associated with involvement in

preventive behaviors. Barriers identified as likely to

interfere with carrying out behaviors were monetary cost,

complexity of the regimen, side effects, degree of change

needed, and accessibility of facilities. The health

motivation component of the Health Belief Model was








proposed in 1975 (Becker & Maiman, 1975); results of later

investigations will be reported below.

A survey of research utilizing the Health Belief

Model found sufficient evidence to uphold the components

as distinct, separate areas of beliefs and as having

significant correlation to many differing areas of

behavior (Becker, Haefner, et al., 1977). This survey

systematically examined each study for evidence of

significant associations between qualities assessed by the

components and subsequent behavior in both the preventive

and sick-role realms. Both the preventive behavior studies

and the sick-role studies used prospective and

retrospective designs.

A total of twelve preventive behavior studies were

evaluated. The behaviors being examined included seeking

dental visits, getting Pap tests, getting polio

vaccinations, getting examinations for a number of

different potentially harmful conditions, and getting

screened for Tay-Sachs disease. The authors noted the

number of studies that reported results for each component

of the model, allowing construction of a "significance

ratio". This ratio compared the number of times a

significant correlation of beliefs with behavior was found

with the number of times it was tested, with significance

set at the .05 level. For perceived susceptibility, the

ratio was 9 out of 10; perceived severity, 6 out of 8.








With the constructs of perceived benefits and barriers,

the ratios were 6 out of 9 and 3 out of 4, respectively.

Seven sick-role studies were charted in the same

manner, showing that each reported result achieved

significance for the four components. The behavior focus

in these studies was primarily penicillin prophylaxis or

regimens recommended as a result of findings of disease,

with one study investigating use of followup care after a

negative health finding. Perceived susceptibility was

examined in three of the studies, with perceived severity

targeted in four. For perceived benefits the number was

four, and perceived barriers, one (Becker, Haefner, et

al., 1977).

A research team performed a study assessing Health

Belief Model component beliefs specific to compliance

issues, in an attempt to explain the different responses

of low income mothers of children with otitis media

(Becker, Drachman, & Kirscht, 1974) Their work was

successful in differentiating compliers with medical

regimens from non-compliers, as measured by a complex

behavioral protocol. The study was intended as an initial

step in a program to improve compliance through changing

health beliefs. While the composite model was found to be

a better predictor, their conclusions stated the aim of

such research in behavioral change "Thus, by knowing which

model components are below a level presumed necessary for







compliance, the health worker may be able to tailor

intervention to suit the needs of the individual."

(Becker, Drachman, & Kirscht, 1974, p. 215). Although the

focus in the current study is on self-initiated changes as

opposed to compliance, the mechanism of action of beliefs

upon behavior was expected to be the same.

Cues to action, a belief construct that was part of

the original formulation of the Health Belief Model, has

proven to be difficult to research. Cues are hypothesized

as stimuli or events that trigger the individual to act

once the other model components have provided the

motivation and direction of the action. Cues may be

internal, such as perceptions of bodily state, or

external, encompassing many different kinds of messages

from media or significant others. Health Belief Model

researchers had early general agreement about their

importance to the outcome of a behavioral intention.

After some attempts to identify salient cues in

investigations of Health Belief Model factors, the

conclusion reached by one of the originators of the

component was that most settings preclude obtaining an

adequate measure of the contribution of this variable to

behavioral outcomes (Rosenstock, 1974a). Without some way

of recording the cues as they occur in the life of each

subject, (e.g. chest pain, or a public health poster about

the need to seek vaccinations) too much is lost when

asking for recall of relevant, possibly related cues. This









construct quickly faded from the Health Belief Model

research method. With the development of new tools and

methods for adequate measurement of cues in future

researches, the variable may well be revived as a relevant

part of the Health Belief Model.

When Health Belief Model components were compared to

other explanatory models of health behavior, many of the

common elements among the models turned out be wholly or

in greater part contained within Health Belief Model

factors (Cummings, Becker, & Maile, 1980). The six

subsuming factors found to be organizing elements for the

theories examined in the study were: 1) accessibility to

health care, 2) evaluation of health care, 3) perception

of symptoms and threat of disease, 4) social network

characteristics, 5) knowledge about disease, and 6)

demographic characteristics. Thus the Health Belief Model

has links with the breadth of health behavior research,

with the only major omission of the model being a part

that explicitly addresses social network characteristics.

In that category, it appears that descriptors of such

characteristics would fit in the benefits component of the

model.

Studies incorporating the Health Belief Model have

proliferated following the initial evaluative research of

the seventies (e.g. Cummings, Jette, & Rosenstock, 1978)

with many different applications developed and







investigated. Although the methods used became more

sophisticated, the tendency for each researcher to use

instruments specific to their own studies continued. This

practice resulted in refinement of statistical technique

to be applied in this kind of research, but made

comparative conclusions difficult because of the

differences between instruments or constructs. This trend

has been bemoaned by researchers centrally involved with

the Health Belief Model, (Janz & Becker, 1984) but the

pattern continues even in the most recent research.

A major shift has occurred in the area of application

of the Health Belief Model, toward investigating

preventive behaviors and with a corresponding lessening of

concern with sick-role behaviors. The first use of the

methodology in connection with exercise was in 1980

(Riddle, 1980), with most intervention studies using the

Health Belief Model having been performed primarily in the

past decade. A close examination of them follows.

Illustrating the durability and continued utility of

the Health Belief Model methodology is a study by Chen and

Tatsuoka (1984) that measured the impact of beliefs on

preventive dental behavior. Preventive dental behavior

includes dental visits, brushing, and flossing of teeth.

Patterned after early, similar research, the study used

canonical correlation, a measure of association especially

appropriate for evaluating the degree of contribution the

Health Belief Model components make in predicting








different preventive dental behaviors. Conducted using 685

married Caucasian women as subjects, the study assessed

beliefs and behavior simultaneously in a cross-sectional

design, using a questionnaire format. Overall success or

failure to engage in preventive dental health behavior was

significantly related to health beliefs (rc=0.436,

P < 0.001). Among the separate Health Belief Model

components, the ones having the greatest predictive value

with preventive dental behavior, as measured by structure

coefficients, were perception of barriers and perception

of benefits. Another element highly predictive of behavior

was a construct examined in the study by Chen and Tatsuoka

(1984), perception of salience of beliefs. Much lower

coefficients were obtained for perception of

susceptibility and perception of severity.

The study by Chen and Tatsuoka (1984) is

representative of the method and general findings in

modern Health Belief Model research. The perceived

barriers component appears to be the construct of the

Health Belief Model holding the strongest association with

preventive behavior, as measured by findings of

significance in Health Belief Model investigations (Janz &

Becker, 1984). General findings differ from the dental

study in that perceived susceptibility was the second

strongest association overall, with perceived benefits

third.









One of the limitations of the Health Belief Model has

been that a causal relationship between beliefs and

subsequent behavior has been difficult to establish

(Cummings, Becker, & Maile 1980). This difficulty arises

from the need to apply the theory using knowledge about

the population, about the setting, and about the specific

behavior under investigation. Isolating the relationship

of a particular personal feature such as beliefs to a

behavioral outcome is made difficult by natural limits on

the amount of information obtainable about

person-environment interactions. Thus, statements about

the variables examined in Health Belief Model

investigations have typically been couched in terms of

correlations and coefficients, with some few studies able

to draw limited conclusions about causality based on

controlled interventions.

The most comprehensive assessment of Health Belief

Model research that has been performed since the model was

organized as a foundation for investigating health

behavior was that of Janz and Becker (1984). The authors

note many empirical findings supporting the ability of all

the model components to achieve significance as

explanatory and predictive mechanisms. Even the component

of the model found to be suspect in its utility in earlier

reviews, that of perceived severity, was upheld as a valid

and reliable measure throughout its many contexts of








application. As a model, the Health Belief Model has shown

more activity and corroboration than any other model

addressing health-related behavior. Within this framework,

assessing the role of beliefs in mediating or determining

health behavior has been a primarily retrospective,

cross-sectional undertaking.

Researchers' limited resources continue to hinder

studies thorough enough to enable firmly ruling out

competing or confounding explanations with some of the

model components, but a solid base for ongoing research

and applications within the health professions has been

established. Longitudinal studies that would allow firmer

statements about causality are costly, with few occurring

in such beliefs research. Prospective studies that have

been performed have shown levels of significance

comparable to retrospective studies, showing promise for

this method of research on the Health Belief Model (Janz &

Becker, 1984).

Experimental studies based on the Health Belief Model

have shown the impact of health beliefs on behavior to be

a useful, potent phenomenon in behavior change

manipulations. Interventions using Health Belief Model

components as a basis for determining the degree of change

in beliefs have met with success. Using beliefs change as

the mechanism to change certain target health behaviors,









researchers have documented subsequent differences in the

level of the target behavior.

In a study investigating the effect of health threat

communications on preventive dental behavior, Beck and

Lund (1981), assigned subjects to four groups. Each group

got an educational slide show about peridontal gum disease

with a stated level of susceptibility to the disease (high

or low) and the alleged severity of the disease if it were

to occur (again high or low). This assignment process

formed four cells of different combinations of

communications, allowing the experimenters to assess the

impact on subsequent beliefs, flossing intentions, and

flossing behavior. Patients who had a higher level of

perceived seriousness of gum disease as a result of

viewing the communication subsequently increased in their

intention to floss as well as their actual frequency of

flossing, showing significant differences from the low

perceived seriousness group. Patients with high

susceptibility were more likely to floss more often,

though the differences were not significant.

Other investigations have demonstrated the efficacy

of interventions designed to affect peoples' health

beliefs and thus behavior. A major review study about

research on the modification of beliefs and thus behaviors

(Kirscht, 1974) found many efforts to utilize the

situational and behavioral specificity of the Health








Belief Model in devising experimental regimens. Most of

the studies reviewed used communications to heighten

threat or arouse fear on the part of the recipient, based

on the premise that the resulting higher degree of

perceived severity or perceived susceptibility would

result in a greater likelihood that the behavior would

change as a result. Results were complex, as might be

expected in a multivariate model, but generally supported

both the Health Belief Model and the effectiveness of the

change regimens. Important mediating effects found to have

an impact on the results were the presence of a response

option perceived as effective in reducing the threat and

other beliefs about the new behavior.

In a test of the effects of interventions targeted on

behavior change, Becker, Maiman, Kirscht, Haefner, and

Drachman (1977) performed a study in which mothers of

overweight children were exposed to fear-arousing

communications based on Health Belief Model dimensions.

Measures of subsequent appointment-keeping and children'

weight loss showed both significant independent effects of

the fear-arousing messages and significant correlations of

the Health Belief Model variables with the behaviors. This

experiment demonstrated the usefulness of the Health

Belief Model as a whole through several multiple

regressions, along with demonstrating that the influence









of the fear-arousing messages was independent of the

impact of beliefs.

Kirscht (1983) reports mixed results later from

Health Belief Model investigations utilizing belief-change

interventions. An outcome of adoption of the target

behavior depended, generally, on the simultaneous presence

of perceived susceptibility and severity, and on the

behavior in question being perceived as beneficial and

with few accompanying barriers or costs. Thus, the

employment of the Health Belief Model in devising

strategies for behavior change through persuasive

communications requires the recognition that all the

components play a role in the final outcome. A narrow

focus utilizing only one component is susceptible to

misapplication and misinterpretation of the results. When

correctly utilized and with an appreciation for

uncontrolled variables, interventions based on the

components of the model appear effective in altering

behavior. Kirscht (1983) cautions that too few studies

exist that have examined long-term changes in behavior for

definitive conclusions to be possible.

Applying the Health Belief Model
to the Exercise Adoption Problem

The Health Belief Model has proven to be a successful

tool for predicting the likelihood and frequency of many

different kinds of preventive health behaviors. Another

potential application receiving increasing research








attention has been in evaluating the effectiveness of

model-based interventions designed to positively affect

personal behavior patterns. A tool for explaining,

predicting, and changing health-related behaviors offers

potential utility in the specific case of aerobic

exercise. The need for mechanisms to support exercise

adoption is easily documented. In a review of exercise

participation, Powell (1988) found that only 9% of the

population in the U.S. exercises at the level necessary to

derive the benefits exercise has to offer.

Current patterns in exercise research show widespread

use of the findings of the American College of Sports

Medicine (1978) on the minimum level of exercise necessary

to develop and maintain fitness. These guidelines specify

the following parameters for exercise quantity and

quality: 1) frequency of training 3-5 days per week, 2)

intensity of training 60%-90% of maximum heart rate

reserve, 3) duration of training 15-60 minutes of

continuous aerobic activity, with duration dependent on

intensity, and 4) a mode of activity that involves large

muscle groups, that can be maintained continuously, and is

rhythmical and aerobic in nature (American College of

Sports Medicine, 1978, p. vii). Given the almost universal

employment of all or parts of these guidelines in fitness

programs and in evaluative research, the current study was

formulated to examine the association of Health Belief







Model variables with exercise conforming to the American

College of Sports Medicine parameters.

With a 50% dropout from aerobic exercise programs at

six months (Dishman & Gettman, 1980), clearly a reliable

regimen for assisting exercise adoption would play a

useful role in such programs. The utility of such a

program would apply whether the target was an individual

or a group, whether the exercise adoption goal was

self-motivated or prescribed. The development of such a

regimen is dependent on adequate knowledge of what changes

are being attempted. Dishman (1988) has observed, "one

barrier to implementing public health promotions of

exercise as well as uniform interventions in exercise

programs is the absence of a consensus over the methods

that might be effectively employed and the exercise

determinants to be targeted" (p. 2). Research on exercise

adoption must contain attempts to identify the key

determinants and begin to apply them in experimental

trials to determine their utility in effecting change.

Hughes (1984) cites the need for more research to

identify relevant characteristics of individual exercisers

and to identify the best method to prescribe exercise as

treatment. Such a base of knowledge is needed to develop

the method for prescribing exercise as well as the best

ways of insuring positive outcomes once the program has

been undertaken. The need to take into account individual

factors affecting exercise adoption, and the challenge









therein, was cited in discussion of a recent attempt to

isolate factors influencing exercise adherence (Gale,

Eckhoff, Mogel, & Rodnick, 1984). The authors noted the

complexity and the difficulty in distinguishing different

individual characteristics while controlling for other

factors, observing that the phenomenon of adherence is

likely the product of interaction between personal and

program characteristics. Along with many other writers in

the exercise adoption area, they cite the need for much

more research in this area before a reliable program for

change can be developed.

As regards exercise and other forms of preventive

health behavior, calls for more research on social and

psychological predictors of health-related behavior have

appeared in different quarters, emphasizing the paucity of

experimental research that could provide the foundation

for programs of change (McAlister, Farquhar, Thoresen, &

Maccoby, 1976; Becker, Haefner, et al. 1977). For the

example of aerobic exercise, additional steps are needed

to identify specific beliefs (and the degree to which each

are held) that are associated with successful and

unsuccessful attempts at exercise adoption. A reliable

method for determining relevant beliefs would provide for

the development of effective psycho-educational and belief

change regimens to augment the known precursors of

behavioral change.









An antecedent of the current interest in the possible

contributions of the Health Belief Model to the

explanation of and support of exercise adoption behavior

was contained in an earlier discussion of the mechanisms

of the model:

Surely, the exercise and dietary mania observed
over the last decade represent behaviors that could
be regarded as striving toward improved health, but
it is just as easy to explain them (insofar as they
are health related at all) as behavior undertaken to
avoid a deleterious situation. Again, there are
individuals who exercise and engage in other health
actions having health implications but who do so for
reasons quite unrelated to health, perhaps for
aesthetic reasons or for the sheer exhilaration felt
by many by the performance of physical work. Again,
the question of whether the avoidance orientation in
the Health Belief Model is adequate to account for
the so-called positive health actions taken by people
remains unresolved.
(Rosenstock, 1974a, p. 335)

Although written some time ago, the uncertainty voiced in

the discussion remains today, with only a few exploratory

efforts having been undertaken to answer the questions

raised. Such research serves as the foundation and

guidance for specifying elements of the present study.

Reviewers of recent research have verified the

relationship of beliefs and attitudes to intentions to

exercise and the exercise behavior itself (Godin &

Shepard, 1986). In examining research up to now in the

area of exercise behavior, Godin and Shepard support the

theory of Ajzen and Fishbein (1977), noting that recent

attitude-behavior research has already proceeded from "is"

questions (Is there an effect?) to "when" and "how"








questions. The second brace of questions, when answered,

will provide information about when there is a relation

between attitude and behavior plus how the effect of

attitudes on behavior is mediated. Godin and Shepard

(1986) lobby for the application of theory to this problem

as opposed to the "shotgun" approach of trying out

variables in varying combinations.

The authors who have stimulated much of the recent

interest and activity in attitude-behavior work did so by

distilling and thematizing much of the research that had

occurred up to the time of their extensive review (Ajzen &

Fishbein, 1977). Of particular interest was the finding,

cited above (pp. 53-54), about the patterns of

correspondence between attitudes and behavior. The result

has been studies that test the relationship in specific,

particular instances of behavior as well as the more

global, broad questions. Exercise provides the opportunity

for both kinds of research, from the likelihood of doing a

particular form of exercise at a particular time to

general adherence over time.

Riddle (1980) devised a test of attitude-behavior

relations for a particular form of exercise--jogging--and

utilizing one of the main models of the relationship

between attitudes and behavior. Choosing Fishbein's

Behavioral Intention Model, she developed a beliefs

measure using an elicitation technique that resulted in a







68 item measure. With the criterion behavior identified as

regular jogging, the survey assessed behavioral intention,

attitude toward behavior, attitude toward the object,

subjective norms for behavior, motivation to comply,

consequences of behavior, and evaluation of the

consequences. The focus of the study was on testing the

validity of the Behavioral Intention Model, but clearly

the Health Belief Model methodology served as a guide in

the construction of the beliefs items on the survey. The

survey was administered to 369 men and women of age 30 and

over, with 296 usable surveys received. Joggers numbered

149, non-exercisers totalled 147.

Results of the research, confirmatory of the

Behavioral Intention Model, also showed marked belief

differences between joggers and non-exercisers. Riddle

(1980) documented differences between joggers and

non-exercisers on beliefs corresponding to two Health

Belief Model dimensions, perceived benefits and perceived

barriers. On 17 of the 19 belief scales, differences

between joggers and non-exercisers on the

benefits/barriers items were significant at the p < .001

level. The items were structured as follows: "Taking part

in regular jogging in the next two weeks would...".

Examples of the items showing significant differences are:

"make me feel too tired," "take too much time," "make me

feel good mentally," "make me have a fatal heart

attack/stroke." Beliefs about the consequences of exercise








in the form of regular jogging accounted for much more of

the variance than did evaluation of the consequences.

Joggers had strong beliefs about the positive consequences

of jogging, while nonexercisers had neutral beliefs about

the positive and negative consequences. Joggers thought

regular jogging would benefit their physical and mental

health, while non-exercisers thought it would require too

much time and discipline and make them too tired.

In another study utilizing techniques suggested by

Fishbein's Behavioral Intention Model, Slenker, Price,

Roberts, and Jurs (1984) established a promising

precedent for the employment of the Health Belief Model

with regard to predicting aerobic exercise adoption. Their

investigation applied the model in a specific exercise

situation, that of jogging. An instrument was developed

and validated for purposes of measurement of beliefs

specific to expectations about jogging, for the five

customary belief areas examined in Health Belief Model

research: perceived benefits, perceived barriers,

perceived susceptibility to disease, perceived severity of

possible disease conditions, and general health

motivation. Other variables under investigation were also

part of the instrument, including knowledge about jogging,

complexity of jogging, cues, and health locus of control.

The completed instrument was administered to 40 joggers

and 39 nonexercisers.








A composite regression analysis of Health Belief

Model variables that included one other variable accounted

for 56% of the variance in jogging behavior. The Health

Belief Model variables of perceived benefits, perceived

barriers, and general health motivation comprised a

multiple R of .707, or 50% of the total variance. The

single variable perceived barriers accounted for 40% of

the variance. The composite figure was a strong

performance of the Health Belief Model variables in

explaining exercise behavior. Given the careful instrument

development and sound statistical analysis, the finding is

one of the most impressive performances of a Health Belief

Model-based instrument for any study examined. Using a

discriminant analysis procedure, 92 percent of the

subjects were correctly assigned to their proper category

of jogger or nonexerciser based on the independent Health

Belief Model variables assessed.

Similar methods with an adolescent population yielded

more equivocal results. The predictive measures produced

more modest percentages of correct classification and the

amount of variance accounted for by the measure employed

for the study was also less (O'Connell, Price, Roberts,

Jurs, & McKinley, 1985). Performed on a sample of 69 obese

and 100 nonobese high school freshmen and sophomores, the

study used the beliefs measures to predict status of obese

or nonobese and to predict membership as an exerciser or a







nonexerciser. The beliefs assessed differed from the study

performed by Slenker, Price, Roberts, and Jurs (1984) in

that measures of social approval for dieting and exercise

were added, along with cues for the behavior in question

(dieting or exercise). General health motivation was not

assessed, while the other Health Belief Model

variables--susceptibility, severity, barriers, and

benefits--remained. Each of the four variables also

obtained a measure of each of the subject's beliefs about

the behaviors of diet and exercise.

For obesity, the independent variables employed in

the study correctly classified 69% of the subjects as

obese or nonobese in a discriminant analysis procedure.

The authors observe that obesity is not a behavior,

identifying a likely reason for the relatively poor

performance of their predictors. When employed to classify

dieters versus nondieters, 83% of the subjects were

correctly classified. When the discriminant analysis

procedure was applied to exercisers versus nonexercisers,

75% of the overall classifications were correct.

The amount of variance accounted for by the

independent variables in the case of dieting behavior for

obese and nonobese adolescents was 23% and 19%,

respectively. For exercise behavior, 14% of the variance

was explained for obese subjects, while none of the

variables emerged as significant in predicting exercise

behavior for the nonobese subjects. These weak findings in








the application of the Health Belief Model to diet and

exercise behavior are noted by the authors as an indicator

that age may be an important factor in assessing health

behaviors using the Health Belief Model. Such an

observation is supported by Weinstein (1984), who found

that college students have an unrealistically low

assessment of their susceptibility to health problems and

accidents.

A study undertaken to determine the relationship of

Health Belief Model components to compliance with an

exercise prescription found that the beliefs of coronary

bypass patients played a role in subsequent adherence to a

walking exercise regimen (Tirrell & Hart, 1980). Five

standard Health Belief Model components were assessed,

perceived benefits, perceived barriers, perceived

susceptibility, perceived severity, and general health

motivation. Using a somewhat abbreviated measure

consisting of 19 items, the researchers found that two of

the components had significant correlations with compliant

behavior, perceived barriers and perceived susceptibility.

The correlation of the barriers measure with participation

in the "heart walk" regimen was 0.64; with the

susceptibility component the correlation was 0.35. The

highest degree of significance was achieved with the

perceived barriers component, which was significant at the








p < .001 level; perceived susceptibility was significant

at the p < .05 level of significance.

In a study designed to relate two theories of

behavior change, Sonstroem (1987) provides more

verification of the role of beliefs in behavior change.

Based on Ajzen and Fishbein's Theory of Reasoned Action

(1977), the study examined belief statements of subjects

with the aim of differentiating between various stages of

change as detailed by Prochaska and DiClemente (1983).

While the study did not explicitly base the beliefs survey

utilized on the Health Belief Model, the categories of

beliefs that emerged from the analysis were coincident

with the categories utilized in most Health Belief Model

studies.

A background questionnaire determined the exercise

status of each subject, allowing them to be assigned a

category precontemplationn, contemplation, action,

maintenance, relapse) in the Prochaska and DiClemente

stage schema. Two hundred fourteen respondents were asked

to react to belief statements that had been supplied by

exercise leaders at various exercise programs. Likert

items providing a graded response accompanied each belief

statement. Each statement had the root "My participation

in a regular program of exercise would:" followed by the

statement. Examples of the belief statements are: "help me

in controlling my weight"; "increase my cardiovascular







endurance"; "be boring". Principal component analysis on

the resulting responses yielded seven usable components,

labeled Benefits, Barriers, Novelty Outlet, Social Outlet,

Fear, Appearance, and Psychological Gain.

A stepwise discriminant function analysis utilizing

the belief findings as a base was constructed to classify

the expected Prochaska and DiClemente category for the

precontemplation, contemplation, action, and maintenance

stages. Mean item responses were computed for each subject

for each component, which were then standardized to mean

of 50 with a standard deviation of 10. Two functions which

accounted for 99.99% of the total Eigen values resulted.

The three components which entered the equation were

Barriers, Benefits, and Psychological Gain. For the two

functions constructed, 60% and 55% of the categories were

correctly identified when compared to the subjects' actual

status. For research of this type, the percentage of

subjects correctly assigned was relatively low. Sonstroem

(1987) cites a small sample size in some of the stage

categories as a limitation, along with limitations in

instrument development.

A survey of research investigating the utility of the

Health Belief Model quickly reveals that almost as many

different instruments exist as do studies! While there are

many idiosyncratic measures that have been employed in

beliefs research, there are many parallels between

studies, and a consensual choice of method. A consistent








call in reviews of Health Belief Model research, however,

is for a standardization of instruments and, where

possible, in methods. Little Health Belief Model research

is designed to replicate or refine earlier findings, nor

are the positive findings that result later employed to

design interventions that may provide for support of

individuals attempting behavior change programs.

Support for the reliability and validity of properly

designed and administered instruments has recently

emerged, documenting that such research tools surpass

minimum requirements for such measures. A number of

studies examining appropriate techniques for constituting

and evaluating measures of health beliefs have validated

the different beliefs posited for the Health Belief Model

as discrete categories that do not overlap. Construct

validity for the commonly researched components of the

model has been consistently demonstrated, with compelling

statistical evidence supporting the independence of the

constructs (Maiman, Becker, Kirscht, Haefner, & Drachman,

1977; Cummings, Jette, & Rosenstock, 1978; Jette,

Cummings, Brock, Phelps, & Naessens, 1981; Given, Given,

Gallin, & Condon, 1983; Champion, 1984). The utility of

the combined components of the model in predicting and

explaining behavior has already been noted above. An

important byproduct of this research activity has been the








development of an item pool where the elements have been

found to be valid and reliable.

The most complete test of a single Health Belief

Model instrument intended for use in the measurement

process was performed by Champion (1984). Designed to

verify the validity of scales for use in predicting breast

self-examination behavior, the investigation subjected

each scale to an internal consistency examination, a

test-retest reliability assessment, a test of exclusivity

for each scale, and an overall measure of construct

validity as demonstrated by the performance of the

instrument in predicting breast self-examination behavior.

The scales examined included perceived benefits of and

barriers to performing breast self-examinations, perceived

susceptibility to disease, perceived severity of possible

disease conditions, and general health motivation. The

final scales used in statistical analysis had between five

and twelve items each. The instrument was administered to

301 women of varied ages, education, ethnicity and

socioeconomic status.

The five scales had internal consistency reliability

coefficients (as determined by computing a Cronbach's

Alpha for each scale) ranging from 0.60 to 0.78. Three of

the scales had coefficients of 0.76 or higher:

susceptibility, severity, and barriers. When test-retest

reliabilities were calculated, all the scales except







benefits exhibited coefficients of 0.76 or higher. The

author observes that the testing process may have

sensitized subjects to the benefits of breast

self-examination, making for lower, but still significant,

test-retest reliability.

A principal component analysis utilizing orthogonal

rotation with a varimax criterion showed the five scales

to be mutually exclusive, with only one of the resulting

factors having items from more than one of the scales. The

severity scale lacked unidimensionality, with distribution

of the items from that scale across three factors

resulting. When scale scores were used to predict breast

self-examination behavior, a multiple R of 0.51 was

obtained, accounting for 26% of the variance. The barriers

scale accounted for the largest portion of the variance,

with health motivation second. The remaining three

variables did not account for a significant portion of the

variance.

Champion (1984) saw the resulting instrument as a

good basis for application with other behaviors,

concluding that the method and the model are of sufficient

validity for application in many areas. The study

represents an important advance in Health Belief Model

methodology, providing a useful framework and point of

departure in developing a standard procedure for beliefs

assessment. Champion assigned importance to the strategy

of utilizing previously researched items as a mechanism








for standardizing research results. Evaluative criteria

generally accepted as standards in test construction but

seldom employed in Health Belief Model investigations are

also important elements of the study.

Other studies using similar tests with instruments of

like construction report results of similar magnitude.

Wagner (1983) found that the Cronbach's alpha reliability

coefficients for her Health Belief Model scales ranged

from 0.65 up to 0.91, with all but one of the scales above

0.76. Her scales employed Health Belief Model components

in ways specific to medical applications that assessed the

psychosomatic elements of beliefs and behaviors. Using a

factor analytic approach to evaluate Health Belief Model

components yielded inconclusive results as to the unitary

dimensionality of the usually robust barriers and benefits

scales. This appears to be due to the inclusion of a wide

range of items representing too many different activities

and symptoms.

Another study that also did an evaluative appraisal

of an instrument development effort found distinct

components and moderate reliabilities (Jette, Cummings,

Brock, Phelps, & Naessens, 1981). These results were

achieved despite the use of only a few items in each

scale. A factor analysis yielded a concordance with

existing Health Belief Model constructs of perceived

barriers, perceived severity of disease, and concern for







health. These results are of interest given that they were

noted in a study that primarily investigated behavior

associated with medical regimens. Even without items

intended to measure the Health Belief Model components,

the factors were identified. Examination of the

methodology of instrument construction and use in Health

Belief Model studies led to the conclusion that

"moderately reliable indices of a wide spectrum of health

beliefs can be constructed and replicated across samples."

(Jette, Cummings, Brock, Phelps, & Naessens, 1981, p. 92).

The authors further note that the obstacles to

replicability of results stem from the lack of

standardization of instruments and of sufficient rigor in

the construction of beliefs measures.

The previously cited study by Slenker, Price,

Roberts, and Jurs (1984) (p. 82) also serves as a good

foundation for the present study. It found the Health

Belief Model to be both valid and reliable in the specific

exercise application of jogging. As with the Champion

(1984) study, measures of internal consistency and

construct validity were applied. The five Health Belief

Model scales of perceived benefits of jogging, perceived

barriers to jogging, perceived susceptibility to disease,

perceived severity of possible disease conditions, and

general health motivation achieved KR-20 reliability

coefficients of 0.83 or greater with the exception of

general health motivation, which has a coefficient of









0.57. Given that the KR-20 procedure is intended for

dichotomous variables, it is unclear what a more

appropriate analysis would yield in evaluating the

components. Factor analysis employing an orthogonal

varimax rotation revealed the five constructs to be

distinct factors.

Of the possible survey/questionnaire mechanisms for

obtaining an accurate measure of an individual's beliefs,

the most frequently used method in Health Belief Model

research has been the Likert scale. The seven-point Likert

method of assessing beliefs has been found to have

convergent validity with interview and multiple choice

methods of assessment, two other popular assessment

strategies. Of the methods compared, the Likert method

showed the greatest validity, with zero method effects as

opposed to 10% for both multiple choice and interview

methods (Cummings, Jette, & Rosenstock, 1978). Likert

items have been the assessment method employed in most

Health Belief Model studies, with consistent results.

Lau, Hartman, and Ware (1986), performed a study

examining the impact of health values on certain health

protective behaviors. In a population of university

students, those with higher valuing of health were more

likely to report doing the behaviors, such as seat belt

use, breast self examination, and exercise. Those with

higher health-as-a-value scores were significantly more







likely to have done the behaviors than those with low

scores. Also, the authors cite an increase in

health-as-a-value scores with increasing age within their

subject pool. These values stabilized as students moved

into adult age ranges, at a level still somewhat lower

than in the adult population.

The finding that age is an important factor to be

taken into account in the measurement of beliefs and their

impact on subsequent behavior corroborates a finding cited

above (O'Connell et al., 1985). To avoid confounds in

developing instruments and interventions utilizing the

Health Belief Model, the use of adults of 25 years of age

or older in preliminary studies appears warranted. Age 25

appears to clear by a safe margin the age range where

detectable changes in beliefs are occurring.

Hypotheses

The Health Belief Model has shown a consistent

capacity for prediction with many different kinds of

health behaviors, as well as an ability to account for

significant amounts of the variance in retrospective and

prospective health behavior investigations (Janz & Becker,

1984). The pattern in the choice and application of

instrumentation has been variable, with different methods

adopted according to the particular perceived needs of

each separate research effort. With the recent spate of

research, particularly studies applying the model to the

area of exercise behavior, study design has begun to








stabilize into a well-documented and proven methodology

with a foundation in accepted assessment practice.

Extending Health Belief Model research into the area

of exercise behavior is a step indicated by many other

sick-role and preventive health behavior studies, and

shows good potential utility for the specific example of

general aerobic exercise. No empirically verified,

replicable link has been established between the major

Health Belief Model components and systematically defined,

general aerobic exercise as a dependent variable. No

prospective studies have examined the ongoing impact of

beliefs in this area. The success of intervention

investigations in the area of aerobic exercise depends on

the existence of a reliable, repeatable measurement

instrument that provides a basis for assessment of

effects.

Interaction between beliefs and behavior precludes

absolute statements of causality, but repeated measures

with a proven instrument provides a basis for first steps

in defining the evolution of individual patterns of

behavior. As was observed early in the first burst of

research activity to use the Health Belief Model, "the

hypothesis that behavior is determined by a particular

constellation of beliefs can only be tested adequately

where the beliefs are known to have existed prior to the

behavior that they are supposed to determine" (Rosenstock,








1974b, p. 362). Such findings hang on the existence of a

tool with known characteristics and validity to as

accurately as possible determine belief constellations

specific to the area being researched.

The present study sought to develop and verify the

attributes of just such an instrument, with regard to a

carefully delineated yet general (many different kinds of

exercise) definition of aerobic exercise. Following the

evaluation of the validity, reliability, and exclusivity

of the constructs, a test of the predictive power of the

overall measure over a time period of one month was

attempted.

The following hypotheses were advanced:

1. The five Health Belief Model constructs of
perceived benefits of aerobic exercise, perceived barriers
to aerobic exercise, perceived susceptibility to disease,
perceived severity of possible disease conditions, and
general health motivation, as operationalized by subscales
of the Personal Beliefs Questionnaire, will be found to be
reliable when evaluated by computed internal reliability
coefficients and test-retest statistics.

2. The five Health Belief Model components, as
measured by the Personal Beliefs Questionnaire, will be
found to be valid, independent constructs when assessed by
measures of intercorrelation and factor analysis
statistics.

3. The refined composite of health belief components,
resulting from item analysis of the Personal Beliefs
Questionnaire and utilizing the five belief components
employed in this study, will account for a significant
portion of the variance in classification of self-assigned
aerobic exercise adoption categories.

4. The refined composite of health belief components
from the Personal Beliefs Questionnaire will account for a
significant portion of the variance determining class of
actual activity/exercise participation as measured by
subjects' self-report.





89


5. The refined composite of health belief components
from the Personal Beliefs Questionnaire will successfully
predict membership in subjects' self-assigned categories
of aerobic exercise adoption.

6. The refined composite of health belief components
from the Personal Beliefs Questionnaire will successfully
predict membership in classifications of actual
activity/exercise participation as measured by subjects'
self-report.

7. The refined composite of health belief components
from the Personal Beliefs Questionnaire will successfully
predict self-reported changes in subjects' exercise status
occurring over the period of one month.













CHAPTER 3
METHOD

Subjects

Participants in the study were male and female adult

faculty at a mid-size college in central New Jersey. The

sample was drawn from a pool of all full-time, on-campus

faculty actively teaching during the Spring, 1990

semester. To avoid confounds from maturation effects noted

in past investigations with younger populations,

(O'Connell et al., 1985; Lau et al., 1986), only

individuals over 25 years of age were utilized in the

study. A convenience sample of 334 faculty at the college

was selected from those in the available pool.

Those faculty members selected were sent a mailing

containing a letter describing the study and requesting

participation, (Appendix A) a reply/consent form (Appendix

B), the "Personal Beliefs Questionnaire" (Gage, 1990;

Appendix C), a survey return envelope, and a reply/consent

form return envelope. One month following receipt of each

subject's response to the first questionnaire, a second

mailing was sent soliciting a second response. It

contained a contact letter, the second form of the

questionnaire (Appendix D), and reply materials. This

response was sought to enable assessment of test-retest




Full Text

PAGE 1

AN EXAMINATION OF THE UTILITY OF THE HEALTH BELIEF MODEL FOR PREDICTING ADULT PARTICIPATION IN AEROBIC EXERCISE By LARRY GAGE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1990 UNIVERSITY OF r-LORI A umv .. :-:iES

PAGE 2

ACKNOWLEDGMENTS I have been fortunate to have a committee of scientist-practitioners who have had the time whenever I have asked for it, and who each by their example have inspired me. A piece of each now resides within me as I continue to develop my own identity as a counseling psychologist. Anchoring this diverse and talented group has been Dr. Harry Grater, there to support and to challenge. His guidance has made this project feasible, with encouragement that has helped to motivate me. His willingness to step outside his usual purview enabled me to mount this project; I hope the experience has been mutually edifying. Especial thanks go as well to Dr. Jim Archer, whose mentoring has been invaluable over many spheres, not just this project. His role as cochair has included prodigious energy invested in early drafts and discussions of this work. It was his critical eye that provided for such meaningful results as occur with this study. Dr. Peggy Fong provided much of the early impetus and knowledge that resulted in my choice of this topic, and her input has been very helpful. In a similar way Dr. Shae Kosch has provided much expert guidance in my development as a psychologist in the health area; her matter-of-fact imparting of substantive guidance and knowledge has been ii

PAGE 3

much appreciated. A special debt of gratitude is owed Dr. Barbara Probert, whose continuous unconditional positive regard from my first association with her forward has helped me to heal past hurts and to believe in myself. Her support has been a spring whose waters I have gratefully imbibed, even at a distance. I must acknowledge the help of Dr. Henry Wang, who helped advise about the best course to chart in wrestling with abstract data and the arcane structure of SAS. Also aiding in the process was Toby Gelman, whose support and guidance helped me sidestep some deep electronic chasms. Trenton State College, through the offices of the Psychological Counseling Services and of Institutional Computing, has provided critical resources that allowed the execution of this project. Finally, I owe deepest thanks to those whose support has enabled me to approach this threshold. My parents, Joe and Phyllis Gage, have provided support and encouragement throughout my life. My wife, Karen Forbes, has supplied substantial professional advice and consultation, over and above the many kinds of support I have also gotten from her as my life partner. She has been an in-house example of how to set the standard as a new professional, and an inspiration to me. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGMENTS ............. ........................... ii LIST OF TABLES .......................................... vi ABSTRACT ................................. Viii CHAPTER 1 INTRODUCTION ........................................ 1 The Costs of Illness and Solutions Offered by Counseling Psychology ...................... 1 Attrition From Exercise Programs ................. 6 The Health Belief Model and Its Utility In Exercise Adoption Research ................ 11 The Role of an Attitude/Beliefs Study in Exercise Adoption Research ................ 15 2 REVIEW OF THE LITERATURE ........................... 2 5 The Benefits of Exercise ........................ 25 Exercise Adoption ............................... 37 Factors in Adherence to Exercise ................ 41 The Health Belief Model and Its Role in Behavioral Research ....................... 51 Applying the Health Belief Model to the Exercise Adoption Problem ................ 67 Hypotheses ...................................... 8 6 3 METHOD .................................... ....... 90 Subjects ........................................ 9 O Instrumentation ................................. 9 3 Procedure ...................................... 1 o O 4 RESULTS ........................................... 10 4 Evaluation of the Personal Beliefs Questionnaire ............................... 104 A Test of the Predictive Validity of the Personal Beliefs Questionnaire .............. 111

PAGE 5

5 DISCUSSION . . . . . . . . . . . . . . . . . . . . 12 5 The Comparative Efficacy of the Personal Beliefs Questionnaire ....................... 125 Assessing the Criterion Validity of The Personal Beliefs Questionnaire .......... 127 Limitations Affecting the Results of This Study ....................................... 134 Implications for Future Research on the Precursors to Aerobic Exercise Behavior ..... 138 Conclusion ..................................... 14 3 APPENDICES A CONTACT LETTER FOR QUESTIONNAIRE .................. 146 B REPLY-CONSENT FORM ................................ 147 C PERSONAL BELIEFS QUESTIONNAIRE .................... 148 D SECOND ADMINISTRATION OF QUESTIONNAIRE ............ 156 E LIST OF ORIGINAL ITEMS AND ASSIGNED CODES ......... 165 REFERENCES ............................................. 1 71 BIOGRAPHICAL SKETCH .................................... 181

PAGE 6

TABLE 1 2 3 4 5 6 7 8 9 10 11 12 13 LIST OF TABLES PAGE Summary of Demographic Descriptors For All Respondents to Questionnaire One ......... 92 Coefficients Alpha for Initial Groupings of Questionnaire Items ...................... 105 Questionnaire Items Screened Due to Poor Relative Association With Subscale .......... 106 Questionnaire Subscale Test-Retest Correlation Coefficients .................... 107 Coefficients Alpha for Final Groupings of Questionnaire Items ...................... 108 Questionnaire Subscale Test-Retest Correlation Coefficients for Final Item Groupings ................................... 109 Subscale Correlation Matrix for Final Item Groupings .............................. 110 Summary of Factor Analysis Results .......... 111 Distribution of Subjects in Exercise Adoption Categories ......................... 112 Individual Items Correlating Most Highly With Adoption Stage Criterion ............... 113 Sources of Variance in Exercise Adoption Categories .................................. 114 Sources of Variance From Health Belief Model Belief Subscales in Exercise Adoption Categories ......................... 115 Sources of Variance in Exercise Behavior Categories ......................... 116 vi

PAGE 7

TABLE 14 15 16 17 18 19 20 Sources of Variance From Health Belief Model Belief Subscales in Exercise PAGE Behavior Categories ......................... 117 Sources of Variance From Health Belief Model Belief Subscales in Antipodal Exercise Behavior Categories ................ 118 Change in Total R-square For Criterion Variables From First to Second Administrations Using Independent Belief Variables Obtained in First Administration .. 119 Number of Observations and Percents Classified Into Exercise Adoption Categories .................................. 121 Number of Observations and Percents Classified Into Exercise Behavior Categories .................................. 122 Number of Observations and Percents Classified Into Antipodal Exercise Behavior Categories ......................... 123 Number of Observations and Percents Classified Into Self-Reported Change in Exercise Behavior Categories ................ 124 vii

PAGE 8

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 AN EXAMINATION OF THE UTILITY OF THE HEALTH BELIEF MODEL FOR PREDICTING ADULT PARTICIPATION IN AEROBIC EXERCISE By Larry Gage August 1990 Chairperson: Harry Grater Major Department: Psychology Prior research on successful exercise adoption has focused on programmatic and behavioral aspects of exercise programs, often omitting personal factors. This study was designed to focus on individual beliefs contributing to aerobic exercise adoption, using the Health Belief Model (HBM) as a foundation for the instrument developed. The beliefs selected for examination in this study were perceived benefits of aerobic exercise, perceived barriers to aerobic exercise, perceived susceptibility to disease, perceived severity of possible disease conditions, and general health motivation. The two purposes of the study were to develop a valid and reliable measure of beliefs related to aerobic exercise and to test the performance of the beliefs in predicting aerobic exercise participation. viii

PAGE 9

A survey based on the HBM was developed to assess respondents' beliefs about exercise and health. The questionnaire was sent twice to a convenience sample of 334 adult faculty at a mid-sized college in central New Jersey. The first administration of the questionnaire yielded 179 usable responses, with 165 received in the retest administration. The questionnaire was evaluated by an item analysis and measures of content validity, internal reliability (alpha coefficients from 0.832 to 0.935 were found for the final five scales), and test-retest reliability (r=0.813 for the total scale). Based on these tests it was concluded that the measure was reliable with indications of validity. Correlational and factor analysis tests of the independence of the beliefs constructs as operationalized in the questionnaire found some association between the scales. Multiple regressions and discriminant function analyses were computed, providing predictive validity of the beliefs measure for the criterion variables of exercise adoption stage and level of exercise reported by subjects. The independent belief variables accounted for 35% of overall variance (multiple r square) for categories of exercise behavior, 40% for exercise adoption stages in multiple regressions on these variables. The instrument performed best in distinguishing between sedentary ix

PAGE 10

subjects and aerobic exercisers, accounting for 51% of overall variance and correctly classifying 82% of the subjects. The main predictor for the criterion categories was identified as perceived barriers to aerobic exercise, corroborating similar research using HBM beliefs to explain other preventive behavior. X

PAGE 11

CHAPTER 1 INTRODUCTION The Costs of Illness and Solutions Offered by Counseling Psychology Americans spend eleven and one half percent of their gross national product for health care in all its forms ("A Solution", 1988). This figure represents a two percent increase since 1984, when concerted efforts began to limit the growth of spending, and a five and a half percent increase since 1965. In 1990, the projected share of the GNP devoted to health care is expected to be twelve percent, for a total expenditure of 750 billion dollars (Kiesler & Morton, 1988). It has been noted that "health" care actually refers primarily to interventions with an illness, not to actual promotion of health; only about two percent of U.S. expenditures on health are for disease prevention and control measures (Knowles, 1977). The rest of the budget primarily services the Medicare and Medicaid programs, programs oriented primarily toward remediation of existing illness or injury. This allocation of budget priorities misses the cause of many of the illnesses treated, allowing the genesis of most health problems to go unchanged and unaddressed. Janis, in an article detailing counselor behavior 1

PAGE 12

2 effective in supporting client behavior change, notes that "the most serious medical problems that today plague the majority of Americans and Europeans are not primarily medical; they are behavioral problems requiring the alteration of personal habits, preferences, and decisions." (Janis, 1983, p. 146). Supporting this observation, Powell (1988) cited an analysis of the causes leading to lost years of life, noting that Lifestyle factors were the largest proportion at 53 percent. Following Lifestyle, in order, were Environment (22 % ), Human Biology (16 % ), and the Health Care System itself (10 % ; note: percentages rounded). The lack of effort in the area of disease prevention and control is not for lack of knowledge about what personal health behavioral changes might be effective. Belloc and Breslow (1972), in a very large longitudinal study of over 7 000 adults, found seven health practices significantly related to health and life expectancy: 1) three meals a day at regular times and no snacking; 2) breakfast every d a y; 3) moderate exercise two or thr ee times a w eek; 4 ) a d eq uate sle e p; 5 ) no smoking; 6 ) moderate weight; 7) no a l cohol or only in mod era t i on. A 45-year-old m a n who practices 67 o f these habit s h as a life e x pect a ncy ten a nd one-hal f y ea rs long e r th a n on e w ho pr a ctices 03 o f th em ( Be lloc & Br esl o w 1 9 7 2 ). Despi t e

PAGE 13

the availability of this study and others like it, few programs to alter behavior exist. 3 Part of the problem in adopting a health promotion focus is with the lack of immediacy of results from efforts directed at changing health habits. Knowles (1977), in comparing such an undertaking to traditional public health practice, found "the knowledge required to persuade the individual to change his habits is far more complex, far less dramatic in its results, far more difficult to organize and convey--in short, far less appealing and compelling than the need for immunization" (p. 61). For the person with questionable health habits, the personal consequences of those habits are not visible in their daily accretion or change, making programs for behavior change difficult to target and execute. In a description of the World Health Organization's Health for All by the Year 2000 (HFA/2000) program, Diekstra and Jansen (1988) note a large potential role for the profession of psychology in accomplishing the ambitious goals set forth there. They specify the fields of psychological medicine, behavioral medicine, and health psychology. As HFA/2000 concerns itself chiefly with prevention and health-enhancing lifestyles, imagining its realization without psychological support is indeed difficult (WHO, 1981).

PAGE 14

4 Many possible contributions of the field of psychology with the goals of HFA/2000 can be identified. Kiesler and Morton (1988) describe such opportunities for "psychology in the public interest", citing facets of psychology that well suit it to preventive efforts in the area of health: 1) an empirical orientation; 2) expertise not traditionally allied with medicine and health care; 3) emphasis on both prevention and wellness; and 4) knowledge about the effect of the environment on behavior and well-being Such e x pertise is pivotal to the ultimate aim of changing behav i or to reduce risk factors in the population at large. Indeed, the concept of "behavioral health" has been discussed for most of this century, with some recent attempts to specify and codify what is meant by the term. Ma tarazzo (1980) defines behavioral health as a n e w interdisciplinary field d e dicated to promoting a ph i losophy o f health that stresses individu al res pons i b il ity in the application of behavioral and b i omed i c a l sci e nce knowledge and techniques to th e ma inten a nc e of health and the prev e ntion of illne s s a nd d y s f un c tion by a variety of sel f -initiated ind i v i du al or shared activities." (p. 813. Emphasis i n or i g inal ) He proposes this fie ld as a w ay o f p r og ressi ng b ey o n d th e limits of traditional appr o ac h es to h eal th. In a discussion of the pot e nt ia l r ol e o f c oun seling psychology in public heal t h Tho rese n a nd Ea gl es ton ( 1985 ) emphasize the traditional ed u cati on al mode l th at serves as a foundation for the field This approach le n ds itself to

PAGE 15

5 health promotion in two related ways: "(1) the promotion and maintenance of good health through the development of healthy habits and lifestyle. (2) the prevention and treatment of disease through changing unhealthy behaviors, especially those that may be associated with certain chronic diseases and health problems." (Thoresen and Eagleston, 1985, p. 70). They argue for consideration of the cognitive, behavioral, and environmental factors that have an impact on the issue of health along with physiological ones. Harris and Guten (1979) point out that an avenue little explored in the area of health protective behavior is how individuals attempt to attain health as producers of this commodity rather than be consumers of others' services. This suggestion is made in a discussion of their findings in a study of how individual beliefs relate to self-identified health protective behavior. Such a view represents a challenge to the paradigm of conceptualizing health; it also argues for investigation of the personal characteristics that affect involvement in health protective or enhancing activities. A principal health protective and enhancing activity that is receiving more research attention is aerobic exercise. Regular aerobic exercise has been shown to be beneficial in a number of different human functions, including both physiological and psychological realms (Folkins & Sime, 1981; Morgan, 1981; Paffenbarger and Hyde

PAGE 16

6 (1988); Sinyor, Schwartz, Peronnet, Brisson, & Serganian, 1983; Stern & Cleary, 1981; Taylor, Sallis, & Needle, 1985). The need for exercise can be self-assessed or prescribed by a health professional who has determined an appropriate application. Exercise has been used successfully in the treatment of depression (Greist, Klein, Eischens, Faris, Gurman, & Morgan, 1979), coronary heart disease risk modification (Stern & Cleary, 1981), and obesity (Epstein & Wing, 1980). As has been observed, the problem of adherence has compromised better assessment of the efficacy of exercise in treatment regimens (Martin & Dubbert, 1982). Even moderate amounts of exercise have been shown to be beneficial. In a very large study examining the relationship varying levels of fitness to subsequent mortality, a group of researchers found a "strong, graded, and consistent inverse relationship between physical fitness and mortality in men and women" (Blair, Kohl, Paffenbarger, Clark, Cooper, & Gibbons, 1989}. In reporting their findings the authors observe that the prevalent sedentary lifestyle presents an important public health problem. Attrition From Exercise Programs The casual observer might well form the impression that most people in the population at large exercise. Exercisers are quite visible, as are advertisements for

PAGE 17

7 health clubs and other "health-oriented" services. Such an impression would be in error, however, when compared to figures describing actual exercise behavior. A recent overview of American participation in exercise found that only 20% of the population exercise intensely enough and often enough to meet current guidelines for fitness or reduce the risk of chronic disease or premature death. (Dishman, 1988). Another 40% are active to some level, but not enough to derive a fitness benefit. This shows that most Americans fail to perform a behavior with proven efficacy in health promotion and remediation of health problems. Despite the growing recognition of the value of exercise among professionals and laypersons alike, a continuing pattern is the failure of most beginners in exercise programs to persist in their efforts. This pattern is observed even in the face of urgent medical conditions for which exercise has proven an effective treatment (Carmody, Senner, Malinow, & Matarazzo, 1980; Dishman, Ickes, & Morgan, 1980; Martin & Dubbert, 1982). Martin and Dubbert (1982) summarized findings on clinical applications and promotion of exercise in behavioral medicine, contending that measuring the efficacy of exercise as a clinical treatment regimen was hampered by poor adherence rates. They maintain that this pattern introduces doubt even in areas where exercise is generally agreed to be effective, as in obesity or

PAGE 18

8 cardiovascular risk reduction. Commenting on the lack of an effective means of keeping patients identified as at-risk in a program of exercise, they state: "we believe that far too little attention has been directed toward the cognitive/psychological/social variables controlling exercise adherence." (Martin & Dubbert, 1982, p. 1013). They recommend the tailoring of individual packages to the needs and profile of the individual undertaking an exercise program. They document a need for more information about the impact of different personal variables on the success or failure of an exercise program before being able to employ such packages successfully. Attrition, or its parallel phenomenon relapse, is not a problem unique to exercise (McAlister, Farquhar, Thoresen, & Maccoby, 1976; Meichenbaum & Turk, 1987). Representative of a number of writers is the Martin et al. (1984) observation that the exercise relapse curve appears similar to the negatively accelerated curve found for the addictions; the majority of the relapses occur during the first several months, followed by continued, more gradual rates, finally leveling off at one year between 55% and 75% dropout. For exercise, relapse refers to the phenomenon of discontinuing the behavior, as compared to addiction where reference is to resuming an undesirable behavior. Dishman, Ickes, & Morgan (1980), discussing a similar observation, speculate that similar mechanisms may

PAGE 19

operate across many behavior change problems. As put in a related study: the remarkable correspondence between dropout patterns in exercise programs ... and programs of psychotherapy, as well as drug, alcohol, and smoking treatment, and hypertension control ... suggests that similar influences may operate across many health care settings. This similarity at least implicates characteristics of the participant as potential adherence predictors. (Dishman & Gettman, 1980, p. 296) 9 Activity in exercise research has begun to address the problem of attrition from exercise programs, examining the phenomenon in an attempt to identify factors affecting the successful adoption of an exercise regimen (Martin & Dubbert, 1982; Dishman, 1982; Gale, Eckhoff, Mogel, & Rodnick, 1984). In a comprehensive review of these factors Dishman (1982) organized them into areas that take into account all the variables likely to have an impact on the outcome of an exercise adoption attempt. The main areas named include 1) psychological features of the exerciser; 2) biological features of the exerciser, including health status; 3) demographic features of the exerciser; 4) situational characteristics of the exercise setting; 5) strategies/clinical interventions used to encourage adoption of exercise habits; and 6) aspects of the person-setting interface, including ways the interaction between the person and the setting affects subsequent decision-making and behavior. Such a summary is not unique to the particular challenges posed in the area of exercise behavior; similar findings are noted in an analysis of

PAGE 20

10 adherence factors affecting the outcome of treatment regimens prescribed by health care providers (Meichenbaum & Turk, 1987). Most research on exercise has addressed the social, environmental, and behavioral aspects of the exercise adoption process. Typical of research in the area of exercise adherence is the comprehensive series of studies performed by Martin et al. (1984). In these studies, a decidedly behavioral focus yielded much valuable data about the effects of social support, cognitive strategies to accompany the exercise, relapse prevention training, attendance lotteries, personalized feedback, and the structure of the goal-setting for the individual's exercise program. Using this example in light of the Dishman categories cited above, one can see the emphasis on 4) situational characteristics of the exercise setting; 5) strategies/clinical interventions used to encourage adoption of exercise habits; and 6) aspects of the person-setting interface. In an already-quoted review that presents a summary of much of the research on exercise adoption, Martin and Dubbert (1982) make the puzzling observation that one's attitude toward exercise does not appear to predict participation or later adherence to a program of exercise. Given the established links between attitudes and b e havior (Ajzen & Fishbein, 1977; Cooper & Croyle, 1984) and

PAGE 21

11 specific studies establishing the relationship in the case of other behaviors (Becker, Haefner, Kasl, Kirscht, Maiman, & Rosenstock, 1977), such an observation cannot be accepted as a given without supporting research. The research necessary to determine the presence of such a relationship has only recently commenced (Sonstroem, 1988). The Health Belief Model and Its Utility in Exercise Adoption Research As a tool for comprehensive measurement of psychological process, the Health Belief Model has shown consistency and robustness, having value for health practitioners and researchers addressing many varied kinds of behavior (Janz & Becker, 1984). It enables a comprehensive view of many beliefs that influence a person's predisposition to act in a certain way, with the exact mechanism of that influence still under study. A belief is considered the cognitive part of an attitude toward an object. A belief represents the information an individual has about an object, as compared to the attitude, which represents the individual's overall evaluation of that object. The belief links the object to a quality in the person's mind. Some have postulated a belief to have an emotional component modifying the overall response to an object or action (Sonstroem, 1982). That is, the person's feeling about the outcome of the action in question becomes part of their beliefs about

PAGE 22

that action, affecting the valence of their specific and general attitudes, beliefs, and behavioral intention. 12 The relationship between attitudes and behavior has been a well-documented phenomenon, especially when many elements of the attitude-behavior pair correspond (Ajzen & Fishbein, 1977). Theorists in the area of attitude-behavior relations note the importance of comparable levels of specificity in the attitude and the behavior (Cooper & Croyle, 1984). A general attitude wil l predict a multiple-act criterion better than a single-act criterion, while a specific attitude works conversely, predicting the single-act criterion better than the multiple-act criterion. The need, as detailed by Godin and Shepard (1986), is for appropriate measures of attitude to be employed in the particular area of exercise behavior. There has been a long-standing interest in documenting the relationship between attitudes or beliefs and behavior. The Health Belief Model has been one o f the most widely utili z ed of theories that have been formul a ted to determine the nature and impact of beliefs in the are a of health protective behavior. What started as an att em pt to identify the factors underlying he a lth behavio r s h as evolved into a w e ll-documented model w i th utility in a variety of settings. The Health Beli ef Mod e l was d e v e loped in a n a tt emp t to understand why p e ople did not pr a ct i c e h ea lth prot e ctiv e b e h a vio rs f ir s t in r egard t o asymp t oma t ic

PAGE 23

13 diseases, later in an expanding number of conditions and venues. Use of the model was later expanded to investigate factors contributing to compliance with medical regimens, (Becker, Maiman, Kirscht, Haefner, & Drachman, 1977; Becker, Drachman, & Kirscht, 1974) with recent investigations broadening its application to preventive health behaviors (Riddle, 1980; Slenker, Price, Roberts, & Jurs, 1984; O'Connell, Price, Roberts, Jurs, & McKinley, 1985). An early overview of research employing the model included a statement detailing its utility for health care providers: "by knowing which Health Belief Model components are below a level presumed necessary for behavior to occur, the health worker might be able to tailor intervention to suit the particular needs of the target group or population" (Becker, Haefner, et al., 1977, p. 30). Other investigations into the efficacy of interventions designed to affect peoples' health beliefs and thus subsequent behavior have shown them to be successful (Becker, Maiman, Kirscht, Haefner, & Drachman, 1977; Kirscht, 1974). The components of the Health Belief Model were posited to identify cognitions and perceptions making up the beliefs that early developers of the model found to have the most impact on health behaviors (Kirscht, Haefner, Kegeles, & Rosenstock, 1966). They include the

PAGE 24

14 individual's subjective estimate of their susceptibility to disease states; the perceived severity of the conditions the individual feels vulnerable to contracting or suffering; perceived benefits of the health protective behavior being researched; and the perceived barriers to performing that behavior. Added later to the Health Belief Model was a construct addressing health motivation, which was seen as a necessary part of the model when examining behaviors with a more preventive or health-enhancing nature (Becker & Maiman, 1975). Finally, having an impact within the framework of the model are cues to action which trigger the individual to act. The Health Belief Model was described by theoreticians associated with early work in its development as measuring specifically elements of two variables essential to the prediction of behavior: 1) the value placed by an individual on a particular goal (i.e. health or the avoidance of a disease); and 2) the individual's estimate of the likelihood that a given action will result in that goal (Maiman & Becker, 1974). A summation offered by the framers of the model who participated in its refinement includes all the elements: the Health Belief Model hypothesized that persons will generally not seek preventive care or health screening unless they possess minimal levels of relevant health motivation and knowledge, view themselves as potentially vulnerable and the condition as threatening, are convinced of the efficacy of intervention, and see few difficulties in undertaking the recommended action. (Beck e r, Haefner, et al., 1977, p. 29)

PAGE 25

15 Investigations employing the Health Belief Model have been successful in their aim of predicting or explaining the level of non-exercise preventive health behaviors. Two evaluations of Health Belief Model research found it to have significance in the degree of association of its principal components with almost every specific behavior under study. The appraisals covered 46 different studies that utilized the Health Belief Model. The pattern of significance held when the type of behavior being investigated was sick role behavior as well as with preventive health behavior, and for earlier as well as more recent studies (Becker, Haefner, et al., 1977; Janz & Becker, 1984). The level of significance for the components was also consistent in both retrospective (n=28) and prospective (n=l8) studies. The model appears to have potential as a useful tool for the investigation of factors shaping exercise behavior, one of the most global forms of health-related behavior. The Role of an Attitude/Beliefs Study in Exercise Adoption Research The relatively new area of research into exercise adoption has been yielding specific data about the factors that influence successful adoption and maintenance of an exercise habit, particularly from the point of view of the practitioner working with beginning exercisers. Behavior modification approaches, as detailed by Knapp (1988), Martin et al. (1984), and Franklin (1978), have shown

PAGE 26

16 effectiveness at increasing adherence rates. Such approaches now include relapse prevention strategies, integrating newly gained knowledge about habit change. The attractiveness of the exercise itself for the beginning exerciser has been shown to be an important factor (Franklin, 1978: Gale, Eckhoff, Mogel, & Rodnick, 1984), along with the approach used by the instructor in an exercise program. The term exercise adoption was intentionally used for the current study to help provide emphasis on the dynamics whereby an individual actively seeks to adopt an exercise program as a part of his/her daily behavior. The adoption process is illustrated in a stage theory of change detailed by Prochaska and DiClemente (1983). The process begins with a precontemplation stage, with movement proceeding to a contemplation stage. Actual behavior change begins in the action stage, which becomes the maintenance stage if the change is long-standing and durable. Failure to maintain the change moves the person into a relapse category. Such a schema helps to capture more of the phenomenon of behavior change than just the actual presence or absence of behavior. The thoughts and feelings of the exerciser have been relatively neglected in research on the exercise adoption process, as compared to examination of the appropriate techniques. An example of one such conceptualization of

PAGE 27

17 personal cognition/self perception is self-efficacy theory (Knapp, 1988; Bandura, 1982). While indications are that exercise self-efficacy plays a role in the subsequent outcome of an attempt to begin an exercise program, results are not conclusive (Sonstroem, 1988). Discussions of psychological characteristics having an impact on habit change outcomes usually mention this construct, but it has not been operationalized for the exercise adoption problem. The need for such socio-psychological correlates of exercise behavior has been asserted by those conducting cognitive-behavioral programs for change in exercise adherence studies (Martin et al., 1984). The personal characteristics and, occasionally, beliefs that influence decision-making and subsequent behavior have been well documented in the case of other behaviors, even other health-related behaviors, but exercise behavior remains largely unexplored in regard to beliefs. The Health Belief Model method of assessment has proven validity and reliability in determining individual beliefs, and has become established as a tool for predicting health behavior (Janz & Becker, 1984). The first applications of the model in the area of exercise behavior have proven encouraging, with limited generalizability to all forms of exercise (Riddle, 1980; Sonstroem, 1987; Slenker, Price, Roberts, & Jurs, 1984).

PAGE 28

18 Dishman (1982) states the need to "determine the effectiveness of various potential adherence strategies . Consider the relative effectiveness of manipulating abstract, conceptual beliefs" (p. 261). To date little activity has focused on the development of interventions to effect changed beliefs in those seeking to adopt exercise behavior as part of their lifestyle. The primary focus with the Health Belief Model up to now has been an individual's participation in medical or disease-prevention strategies. The current study was proposed to provide a basis for a structured intervention to increase the success rates for exercise adoption efforts. With the capability to reliably assess beliefs about health and exercise, the efficacy of such interventions may be evaluated. Aerobic Exercise and Health Beliefs Extending the Health Belief Model mode of research to the area of aerobic exercise behavior is a step indicated by other prior studies that have explored the role of health beliefs in preventive health behavior, including specific forms of exercise. The present study was designed to show that a reliable and valid mechanism of measurement of beliefs regarding exercise can be developed, including a demonstration of predictive validity of the instrument in reference to self-reported aerobic exercise adoption stages and levels of aerobic exercise.

PAGE 29

19 Individual tailoring of programs to each person seeking to begin an aerobic exercise program (whether self-motivated or prescribed) has been recommended for greater success achieving lasting change and implementing the goal of an exercise regimen (Gale, Eckhoff, Mogel, & Rodnick, 1984). Circumstances for a prescribed aerobic exercise program may be weight reduction or cardiovascular rehabilitation, while many reasons have been cited by people undertaking exercise programs voluntarily, from appearance to preparation for participation in a sport. A specific, comprehensive beliefs measure would provide for identification of personal beliefs relevant to entry into an aerobic exercise program. Martin and his fellow researchers have identified many of the program strategies and interpersonal approaches that enable the fitting of an exercise adoption approach to the needs of the person starting a program of exercise (Martin et al., 1984; Martin & Dubbert, 1982). Missing from the daily practice of health care workers who address the exercise adoption challenge is a consideration of the individual's attitudes and beliefs about the undertaking. Such an omission is, in part, due to lack of a means with which to assess a person's beliefs about health and exercise. An historical drawback in Health Belief Model research is that measures employed to assess belief

PAGE 30

20 constellations for the behavior under study have been idiosyncratic to each particular investigation. Surprisingly, no one procedure of instrumentation has emerged as a standard in any area covered by the dozens of studies utilizing the Health Belief Model. In one attempt to organize and verify the reliability and validity of constructs within the Health Belief Model, the researchers comment on the limitations imposed by the lack of such work: The absenc e of reliable and valid measures not only limits the practical utility of the theoretical formulation, but also reduces the potential for developing a reliable body of knowledge on which to design intervent i on strategies to change personal health behavior. (Jette, Cummings, Brock, Phelps, & Naessens, 1981, p. 83) No attempt has been made to develop a standard instrument to be employed as an adjunct for aerobic exercise ado p tion situations, w hether for research purposes or for use as a treatment co m ponent. App l ying th e Heal th B eli ef Model to Aerobic E xe rci se B e havior In studies w h e re interventions utilizing H e alth Beli e f Mod e l co m pon e nts have b ee n performed, s ucc essf ul ch a n g es i n t ar g e t e d preventive h ea lth b e havio rs were effec t ed ( Beck & Lund, 1 98 1; Becke r, Ma i ma n, Kirsc h t Haefner, & Drachman 1 977 ). Behavi o rs inc lu ded oral h ygie ne ca r e a nd m oth ers s u ccessf ul f ollo w -th ro u g h o n treatment for their childrens obesity The technique of b eliefs modificatio n has favorable potential as a m e thod

PAGE 31

for influencing the process of behavior change, specifically the adoption of aerobic exercise behavior. 21 No definite relationship has been established between the Health Belief Model components and aerobic exercise as a dependent variable. The purpose of the present study was to test for a link between the belief variables of the Health Belief Model and aerobic exercise, using a methodology paralleling prior, successful preventive health behavior investigations. In these earlier studies, the use of beliefs assessment instruments has provided for successful prediction of a variety of targeted preventive health behaviors (Tirrell & Hart, 1980; Chen & Tatsuoka, 1984; O'Connell, Price, Roberts, Jurs, & McKinley, 1985). Previous studies have shown the Health Belief Model to have similar potential in the successful measurement of relevant beliefs and the prediction of aerobic exercise behavior (Riddle, 1980; Slenker, Price, Roberts, & Jurs, 1984) A valid instrument for assessing beliefs in reference to aerobic exercise will provide a basis for successful employment of belief component manipulations in public health, recreation, and counseling settings, toward the end of positive improvements in the adoption of aerobic exercise behaviors. Belief manipulations can range from personalized, one-on-one discussions to mass media campaigns that involve an entire community. Such regimens have been proven effective even on a very large scale. An

PAGE 32

example of one such intervention often cited is the Stanford Five-City Project, where significant reductions in cardiovascular risk factors occurred over a thirty month period following an ongoing public health information campaign (Thoresen & Eagleston, 1985). 22 The method proposed parallels that used in successful applications of the Health Belief Model to explain and predict other preventive health behavior, as well as that used in earlier applications of the model in the e x ample of illness avoidance/sickness reduction. As has been noted, correlations of model components with target behaviors that were obtained in past investigations have been sign i ficant, wi th some variation in the deg r ee of signi f icance. I ni t ia l e x ploration of the constellation of health belie f s f or e xe rcisers has yielded encouraging results Th ese f indings h a ve provided a found a tion f or a n effort to dev el op a n instrument for determining Health Beli ef Mod e l b e l i e fs in r e fer e nc e to a e robic exe rci se (Sonst r o em, 1 9 87; S le n k er, Price, Roberts, & Jurs 1984 ; Riddl e 1 98 0) S u c h a n in s tru m ent would prov i d e th e b as i s f o r measureme nt th a t w o ul d e n a bl e a ss e ssmen t o f a n y s u bsequen t ma ni pu l a t i o n o f b e l iefs, i n t h at c h a n ge could t hen be relia bl y measured Posi n g the Health Belief Model constructs in the context of a standard definition of aerobic exercise (American College of Sports M e dici n e 1978) allows for

PAGE 33

23 corroborative research and provides a basis for comparisons to be made on the basis of a general consideration of aerobic exercise, not just one particular form. The College guidelines specify exercise quantity and quality: 1) Frequency of training 3-5 days per week; 2) Intensity of training 60%-90% of maximum heart rate reserve; 3) Duration of training 15-60 minutes of continuous aerobic activity, with duration dependent on intensity; and 4) A mode of activity that involves large muscle groups, that can be maintained continuously, and is rhythmical and aerobic in nature (American College of Sports Medicine, 1978, p. vii). The structure of the current investigation was intended to provide a meaningful foundation for future intervention studies to be attempted, as well as other prospective studies that monitor behavior change given an initial constellation of beliefs. In the following chapter, a more detailed examination of the literature in the areas of exercise and exercise adoption will be presented, including theory and practice in exercise adoption, adherence, and compliance contexts. A close inspection of factors affecting the exercise adoption process will follow, including cognitive, behavioral, and interpersonal elements. The Health Belief Model and its application in other studies will then be presented, along with investigations documenting the suitability of the Model for aerobic exercise adoption

PAGE 34

24 research and practice. Particular issues affecting the development and use of an instrument to measure beliefs about aerobic exercise will also be examined. Chapters 3 and 4 will describe the development, evaluation, and testing of the beliefs/exercise assessment instrument. In Chapter 5, specific findings and the implications of the results for future investigations utilizing Health Belief Model theory will be discussed, as well as indications for the area of exercise adoption research in general.

PAGE 35

CHAPTER 2 REVIEW OF THE LITERATURE The following review will examine the literature on exercise, exercise adoption, adherence to exercise programs and the factors affecting adherence. Also to be examined is the Health Belief Model and its use as an explanatory tool in research on peoples' health-related decision making and behavior. Finally, research relevant to the decision to apply the Health Belief Model in the exercise adoption context will be reviewed. The Benefits of Exercise The benefits of regular exercise seem almost without question; the findings, however, are not as unqualified as the casual observer might expect. Many di f ferent kinds of studies have been performed in an effort to ascertain exercise effects along a continuum ranging from very narrow, specific questions to very broad, inclusive ones. An example of the former is a study by Sohn and Micheli (1984) that investigates the effect of running on the development of a specific form of arthritis, an example of the latter is Folkins & Sime's (1981) analysis of exercise effects on me nt al health. Their review was a critical examination of the results of over 39 studies 25

PAGE 36

investigating the impact of exercise on psychological states. 26 Research studies on exercise also vary according to type of exercise, length of exercise, the method of monitoring adherence to an exercise program, and the various groups of subjects studied. The following review will focus on research that utilizes a more general, inclusive definition of aerobic exercise, that is exercise that produces an elevated heart rate for extended periods. Another factor assumed and sought as an element of the research cited is regular, versus episodic, participation in programs of exercise. Factors utilized in the selection follow from American College of Sports Medicine (1978) recommendations about the quantity and quality of exercise needed for "developing and maintaining cardiorespiratory fitness and body composition in the healthy adult" (American College of Sports Medicine, 1978, p. vii). The effects "bias" in selecting research on benefits is toward psychological results. Stern and Cleary (1981) citing Naughton (1974) list a summary of the clinically observable physiological benefits of exercise, illustrating the pattern of changes that accompany improved fitness. They include 1) significant reduction of heart rate and systolic blood pressure, at rest and at work; 2) significant increases in peak oxygen uptake; 3 ) significant decreases in myocardial

PAGE 37

27 (heart muscle) work at rest and at work; 4) changes in body composition, including reduced fat and increased muscle mass; and 5) changes in circulation to patterns observed in healthy, physically active subjects. This pattern is noted in many studies as the training effect, and various of the features are often used to test for fitness levels, as they were in the Stern and Cleary (1981) study on changes in a population of exercisers who had suffered myocardial infarct. In a chapter examining the effect of exercise on coronary heart disease (CHO) rate and risk, Paffenbarger and Hyde (1988) found significant differences in longitudinal studies of two very different populations. For a group of 3,686 San Francisco longshoremen, those in the group who had the highest levels of physical activity as part of their jobs were found to have significantly less risk of CHO than their counterparts who did not work as hard. This effect held for all age groups, and was strongest for the youngest age group. The groups were studied over a 21-year period, with the work level of 8,500 kilocalories per week used as the cutoff between lowand high-energy-output groups. In an impressive finding, the degree of increased CHO risk was greater for those who were in a low energy output group (versus a high output group) than the increased risk that accrued for being a heavy cigarette smoker (greater than one pack a

PAGE 38

day) or for having high blood pressure (greater than or equal to the mean level for the age cohort)! 28 A parallel study, involving 16,936 Harvard alumni, found similar results. Alumni engaging in strenuous activity plus a minimum of 1000 kilocalories a week of other light activities had less than half the CHO incidence of their nonathletic, sedentary classmates (Paffenbarger & Hyde, 1988). Exercise was found to be the second most influential factor affecting mortality from CHO causes, preceded by smoking and followed by hypertension. Exercise was second only to parental history of hypertension in its impact on whether the alumni developed hypertension themselves. Recent results demonstrate some of the likely mechanisms by which regular exercise may contribute to lower CHO mortality. A study investigating the impact of exercise on blood cholesterol levels found that "good" cholesterol (HOL or high density lipoprotein) began increasing while "bad" cholesterol (LOL/VLOL or low/very low density lipoprotein) began decreasing after a seven week exercise program, a fairly rapid effect of exercise. The program consisted of four sessions per week of treadmill running for twenty minutes per session to 80 % predicted maximum heart rate. Subjects ate a control le d diet and underwent frequent blood testing to assess lipoprotein and other blood element levels. A parallel

PAGE 39

29 finding was a significant decrease in postprandial levels of lipoproteins, showing a faster uptake of cholesterol (Weintraub, Rosen, Otto, Eisenberg, & Breslow, 1989). Exercise was demonstrated to have a positive effect on the coping responses to a stressor when trained and untrained subjects were compared. Heavy aerobic exercisers were assigned to the trained group, while non-exercisers comprised the untrained group. In addition to a standard "training effect", (Higher maximal oxygen uptake, lower blood pressure) subjects showed a faster adaptive biochemical response when subjected to a psychosocial stressor (Sinyor, Schwartz, Peronnet, Brisson, and Serganian, 1983). Catecholamine levels peaked earlier, and at higher levels, for the trained subjects. This effect was described by the authors as a more efficient coping arousal. Additionally, heart rate returned to normal more quickly in the trained group of subjects, reflecting a recovery from the experience of the stressor that was faster than in the untrained group. Stern and Cleary (1981), in the above-noted study on a male post-infarct population of exercisers, noted an increased work capacity for the exercisers as measured in testing on a motorized treadmill. Their program of low-level exercise consisted of two to three sessions per week for six weeks for an average of 15 minutes of exercise per session. This program produced an average ten

PAGE 40

30 percent increase in the work capacity of subjects over the course of the study. Wilfey and Kunce (1986) found the magnitude of training benefit to be greater for subjects with lower initial fitness levels, in both physiological and psychological measures. For low fitness/high stress subjects the overall changes were the greatest, as might be projected for people of the group with the most to gain from completing a two month long comprehensive exercise training program. Heart rate decreases were greatest for the low fitness/high stress group, followed in order by the low fitness/low stress group, the high fitness/high stress group, and the high fitness/low stress group. A similar pattern obtained in measuring gains on a fitness test, with the high fitness/low stress and high fitness/high stress groups exchanging ranks. The finding that persons of lower fitness gain more from training than those at higher fitness levels may seem obvious, but it also illustrates the linear nature of the physiological gains obtainable through an exercise training program. Many psychological benefits have been document ed as a product of regular exercise, with findings consistent through many studies. The positive effects can be identified with a high degree of certainty as associated with the exercise itself and not with other variables. Well-controlled studies do vary in the specific effects

PAGE 41

noted, often as a result of their initial focus or particular changes identified as the target of the investigation. 31 In a review of his own and others' work on the short and long term effects of exercise, Morgan {1981) found some consistent indications that regular exercise, performed for a minimum of twelve weeks, results in lower scores on anxiety and depression scales, while measures of well-being increased. This pattern held across different measures and different periods of exercise beyond the twelve week minimum. Morgan concludes that exercise is clearly associated with changes in the three areas, wi th an additional observation that causality is harder to prove. A related study found reduced levels of anxiety in physically trained subjects following exposure to a psychosocial stressor when compared to untrained subjects {Sinyor, Schwartz, Peronnet, Brisson, & Serganian, 1983). Aerobic points earned and physiological response measures were used to differentiate trained from untrained subjects. The anxiety measure was the Spielberger State Anxiety Inventory, administered prior to and following different stressor tasks. Following the tasks, trained subjects had significantly lower SSAI scores, showing faster recovery from induced anxiety.

PAGE 42

32 One of the largest, most complete studies was conducted on a population of 651 subjects that had experienced a myocardial infarct less than three years prior to their involvement in the research. Stern and Cleary (1981) investigated the psychosocial effects of a low level exercise program on participants, finding a decrease in the percentage of depressed subjects and in depression scores as measured by the MMPI depression subscale. On a group of twelve clinical symptoms, persisters in the exercise program improved on ten of the twelve as rated by spouses. Differences were found to be significant between the initial and followup evaluations, which were administered six weeks apart. Although the authors see the exercise group as benefiting more than a non-exercise group, they qualify the finding as not being wholly attributable to the exercise component. One of the most thorough evaluations of the impact of exercise on mental health was a review of research conducted by Taylor, Sallis, and Needle (1985). They f ound positive effects in many of the studies they examin e d where exercise w a s used as a treatment modality fo r clinical cond i tions, many with a significant d e gr ee o f ch a ng e Mild to mod e rate depr ess ion w as cited a s s h owi n g the most promise for change, with m a ny studi es docu me nt i n g positive e ffe cts. Th e h y poth esize d m ec h a ni s m s o f c h a n ge i n ex e r cis e p r o grams u sed to tr ea t d e p ressi on, w h ile n ot

PAGE 43

certain, were increased neurotransmission of catecholamines, endogenous opiates, or both; diversion, social reinforcement, and improved self-efficacy. 33 Taylor et al. (1985) found both state and trait anxiety to be effectively reduced by varying levels of exercise, with a greater effect noted in subjects whose initial anxiety levels were higher. Effects were more conclusive for state anxiety than for trait anxiety, with both types showing a positive benefit in most studies examined. The primary effective mechanisms here were described as diversion, social reinforcement, experience of mastery and an improved response to stress. Taylor et al. (1985) also noted that improved self-confidence, cognitive function, and sense of well-being were associated with exercise and physical activity. Perhaps the most complete of the major studies reviewing the impact of exercise on mental health is the massive compilation by Folkins and Sime (1981). Their critical review is often cited in other studies as a significant overall finding of the beneficial effects of exercise on human psychological function. Often omitted in a citation of their work is the observation that most of the research on exercise effects that they reviewed were not true experimental designs. They note a preponderance of pre-experimental or quasi-experimental designs, a drawback that is particularly limiting in longitudinal

PAGE 44

studies. Even within the true experimental research, however, a pattern of positive effects is cited. 34 Folkins and Sime (1981) found fitness to be related to the following aspects of mental health: 1) improved work patterns, including reduced absenteeism, fewer errors, and improved output; 2) improvement in mood state, particularly for distressed or initially unfit individuals; 3) improved self-concept and body image; 4) reduced levels of depression where exercise is used for treatment of depression; and 5) improved cognitive function. The authors caution that in some instances the results obtained in a study were not significant; overall they were convinced of the utility of exercise in producing improved psychological states. They single out self-concept and depression as areas where the results are less equivocal. In a carefully controlled study, Bahrke and Morgan (1978) compared exercise, meditation, and quiet rest to determine their relative efficacy at inducing relaxation for experimental subjects. Each "treatment" was employed for twenty minute periods, with the exercise condition being treadmill walking to seventy percent of age-specific maximal heart rate. Relaxation was measured u sing a state anxiety scale that was employed preceding, immediately following and ten minutes following the experimental

PAGE 45

35 conditions. All three treatments were effective in reducing anxiety, causing decrements that differed significantly from the initial measurement. Interestingly, the three treatments did not differ significantly from each other in their effects. The authors found the findings striking in that the mechanism of exercise (physiological arousal) is so different from that of either quiet rest or meditation in the quality of its effects. They also found the exercise effects corroborated earlier investigations into the impact of exercise on anxiety levels. It must be noted that "lack of anxiety" may not be a universal criterion for relaxation; given the arousal aftereffects on physiological measures it was the choice for this study. Morgan and O'Conner (1988) have summarized some limitations in the studies which document the positive effects of various forms of exercise on mental health. They note a unanimity among researchers that the higher the level of fitness, the more desirable the level of mental health. Next is noted the elusive nature of evidence allowing clear statements of causality. Finally, the lack of controlled studies with true experimental designs, as noted above in the review by Folkins and Sime (1981), precludes clear statements about cause and effect. This point holds even in longitudinal studies that would

PAGE 46

support the hypothesis that vigorous exercise leads to beneficial mental health effects. 36 Citing a NIMH workshop that used a strictly limited criterion for selection of statements about exercise effects, Morgan and O'Conner (1988), list findings about the influence of exercise on mental health. These findings use carefully stated points to summarize what can be stated about the positive effects to be gained from exercise/fitness training: 1. Physical fitness is positively associated with mental health and with well-being. 2. Exercise is associated with the reduction of stress emotions such as state anxiety. 3. Anxiety and depression are common symptoms of failure to cope with mental stress, and exercise has been associated with a decreased level of mild to moderate depression and anxiety. 4. Long-term exercise is usually associated with reductions in traits such as neuroticism and anxiety. 5. Severe depression usually requires professional treatment, which may include medication, electroconvulsive therapy, and/or psychotherapy, with exercise as an adjunct. 6. Appropriate exercise results in reductions in various stress indices such as neuromuscular tension, resting heart rate, and some stress hormones. 7. Current clinical opinion holds that exercise has beneficial emotional effects across all ages in both sexes. 8. Physically healthy people who require psychotropic medication may safely exercise when exercise and medications are titrated under close medical supervision. (p. 95) The exercise effect is thus shown to be dependable, applicable to a wide variety of persons, and amenable to use at the same time as other forms of treatment. Peoples' perceptions about the benefits of a regular program of exercise are equally as important as the

PAGE 47

37 documented benefits. As will be seen in the discussion of the components of the Health Belief Model, perceptions are an important aspect of beliefs. In a study in which people were polled about various forms of behavior they performed to "protect, promote, or maintain ... health", Harris and Guten, (1979) found that exercise behavior was cited the third most often--behind nutrition and sleep--as one of the "three most important things that you do to protect your health." (p. 21). The important distinction here was that the researchers recorded what behaviors the respondents considered important for health protection, versus a list of behaviors that clinicians perform or would recommend. The level of utilization of health protective behaviors held even when health condition of the respondent varied from good to poor, with the emphasis shifting from self-initiated lifestyle choices like diet and exercise to utilization of the health care system. Exercise Adoption The phenomenon of exercise adoption may be patterned after an area of research that provides a basis for measurement and interpreting observations. Prochaska and DiClemente (1983) have developed a theory of behavior change which describes a person's movement through a series of psychological/behavioral stages, resulting in several different outcomes. In their work developing this stage theory, Prochaska and DiClemente (1983) have focused

PAGE 48

38 primarily on the cessation of undesirable behaviors, such as smoking cigarettes. As specified in their schema, a person's initial step in the change process is from an immotive or precontemplation stage to a contemplation stage. Using cognitive/affective reevaluation of their situation and behavior, the person is compelled to move to an action stage, changing actual behavior. Should the target behavior change be long term and durable, with accompanying cognitive and behavioral changes to support the target behavior, the stage entered into is that of maintenance, where the behavior is ongoing and maintained at a stable level. If behavior drops below the minimum specified level of change, the person drops into the relapse category, wherein the attempted behavior change is abandoned for a period of time. The period may be brief or extended, depending on the determination and resources of the changer toward renewing the effort at change. As will be detailed below, the relapse outcome is fairly common in behavior change projects. Employing the Prochaska and DiClemente (1983) schema in the context of exercise underscores the n eed for differentiation of adoption from adherence, which, as generally employed by exercise researchers, includ es only the maintenance stage of the process. This stance excludes other important and universal features of human a tt empts to change behavior that are noted in a broader view of

PAGE 49

39 change. The stage process as used in early studies applies whether the change is self-initiated or prescribed by other parties. At present the terms adoption and adherence are often used interchangeably to refer to success or failure to adhere to a program of exercise begun for any reason, with many of the cognitive factors and much personal history not taken into account. Dishman (1988) notes the confusions inherent in the current terms being used in behavioral science to describe the exercise adoption process. He observes that compliance and adherence are often used to describe the same phenomena, yet are not generally recognized in catalogues of psychological literature by definitions that conform to their current utilization. He proposes employing the term adherence as a general descriptor in the area of exercise, to include the area of compliance. This comes closer to meeting current needs and recognizing the entirety of the exercise adoption process, yet the early stages of a self-initiated change can be omitted by this term as well. Too, other researchers explicitly link the terms compliance and adherence (e.g. Oldridge, 1981), serving their own needs in investigating exercise treatment programs, but excluding important population groups of exercisers and parts of their experience in entering into an exercise program.

PAGE 50

40 The research on compliance to exercise programs addresses situations that are primarily prescribed behaviors in response to a stimulus situation like obesity or high blood pressure. The focus of compliance research has been on practitioner behavior and characteristics of the setting, with the interaction between client and health care provider the main interest (Becker & Maiman, 1980; Janz & Becker, 1984). Research in compliance assuredly does include individual characteristics, providing the overlap and confusion of terms alluded to above. Examination of the research findings shows by illustration some of the phenomena to be anticipated in research on exercise adoption. A finding that has results useful for the issue of self-change of habits is the comprehensive study carried out by Perri and Richards (1980). The study compared people who had carried out successful self-change programs with those who had attempted unsuccessful ones, in an attempt to determine what personal characteristics or strategies differentiated the two outcom es. Important indicators emerged for those undertaking self-change programs or working with those who do. Successful self-changers appear to be intuitively skilled at behavioral technique, employing self-reinforcement and persisting far longer in their attempts than did unsuccessful changers. Of note for the present study was the recommendation that persons

PAGE 51

41 undertaking self-change programs insure that their program is comprehensive in nature and should involve a number of individually tailored procedures. Some research (Sonstroem, 1987; Ward & Morgan, 1984) has been performed using the stage schema applied to exercise. Ward and Morgan appear to be the first to apply the stage theory of change to exercise, finding that adherence at 10, 20, and 32 weeks in an exercise program was predicted by different sets of variables. Using the Theory of Reasoned Action (Ajzen & Fishbein, 1977) as a basis for a beliefs investigation, Sonstroem related the belief constellation to a reduced number of the Prochaska and DiClemente stages. He termed the process exercise adoption, although without providing a rationale for the use of that term versus the others available. The main body of work cited in his review of the theoretical bases for his research, he notes, does include subjects who were seeking to change on their own. As the term "exercise adoption" is rarely used, the more widely held "adherence" term will apply through most of this review. Factors in Adherence to Exercise Exercise adherence is examined from many different perspectives, with each providing information to compose a picture of someone in the exercise adoption process. Some studies examine the factors associated with dropout, where others test for personal characteristics associated with a

PAGE 52

42 successful, continued exercise program. Perhaps one of the most complete recent studies on exercise adherence was the collection of researches reported by Martin et al. (1984). The series was designed in an attempt to isolate factors that affect the outcome of exercise adoption attempts by sedentary adults. The investigators' main interest was in determining the efficacy of various cognitive-behavioral procedures in support of exercise programs begun by people attending a community exercise program. Dividing 143 subjects into six treatment conditions allowed the researchers to test for the effectiveness of different elements of behavioral techniques. Each treatment group that began a running program received differing combinations of cognitive-behavioral strategies for the three months of the program, apart from a common structure utilized by all groups. For each group an "ideal" strategy was included, with the ideal based on the past observ ed efficacy of each element. In Group 1, the ideal was personalized feedback, distance goals (for running), and fixed goals. Each group provided varying combinations, allowing the researchers to compare group and personalized feedback; time and distance goals; flexible and fixed goals; distal (long range) and proximal (short range) goals; and associative versus dissociative cognitive strategies. Analyses of variance and t-tests were used to test for significant differences between conditions within

PAGE 53

43 groups. Many of the variables investigated proved a positive influence on the ultimate adherence of the subjects beginning exercise programs. In particular, social support, personal feedback/praise, flexibility in goal setting, and distraction-based cognitive strategies proved most effective in increasing rates of adherence. No advantage was seen for time versus distance goals. Impressively, overall adherence at the end of the three months, for all subjects in the study, varied from 80-85 percent. This was considerably better than the 55 percent adherence usually observed for this time period in an exercise adoption group program (Martin et al., 1984). Of interest here is the fact that 49% of the subjects in the study had previously attempted an exercise program. No differentiation of these "relapsers" from the subjects beginning exercise for the first time was attempted. Martin et al. (1984) also found class attendance and self-report of exercise participation to correlate significantly with "post" measures of fitness. Another study illustrating recent methods used to scrutinize exercise adherence is one by Dishman and Gettman (1980) that examines many different psychobiologic influences on ultimate adherence. The study used physiological measures such as weight and percent body fat along with psychological measures designed to measure attitudes toward and valuing of exercise. A prospective

PAGE 54

44 study with 66 subjects, it used subjects that were beginning exercise programs for the first time, combining healthy exercisers with coronary heart disease rehabilitation subjects. The outcome of the entry into an exercise program was labeled as either adherence or dropout, with anyone not participating at the end of the 20 week period of the study labeled a dropout. Using stepwise multiple regression, three variables of the ones measured were found to significantly enhance the prediction equation for adherence: 1) body fat percentage; 2) a measure of self-motivation; and 3) body weight. Taken together, these three variables accounted for 45 percent of the overall variance and provided for predicting adherence/dropout category in 79 percent of all cases examined. The authors chose only one other independent me a sure in addition to their own on self-motivation, that of the health locus of control (Wallston, Wallston, Kaplan, & Maides, 1976). The other sc a les w ere subsc a les of the Attitude To w ard Physic a l A ctivity scale o r the Physical Estimation and Attraction sca le Th e a bility o f these subsc a les to contribut e to a pre d ic t i on o f s ub se quent exercis e st a tus m a y h a v e b ee n compr o mised b y th eir a pplic a tion within this s tu dy as stand alone variables In a study examin in g th e lon gt erm r ates of adherence for a population of male and female vic t ims o f myocardial infarction, the investigators fou n d only 18 7% remained at

PAGE 55

45 the end of a forty month period (Carmody, Senner, Malinow, & Ma t arazzo, 1980). Two hundred and six subjects were followed up at four month intervals to see which remained active in a prescribed exercise program. In the initial four month period dropout was 30%, with the subsequent rate of dropout reduced in each followup after the initial period. This pattern duplicates the typical finding for relapse/dropout in habit-change studies (e.g. Dishman, 1988), with the dropout rate somewhat greater than that of a general population. studies of this duration are rare, so comparison is difficult. This is nonetheless a surprising finding, given the explicit urgency attached to the successful adoption of the exercise prescription for this group. Summarizing findings in a chapter reviewing studies on adherence to post-myocardial infarction exercise regimens, Oldridge (1981) identified some common characteristics that tended to be associated with dropout. The research reviewed focused primarily on the characteristics of the enrollees, omitting formal measures of their psychological or physical states. A profile constructed using characteristics that recurred across studies shows that a dropout is likely to be a blue collar worker, to be inactive physically in his leisure time, to be a smoker, and to have had more than one previous myocardial infarction. The mechanisms whereby these

PAGE 56

46 characteristics contributed to the outcome of dropping out were described by Oldridge as being unclear. Some of the subjects' negative evaluations of the programs in which they were participating included statements like "difficult to perform," "dislike physical training," or "aversion to the hospital". Franklin (1978) investigated the factors affecting exercise adherence in fitness programs for both healthy adults and patients in rehabilitation programs. Although the analysis focused primarily on individual physical characteristics, fitness level, and behavioral regimens, motivation was mentioned as a main personal factor to be addressed. Franklin's suggested solutions to motivational deficits focused on the structure and form of the exercise itself rather than directly on psychological factors. Program features associated with dropout were lack of encouragement for clients, client injuries incurred while participating, individual participation versus group programs, inadequate instruction, and absence of regular updating of clients on progress made. Other client features cited as factors were presence of support from spouse or peers and a convenient tim e schedule for involvement in exercise. A prospective study that followed 106 subjects for a total of six months was conducted by Gale, Eckhoff, Mogel, and Rodnick (1984) to determine factors that could be associated with exercise adherence. Many measures were

PAGE 57

47 taken of physiological characteristics, self-motivation, demographic descriptors, health related behaviors, and previous involvement in exercise. Subjects were healthy adult men and women; 18% dropped out before attending 10% of the classes, 40% attended between 10 and 40% of the classes, and 42% attended more than half the classes. Factors were examined for correlation to adherence. Findings were disappointing given the array of related factors that were examined. Only one of the factors (marital status) was significantly correlated with adherence to the program of exercise, and the two factors with the highest predictive value were the number of years in present occupation and number of years at present address. Indicators of less stability of overall life were associated with early dropout; features such as being single or being only a couple of years at present address or job. The other variables that distinguished between categories were self-motivation and number of children. Discussing the puzzling results, the authors speculate that the dropouts in their study may have continued to exercise away from the program, or that they may have not found the program to be engaging or challenging enough. Investigations attempting to identify specific personal factors affecting adherence to exercise sometimes yield results in unexpected directions. An example is research cited above by Dishman, Ickes, and Morgan, (1980)

PAGE 58

48 to verify the concept of self-motivation as an explanatory mechanism for differential outcomes for people beginning exercise regimens. Their findings found their construct of self-motivation to be significantly associated with adherence to an exercise program; it also found the most important factor affecting adherence to be percentage of body fat. The higher the percentage of fat (and the greater the weight) the more likely the person was to drop out of the exercise program. Although self-motivation was one of the three factors that reached significance in distinguishing between adherers and dropouts, this finding illustrates the way in which many different--and sometimes unexpectedly prominent--biopsychosocial variables affect behavior. As an aside in surveying factors affecting adherence/compliance, Dishman (1982) notes the importance of taking into account those who may drop out of the formal exercise program under study, but continue to exercise on their own or in another program. Dishman also cautions against losing research information by dichotomizing the population into adherers and dropouts, to the exclusion of the large class that does succ eed in participating in an alternative pattern of exercise. In the above pair of studies Dishman, Ickes, and Morgan, (1980) and Dishman and Gettman (1980) advanced the concept of self-motivation as an explanation for differential outcomes in people beginning exercise

PAGE 59

49 regimens. They dismiss the role of attitudes or beliefs, positing a weak or absent relationship. Godin and Shepard (1986) have responded with the observation that the problem may lie with the method. They reply to Dishman and Gettman (1980) with the assertion that the need is for theory-driven research to test models, versus the "shotgun" approach of associating variables with outcomes and claiming some degree of relationship. For them the questions to ask are "when" and "how": when is there a relationship between attitude and behavior, and how is the effect of an attitude on behavior mediated? Sonstroem (1982), acknowledges the utility of self motivation, but notes that its classification as a trait places it outside the area targeted by research into modifiable psychological factors. As will be seen below, many theorists working with the Health Belief Model believe motivation is determined by the constellation of beliefs held by persons in relation to certain pivotal areas of their lives. Powell (1988) summarizes factors affecting exercise adoption in table form, collecting together potential "determinants of physical activity" as an illustration of the scope of the problem facing those seeking to identify the specific factors having an impact on adherence to exercise. The determinants come from a compilation of past research done to identify the most important and most

PAGE 60

salient influences on an exercise adoption effort. Although research results in the exercise adoption area are far from conclusive, the list of factors is instructive: 50 Personal characteristics--Past program participation; past extraprogram activity; school athletics, 1 sport; school athletics, >1 sport; blue collar occupation; smoking; overweight; high risk of coronary heart disease; type A behavior; health, exercise knowledge; attitudes; enjoyment of activity; perceived health; mood disturbance; education; age; expect personal health benefit; self-efficacy for exercise; intention to adhere; perceived physical competence; self-motivation; evaluating costs and benefit; behavioral skills. Environmental characteristics--Spouse support; perceived available time; access to facilities; disruptions in routine; social reinforcement (staff, exercise partner); family influence; peer influence; physical influences; cost; medical screening; climate; incentives. Activity characteristics--Activity intensity; perceived discomfort. (Powell, 1988, p. 27) Powell's list bears some striking similarities to a similar group of factors identified as having an impact on treatment adherence behavior (Meichenbaum & Turk, 1987). Many of the above findings on adherence to exercise may also be found in their assemblage of factors affecting compliance with treatment regimens. Powell's list is a good tool for guiding research on exercise adoption, in that it reminds the researcher of potential limitations to the conclusions drawn from research on any one variabl e Given this caveat, another finding that may be noted is the number of items on the list that will have

PAGE 61

representation in some form in a Health Belief Model investigation targeted on exercise behavior. The Health Belief Model and Its Role in Behavioral Research 51 The Health Belief Model was developed in the early 1950s in an attempt to understand individual differences in the practice of preventive health behaviors, first in regard to asymptomatic diseases, later in an expanding number of disease states, venues, and situations. The roots of the Health Belief Model lie in social psychology, specifically the work of Lewin {1935). Rosenstock {1974a) identifies the main orientation of the early researchers as phenomenological, centered on the world of the perceiver as opposed to the actual physical world. It was this foundation that led to the formulation of a model detailing different personal "fields" that were believed to affect subsequent motivation and thus behavior. In an explication of the function of personal fields, Rosenstock {1974a) puts the origin of motivation as toward positive fields and away from negative ones. Persons operating within these fields would thus move away from those areas of their lives that were perceived as negative and toward those parts that were valued as positive. The components of the model represent a conceptualization of the fields likely to be identified as affecting decisions about behavior having an impact on health.

PAGE 62

52 In one of the earlier theoretical writings detailing the roots of the Health Belief Model, Suchman (1970) elaborated on the role of perception and interpretation in the etiology of illness. He notes, "People do not perceive the world as it actually is but as they have become accustomed to perceive it" (p. 107). In his conceptualization of the motivation to change behavior, the facts matter less than one's perceptions in making choices. Three levels of influence guide the perceptual process in Suchman's view of individual and social behavior: the anthropological/cultural, the sociological/group, and the psychological/individual. Each level shapes the resultant attitudes and beliefs, thus the behavior. "Motivation to change one's health practices depends, to a large extent, upon the individual's feelings of personal vulnerability and the seriousness with which he views the health hazard." (Suchman, 1970, p. 109-110). It is thus possible to see the scope of the challenge if one thinks in terms of behavior change interventions, as the focus is not just factual information but the particular beliefs th e individual has about the world. Becker, Maiman, Kirscht, Haefner, & Drachman (1977) argue that the Health Belief Model is based on a "value-expect ancy" approach, wherein behavior is predictable if one knows an individual's valuation of a n

PAGE 63

53 outcome and their expectation that a specific action will result in that outcome. Characteristics the authors deem relevant to the formulation of a model utilizing health beliefs include 1) the motivation to avoid illness or to get well; 2) the amount of desire for a specific level of health; 3) the belief that specific actions will prevent or moderate illness or disease conditions. The alignment of components detailed on pages 13 and 14 of Chapter 1 is the grouping commonly studied in Health Belie f Model research, and were devised to assess the presence and strength of the above characteristics. In Ajzen and Fishbein's theory of reasoned action, (Ajzen & Fishbein, 1977) beliefs are considered the cognitive component of the attitude. Kirscht (1974) defines a belief as "any proposition or hypothesis held by a person, relating any two or more psychological elements or objects" (p. 456), while attitudes are posited as collections of beliefs in which there is an evaluative component. Highlighted more within Health Belief Model theory is the evaluative, positive-negative association that the individual has at the belief level. A parallel can be discerned between the process described in the Health Belief Model and attitude-behavior relations, with the outcome, behavior, dependent on collections of attitudes, in turn made up of beliefs. The attitude-behavior link has received much discussion in social psychology, with lack of consensus

PAGE 64

54 about the role of attitudes in determining behavior (Cooper & Croyle, 1984). Generally accepted are the assertions by Ajzen and Fishbein, (1977) that the specificity of correspondence between attitudes and behavior has a lot to do with the degree of the relationship between the two. That is, a general attitude will predict a multiple-act criterion better than a single-act one, while a specific attitude will predict a single-act criterion better than a multiple-act one. An illustration is the comparison between a general attitude toward health with its resultant health-protective behaviors and a specific attitude about tooth-brushing and subsequent brushing behavior. Recent findings bear out these assertions, with a study using a measure of general attitudes and multiple acts showing a high degree of correspondence between the two (Turk, Rudy, & Salovey, 1984). Rosenstock (1974b) documents the first application of the Health Belief Model as a study by Hochbaum in 1952 that attempted to identify factors underlying the decision to obtain a chest X-ray for the detection of tuberculo sis This study assessed specific beliefs about the disease of tuberculosis and related them to the decision to seek an X-ray as a mechanism for detection of the disease. Two specific beliefs were measured, one detailing the respondent's belief that they were susceptible to

PAGE 65

55 tuberculosis, the other their belief in the overall benefits of early detection. Those having both beliefs were four times as likely to have received a chest X-ray during a specified period as those who had neither belief. These findings spurred much subsequent research and the delineation of a full model for the organization and identification of individual beliefs. It was not until the mid-seventies that the structure of the Health Belief Model had stabilized into the form that it has retained to the present. By then it had gained currency as a tool for explaining health behavior, with a review of applications of the model finding a durable pattern to the results of initial investigations (Becker & Maiman, 1975). Examining the testing of the Health Belief Model components, the authors found that the relevant beliefs reliably correlated with beh a vior. The perceived level of susceptibility was found to correlate positively with the performance of health behavior, which is defined as any action taken by an individual who perceives himself to be healthy in order to prevent the occurrence of disease or detect it in and asymptomatic stage (Kasl & Cobb, 1966). Behaviors ranged from cancer screening to obtaining immunizations to dental visits. The perceived severity of the conditions to which the individual feels vulnerable was found to have a weak relationship in early reviews of research (Kirscht, 197 4 ; Becker & Maiman, 1975), with some studies finding that a

PAGE 66

56 higher degree of perceived severity predicts participation in dental care, responding to disease symptoms, and working to prevent accidents. No significant correlation of perceived severity with target behaviors was found in many of the other studies, but the authors of the main review (Becker & Maiman, 1975) saw the component as contributing valuable information to the overall predictive value of the model. The elements of the model found to have the greatest correlation with behavior in these early reviews were the perceived benefits of the health protective behavior being researched and the perceived barriers to or costs of performing that behavior. For the benefits construct, positive, statistically significant correlations were usually found between perceived efficacy of a preventive health action and the performance of that action, including obtaining immunizations, getting screening for cancer and tuberculosis, and seeking prophylactic dental visits (Becker & Maiman, 1975). Similarly, barriers/costs were found to be inversely associated with involvement in preventive behaviors. Barriers identified as likely to interfere with carrying out behaviors were monetary cost, complexity of the regimen, side effects, degree of change needed, and accessibility of facilities. The health motivation component of the Health Belief Model was

PAGE 67

57 proposed in 1975 (Becker & Maiman, 1975); results of later investigations will be reported below. A survey of research utilizing the Health Belief Model found sufficient evidence to uphold the components as distinct, separate areas of beliefs and as having significant correlation to many differing areas of behavior (Becker, Haefner, et al., 1977). This survey systematically examined each study for evidence of significant associations between qualities assessed by the components and subsequent behavior in both the preventive and sick-role realms. Both the preventive behavior studies and the sick-role studies used prospective and retrospective designs. A total of twelve preventive behavior studies were evaluated. The behaviors being examined included seeking dental visits, getting Pap tests, getting polio vaccinations, getting examinations for a number of different potentially harmful conditions, and getting screened for Tay-Sachs disease. The authors noted the number of studies that reported results for each component of the model, allowing construction of a "significance ratio". This ratio compared the number of times a significant correlation of beliefs with behavior was found with the number of times it was tested, with significance set at the .05 level. For perceived susceptibility, the ratio was 9 out of 10; perceived severity, 6 out of 8.

PAGE 68

With the constructs of perceived benefits and barriers, the ratios were 6 out of 9 and 3 out of 4, respectively. 58 Seven sick-role studies were charted in the same manner, showing that each reported result achieved significance for the four components. The behavior focus in these studies was primarily penicillin prophylaxis or regimens recommended as a result of findings of disease, with one study investigating use of followup care after a negative health finding. Perceived susceptibility was examined in three of the studies, with perceived severity targeted in four. For perceived benefits the number was four, and perceived barriers, one (Becker, Haefner, et al., 1977). A research team performed a study assessing Health Belief Model component beliefs specific to compliance issues, in an attempt to explain the different responses of low income mothers of children with otitis media (Becker, Drachman, & Kirscht, 1974) Their work was successful in differentiating compliers with medical regimens from non-compliers, as measured by a complex behavioral protocol. The study was intended as an initial step in a program to improve compliance through changing health beliefs. While the composite model was found to be a better predictor, their conclusions stated the aim of such research in behavioral change "Thus, by knowing which model components are below a level presumed necessary for

PAGE 69

59 compliance, the health worker may be able to tailor intervention to suit the needs of the individual." (Becker, Drachman, & Kirscht, 1974, p. 215). Although the focus in the current study is on self-initiated changes as opposed to compliance, the mechanism of action of beliefs upon behavior was expected to be the same. Cues to action, a belief construct that was part of the original formulation of the Health Belief Model, has proven to be difficult to research. Cues are hypothesized as stimuli or events that trigger the individual to act once the other model components have provided the motivation and direction of the action. Cues may be internal, such as perceptions of bodily state, or external, encompassing many different kinds of messages from media or significant others. Health Belief Model researchers had early general agreement about their importance to the outcome of a behavioral intention. After some attempts to identify salient cues in investigations of Health Belief Model factors, the conclusion reached by one of the originators of the component was that most settings preclude obtaining an adequate measure of the contribution of this variable to behavioral outcomes (Rosenstock, 1974a). Without some way of recording the cues as they occur in the life of each subject, (e.g. chest pain, or a public health poster about the need to seek vaccinations) too much is lost when asking for recall of relevant, possibly related cues. This

PAGE 70

60 construct quickly faded from the Health Belief Model research method. With the development of new tools and methods for adequate measurement of cues in future researches, the variable may well be revived as a relevant part of the Health Belief Model. When Health Belief Model components were compared to other explanatory models of health behavior, many of the common elements among the models turned out be wholly or in greater part contained within Health Belief Model factors (Cummings, Becker, & Maile, 1980). The six subsuming factors found to be organizing elements for the theories examined in the study were: 1) accessibility to health care, 2) evaluation of health care, 3) perception of symptoms and threat of disease, 4) social network characteristics, 5) knowledge about disease, and 6) demographic characteristics. Thus the Health Belief Model has links with the breadth of health behavior research, with the only major omission of the model being a part that explicitly addresses social network characteristics. In that category, it appears that descriptors of such characteristics would fit in the benefits component of the model. studies incorporating the Health Belief Model have proliferated following the initial evaluative research of the seventies (e.g. Cummings, Jette, & Rosenstock, 1978) with many different applications developed and

PAGE 71

61 investigated. Although the methods used became more sophisticated, the tendency for each researcher to use instruments specific to their own studies continued. This practice resulted in refinement of statistical technique to be applied in this kind of research, but made comparative conclusions difficult because of the differences between instruments or constructs. This trend has been bemoaned by researchers centrally involved with the Health Belief Model, (Janz & Becker, 1984) but the pattern continues even in the most recent research. A major shift has occurred in the area of application of the Health Belief Model, toward investigating preventive behaviors and with a corresponding lessening of concern with sick-role behaviors. The first use of the methodology in connection with exercise was in 1980 (Riddle, 1980), with most intervention studies using the Health Belief Model having been performed primarily in the past decade. A close examination of them follows. Illustrating the durability and continued utility of the Health Belief Model methodology is a study by Chen and Tatsuoka (1984) that measured the impact of beliefs on preventive dental behavior. Preventive dental behavior includes dental visits, brushing, and flossing of teeth. Patterned after early, similar research, the study used canonical correlation, a measure of association especially appropriate for evaluating the degree of contribution the Health Belief Model components make in predicting

PAGE 72

62 different preventive dental behaviors. Conducted using 685 married Caucasian women as subjects, the study assessed beliefs and behavior simultaneously in a cross-sectional design, using a questionnaire format. Overall success or failure to engage in preventive dental health behavior was significantly related to health beliefs (rc=0.436, P < 0.001). Among the separate Health Belief Model components, the ones having the greatest predictive value with preventive dental behavior, as measured by structure coefficients, were perception of barriers and perception of benefits. Another element highly predictive of behavior was a construct examined in the study by Chen and Tatsuo k a (1984), perception of salience of beliefs. Much lower coefficients were obtained for perception of susceptibil i ty and perception of severity. The stud y by Chen and Tatsuoka (1984) is representat i ve o f the method and general findings in modern Health Belie f Model research. The perceived b a r ri ers c ompon en t app ea rs to be the construct o f th e Hea lth Bel i e f Mod el holding the strongest as soci a tion wi th prevent i ve b e h a vior, a s measur e d by f indings o f s igni ficance i n Hea lt h B e li ef Mod e l i n vest ig a tio ns (J a n z & Becker 1984 ) General findi n gs d iffer fr o m t he de nt al s t udy in t hat perceived suscepti b i l i t y was t he seco n d str o nges t association overall wi t h perceived benefits third

PAGE 73

63 One of the limitations of the Health Belief Model has been that a causal relationship between beliefs and subsequent behavior has been difficult to establish (Cummings, Becker, & Maile 1980). This difficulty arises from the need to apply the theory using knowledge about the population, about the setting, and about the specific behavior under investigation. Isolating the relationship of a particular personal feature such as beliefs to a behavioral outcome is made difficult by natural limits on the amount of information obtainable about person-environment interactions. Thus, statements about the variables examined in Health Belief Model investigations have typically been couched in terms of correlations and coefficients, with some few studies able to draw limited conclusions about causality based on controlled interventions. The most comprehensive assessment of Health Belief Model research that has been performed since the model was organized as a foundation for investigating health behavior was that of Janz and Becker (1984). The authors note many empirical findings supporting the ability of all the model components to achieve significance as explanatory and predictive mechanisms. Even the component of the model found to be suspect in its utility in earlier reviews, that of perceived severity, was upheld as a valid and reliable measure throughout its many contexts of

PAGE 74

64 application. As a model, the Health Belief Model has shown more activity and corroboration than any other model addressing health-related behavior. Within this framework, assessing the role of beliefs in mediating or determining health behavior has been a primarily retrospective, cross-sectional undertaking. Researchers' limited resources continue to hinder studies thorough enough to enable firmly ruling out competing or confounding explanations with some of the model components, but a solid base for ongoing research and applications within the health professions has been established. Longitudinal studies that would allow firmer statements about causality are costly, with few occurring in such beliefs research. Prospective studies that have been performed have shown levels of significance comparable to retrospective studies, showing promise for this method of research on the Health Belief Model (Janz & Becker, 1984). Experimental studies based on the Health Belief Model have shown the impact of health beliefs on behavior to be a useful, potent phenomenon in behavior change manipulations. Interventions using Health Belief Model components as a basis for determining the degree of change in beliefs have met with success. Using beliefs change as the mechanism to change certain target health behaviors,

PAGE 75

65 researchers have documented subsequent differences in the level of the target behavior. In a study investigating the effect of health threat communications on preventive dental behavior, Beck and Lund (1981), assigned subjects to four groups. Each group got an educational slide show about peridontal gum disease with a stated level of susceptibility to the disease (high or low) and the alleged severity of the disease if it were to occur (again high or low). This assignment process formed four cells of different combinations of communications, allowing the experimenters to assess the impact on subsequent beliefs, flossing intentions, and flossing behavior. Patients who had a higher level of perceived seriousness of gum disease as a result of viewing the communication subsequently increased in their intention to floss as well as their actual frequency of flossing, showing significant differences from the low perceived seriousness group. Patients with high susceptibility were more likely to floss more often, though the differences were not significant. Other investigations have demonstrated the efficacy of interventions designed to affect peoples' health beliefs and thus behavior. A major review study about research on the modification of beliefs and thus behaviors (Kirscht, 1974) found many efforts to utilize the situational and behavioral specificity of the Health

PAGE 76

66 Belief Model in devising experimental regimens. Most of the studies reviewed used communications to heighten threat or arouse fear on the part of the recipient, based on the premise that the resulting higher degree of perceived severity or perceived susceptibility would result in a greater likelihood that the behavior would change as a result. Results were complex, as might be expected in a multivariate model, but generally supported both the Health Belief Model and the effectiveness of the change regimens. Important mediating effects found to have an impact on the results were the presence of a response option perceived as effective in reducing the threat and other beliefs about the new behavior. In a test of the effects of interventions targeted on behavior change, Becker, Maiman, Kirscht, Haefner, and Drachman (1977) performed a study in which mothers of overweight children were exposed to fear-arousing communications based on Health Belief Model dimensions. Measures of subsequent appointment-keeping and childrens' weight loss showed both significant independent effects of the fear-arousing messages and significant correlations of the Health Belief Model variables with the behaviors. This experiment demonstrated the usefulness of the Health Belief Model as a whole through several multiple regressions, along with demonstrating that the influ e nc e

PAGE 77

of the fear-arousing messages was independent of the impact of beliefs. 67 Kirscht (1983) reports mixed results later from Health Belief Model investigations utilizing belief-change interventions. An outcome of adoption of the target behavior depended, generally, on the simultaneous presence of perceived susceptibility and severity, and on the behavior in question being perceived as beneficial and with few accompanying barriers or costs. Thus, the employment of the Health Belief Model in devising strategies for behavior change through persuasive communications requires the recognition that all the components play a role in the final outcome. A narrow focus utilizing only one component is susceptible to misapplication and misinterpretation of the results. When correctly utilized and with an appreciation for uncontrolled variables, interventions based on the components of the model appear effective in altering behavior. Kirscht (1983) cautions that too few studies exist that have examined long-term changes in behavior for definitive conclusions to be possible. Applying the Health Belief Model to the Exercise Adoption Problem The Health Belief Model has proven to be a successful tool for predicting the likelihood and frequency of many different kinds of preventive health behaviors. Another potential application receiving increasing research

PAGE 78

68 attention has been in evaluating the effectiveness of model-based interventions designed to positively affect personal behavior patterns. A tool for explaining, predicting, and changing health-related behaviors offers potential utility in the specific case of aerobic exercise. The need for mechanisms to support exercise adoption is easily documented. In a review of exercise participation, Powell (1988) found that only 9% of the population in the U.S. exercises at the level necessary to derive the benefits exercise has to offer. Current patterns in exercise research show widespread use of the findings of the American College of Sports Medicine (1978) on the minimum level of exercise necessary to develop and maintain fitness. These guidelines specify the following parameters for exercise quantity and quality: 1) frequency of training 3-5 days per week, 2) intensity of training 60%-90% of maximum heart rate reserve, 3) duration of training 15-60 minutes of continuous aerobic activity, with duration dependent on intensity, and 4) a mode of activity that involves large muscle groups, that can be maintained continuously, and is rhythmical and aerobic in nature (American College of Sports Medicine, 1978, p. vii). Given the almost universal employment of all or parts of these guidelines in fitness programs and in evaluative research, the current study was formulated to examine the association of Health Belief

PAGE 79

Model variables with exercise conforming to the American College of Sports Medicine parameters. 69 With a 50% dropout from aerobic exercise programs at six months (Dishman & Gettman, 1980), clearly a reliable regimen for assisting exercise adoption would play a useful role in such programs. The utility of such a program would apply whether the target was an individual or a group, whether the exercise adoption goal was self-motivated or prescribed. The development of such a regimen is dependent on adequate knowledge of what changes are being attempted. Dishman (1988) has observed, "one barrier to implementing public health promotions of exercise as well as uniform interventions in exercise programs is the absence of a consensus over the methods that might be effectively employed and the exercise determinants to be targeted" (p. 2). Research on exercise adoption must contain attempts to identify the key determinants and begin to apply them in experimental trials to determine their utility in effecting change. Hughes (1984) cites the need for more research to identify relevant characteristics of individual exercisers and to identify the best method to prescribe exercise as treatment. Such a base of knowledge is needed to develop the method for prescribing exercise as well as the best ways of insuring positive outcomes once the program has been undertaken. The need to take into account individua l factors affecting exercise adoption, and the challenge

PAGE 80

70 therein, was cited in discussion of a recent attempt to isolate factors influencing exercise adherence (Gale, Eckhoff, Mogel, & Rodnick, 1984). The authors noted the complexity and the difficulty in distinguishing different individual characteristics while controlling for other factors, observing that the phenomenon of adherence is likely the product of interaction between personal and program characteristics. Along with many other writers in the exercise adoption area, they cite the need for much more research in this area before a reliable program for change can be developed. As regards exercise and other forms of preventive health behavior, calls for more research on social and psychological predictors of health-related behavior have appeared in different quarters, emphasizing the paucity of experimental research that could provide the foundation for programs of change (McAlister, Farquhar, Thoresen, & Maccoby, 1976; Becker, Haefner, et al. 1977). For the example of aerobic exercise, additional steps are needed to identify specific beliefs (and the degree to which each are held) that are associated with successful and unsuccessful attempts at exercise adoption. A reliable method for determining relevant beliefs would provide for the development of effective psycho-educational and belief change regimens to augment the known precursors of behavioral change.

PAGE 81

71 An antecedent of the current interest in the possible contributions of the Health Belief Model to the explanation of and support of exercise adoption behavior was contained in an earlier discussion of the mechanisms of the model: Surely, the exercise and dietary mania observed over the last decade represent behaviors that could be regarded as striving toward improved health, but it is just as easy to explain them (insofar as they are health related at all) as behavior undertaken to avoid a deleterious situation. Again, there are individuals who exercise and engage in other health actions having health implications but who do so for reasons quite unrelated to health, perhaps for aesthetic reasons or for the sheer exhilaration felt by many by the performance of physical work. Again, the question of whether the avoidance orientation in the Health Belief Model is adequate to account for the so-called positive health actions taken by people remains unresolved. (Rosenstock, 1974a, p. 335) Although written some time ago, the uncertainty voiced in the discussion remains today, with only a few exploratory efforts having been undertaken to answer the questions raised. Such research serves as the foundation and guidance for specifying elements of the present study. Reviewers of recent research have verified the relationship of beliefs and attitudes to intentions to exercise and the exercise behavior itself (Godin & Shepard, 1986). In examining research up to now in the area of exercise behavior, Godin and Shepard support the theory of Ajzen and Fishbein (1977), noting that recent attitude-behavior research has already proceeded from "is" questions (Is there an effect?) to "when" and "how"

PAGE 82

72 questions. The second brace of questions, when answered, will provide information about when there is a relation between attitude and behavior plus how the effect of attitudes on behavior is mediated. Godin and Shepard (1986) lobby for the application of theory to this problem as opposed to the "shotgun" approach of trying out variables in varying combinations. The authors who have stimulated much of the recent interest and activity in attitude-behavior work did so by distilling and thematizing much of the research that had occurred up to the time of their extensive review (Ajzen & Fishbein, 1977). Of particular interest was the finding, cited above (pp 53-54), about the patterns of correspondence between attitudes and behavior. The result has been stud i e s that test the relationship in specific, particular inst a nces of behavior as well a s the more g l obal, broad questions Exercise provides the opportunit y f or both kin ds o f resea r ch, from the li ke lihood of doing a particul a r fo rm o f e x ercise at a part i cular time to g e n e ral adh erence over tim e Rid d le ( 198 0) dev i sed a test o f a ttitud e -beh a vior re l a tion s f o r a par ti c ul a r form of exe rci se -j og gi n g -and ut ilizi n g on e o f t he mai n models o f th e rela t i on s h ip betwee n attitudes and behavior. Ch oosing Fishbein s Be h avioral Intention Model, she developed a beliefs measure using an elicitation tec h nique that resulted in a

PAGE 83

73 68 item measure. With the criterion behavior identified as regular jogging, the survey assessed behavioral intention, attitude toward behavior, attitude toward the object, subjective norms for behavior, motivation to comply, consequences of behavior, and evaluation of the consequences. The focus of the study was on testing the validity of the Behavioral Intention Model, but clearly the Health Belief Model methodology served as a guide in the construction of the beliefs items on the survey. The survey was administered to 369 men and women of age 30 and over, with 296 usable surveys received. Joggers numbered 149, non-exercisers totalled 147. Results of the research, confirmatory of the Behavioral Intention Model, also showed marked belief differences between joggers and non-exercisers. Riddle (1980) documented differences between joggers and non-exercisers on beliefs corresponding to two Health Belief Model dimensions, perceived benefits and perceived barriers. On 17 of the 19 belief scales, differences between joggers and non-exercisers on the benefits/barriers items were significant at the p .001 level. The items were structured as follows: "Taking part in regular jogging in the next two weeks would ... ". Examples of the items showing significant differences are: "make me feel too tired," "take too much time," "make me feel good mentally," "make me have a fatal heart attack/stroke." Beliefs about the consequences of exercise

PAGE 84

74 in the form of regular jogging accounted for much more of the variance than did evaluation of the consequences. Joggers had strong beliefs about the positive consequences of jogging, while nonexercisers had neutral beliefs about the positive and negative consequences. Joggers thought regular jogging would benefit their physical and mental health, while non-exercisers thought it would require too much time and discipline and make them too tired. In another study utilizing techniques suggested by Fishbein's Behavioral Intention Model, Slenker, Price, Roberts, and Jurs (1984) established a promising precedent for the employment of the Health Belief Model with regard to predicting aerobic exercise adoption. Their investigation applied the model in a specific exercise situation, that of jogging. An instrument was developed and validated for purposes of measurement of beliefs specific to expectations about jogging, for the five customary belief areas examined in Health Belief Model research: perceived benefits, perceived barriers, perceived susceptibility to disease, perceived severity o f possible disease conditions, and general health motivation. Other variables under investigation were also part of the instrument, including knowledge about jogging, complexity of jogging, cues, and health locus of control. The completed instrument was administered to 40 joggers and 39 nonexercisers.

PAGE 85

75 A composite regression analysis of Health Belief Model variables that included one other variable accounted for 56% of the variance in jogging behavior. The Health Belief Model variables of perceived benefits, perceived barriers, and general health motivation comprised a multiple R of .707, or 50% of the total variance. The single variable perceived barriers accounted for 40% of the variance. The composite figure was a strong performance of the Health Belief Model variables in explaining exercise behavior. Given the careful instrument development and sound statistical analysis, the finding is one of the most impressive performances of a Health Belief Model-based instrument for any study examined. Using a discriminant analysis procedure, 92 percent of the subjects were correctly assigned to their proper category of jogger or nonexerciser based on the independent Health Belief Model variables assessed. Similar methods with an adolescent population yielded more equivocal results. The predictive measures produced more modest percentages of correct classification and the amount of variance accounted for by the measure employed for the study was also less (O'Connell, Price, Roberts, Jurs, & McKinley, 1985). Performed on a sample of 69 obese and 100 nonobese high school freshmen and sophomores, the study used the beliefs measures to predict status of obese or nonobese and to predict membership as an exerciser or a

PAGE 86

76 nonexerciser. The beliefs assessed differed from the study performed by Slenker, Price, Roberts, and Jurs (1984) in that measures of social approval for dieting and exercise were added, along with cues for the behavior in question (dieting or exercise). General health motivation was not assessed, while the other Health Belief Model variables--susceptibility, severity, barriers, and benefits--remained. Each of the four variables also obtained a measure of each of the subject's beliefs about the behaviors of diet and exercise. For obesity, the independent variables employed in the study correctly classified 69% of the subjects as obese or nonobese in a discriminant analysis procedure. The authors observe that obesity is not a behavior, identifying a likely reason for the relatively poor performance of their predictors. When employed to classify dieters versus nondieters, 83% of the subjects were correctly classified. When the discriminant analysis procedure was applied to exercisers versus nonexercisers, 75% of the overall classifications were correct. The amount of variance accounted for by the independent variables in the case of dieting behavior for obese and nonobese adolescents was 23% and 19%, respectively. For exercise behavior, 14% of the variance was explained for obese subjects, while none of the variables emerged as significant in predicting exercise behavior for the nonobese subjects. These weak findings in

PAGE 87

77 the application of the Health Belief Model to diet and exercise behavior are noted by the authors as an indicator that age may be an important factor in assessing health behaviors using the Health Belief Model. Such an observation is supported by Weinstein (1984), who found that college students have an unrealistically low assessment of their susceptibility to health problems and accidents. A study undertaken to determine the relationship of Health Belief Model components to compliance with an exercise prescription found that the beliefs of coronary bypass patients played a role in subsequent adherence to a walking exercise regimen (Tirrell & Hart, 1980). Five standard Health Belief Model components were assessed, perceived benefits, perceived barriers, perceived susceptibility, perceived severity, and general health motivation. Using a somewhat abbreviated measure consisting of 19 items, the researchers found that two of the components had significant correlations with compliant behavior, perceived barriers and perceived susceptibility. The correlation of the barriers measure with participation in the "heart walk" regimen was 0.64; with the susceptibility component the correlation was 0.35. The highest degree of significance was achieved with the perceived barriers component, which was significant at the

PAGE 88

p .001 level; perceived susceptibility was significant at the p < .05 level of significance. 78 In a study designed to relate two theories of behavior change, Sonstroem (1987) provides more verification of the role of beliefs in behavior change. Based on Ajzen and Fishbein's Theory of Reasoned Action (1977), the study examined belief statements of subj e cts with the aim of differentiating between various stages of change as detailed by Prochaska and DiClemente (1983). While the study did not explicitly base the be l iefs surve y utilized on the Health Belief Model, the categories o f beliefs that emerged from the analysis w ere coincident with the categories utilized in most Health Be l ief Mo d e l studies. A background questionnaire determined the exercise status of each sub j ect, allowing them to be assign e d a category (precont em plation, contemplation, act io n, m a intenanc e relap s e) in the Prochas k a a nd D i Cl eme nt e st a ge schema. Two hundred fourteen respondents w ere asked to r eact to b e lie f statements that h a d b e en suppli ed by e xe rcise le a d e rs at various exercise pro g ra m s. Liker t it e m s providing a gr a d e d r e sponse a cc o mpa ni e d each belief stat e ment Eac h s ta t eme nt h a d the root My partic i pation i n a r e gul ar pr o gram o f exercise would :" fol l owed by the sta t eme nt Examples of t he belief statements are : help me i n co n tr o lling my weigh t" ; increase my cardiovascular

PAGE 89

79 endurance"; "be boring". Principal component analysis on the resulting responses yielded seven usable components, labeled Benefits, Barriers, Novelty Outlet, Social Outlet, Fear, Appearance, and Psychological Gain. A stepwise discriminant function analysis utilizing the belief findings as a base was constructed to classify the expected Prochaska and DiClemente category for the precontemplation, contemplation, action, and maintenance stages. Mean item responses were computed for each subject for each component, which were then standardized to mean of 50 with a standard deviation of 10. Two functions which accounted for 99.99% of the total Eigenvalues resulted. The three components which entered the equation were Barriers, Benefits, and Psychological Gain. For the two functions constructed, 60% and 55% of the categories were correctly identified when compared to the subjects' actual status. For research of this type, the percentage of subjects correctly assigned was relatively low. Sonstroem (1987) cites a small sample size in some of the stage categories as a limitation, along with limitations in instrument development. A survey of research investigating the utility of the Health Belief Model quickly reveals that almost as many different instruments exist as do studies! While there are many idiosyncratic measures that have been employed in beliefs research, there are many parallels between studies, and a consensual choice of method. A consistent

PAGE 90

80 call in reviews of Health Belief Model research, however, is for a standardization of instruments and, where possible, in methods. Little Health Belief Model research is designed to replicate or refine earlier findings, nor are the positive findings that result later employed to design interventions that may provide for support of individuals attempting behavior change programs. Support for the reliability and validity of properly designed and administered instruments has recently emerged, documenting that such research tools surpass minimum requirements for such measures. A number of studies examining appropriate techniques for constituting and evaluating measures of health beliefs have validated the different beliefs posited for the Health Belief Model as discrete categories that do not overlap. Construct validity for the commonly researched components of the model has been consistently demonstrated, with compelling statistical evidence supporting the independence of the constructs (Maiman, Becker, Kirscht, Haefner, & Drachman, 1977; Cummings, Jette, & Rosenstock, 1978; Jette, Cummings, Brock, Phelps, & Naessens, 1981; Given, Given, Gallin, & Condon, 1983; Champion, 1984). The utility of the combined components of the model in predicting and explaining behavior has already been noted above. An important byproduct of this research activity has been the

PAGE 91

development of an item pool where the elements have been found to be valid and reliable. 81 The most complete test of a single Health Belief Model instrument intended for use in the measurement process was performed by Champion {1984). Designed to verify the validity of scales for use in predicting breast self-examination behavior, the investigation subjected each scale to an internal consistency examination, a test-retest reliability assessment, a test of exclusivity for each scale, and an overall measure of construct validity as demonstrated by the performance of the instrument in predicting breast self-examination behavior. The scales examined included perceived benefits of and barriers to performing breast self-examinations, perceived susceptibility to disease, perceived severity of possible disease conditions, and general health motivation. The final scales used in statistical analysis had between five and twelve items each. The instrument was administered to 301 women of varied ages, education, ethnicity and socioeconomic status. The five scales had internal consistency reliability coefficients (as determined by computing a Cronbach's Alpha for each scale) ranging from 0.60 to 0.78. Three of the scales had coefficients of 0.76 or higher: susceptibility, severity, and barriers. When test-retest reliabilities were calculated, all the scales except

PAGE 92

82 benefits exhibited coefficients of 0.76 or higher. The author observes that the testing process may have sensitized subjects to the benefits of breast self-examination, making for lower, but still significant, test-retest reliability. A principal component analysis utilizing orthogonal rotation with a varimax criterion showed the five scales to be mutually exclusive, with only one of the resulting factors having items from more than one of the scales. The severity scale lacked unidimensionality, with distribution of the items from that scale across three factors resulting. When scale scores were used to predict breast self-examination behavior, a multiple R of 0.51 was obtained, accounting for 26% of the variance. The barriers scale accounted for the largest portion of the variance, with health motivation second. The remaining three variables did not account for a significant portion of the variance. Champion (1984) saw the resulting instrument as a good basis for application with other behaviors, concluding that the method and the model are of sufficient validity for application in many areas. The study represents an important advance in Health Belief Model methodology, providing a useful framework and point of departure in developing a standard procedure for beliefs assessment. Champion assigned importance to the strategy of utilizing previously researched items as a mechanism

PAGE 93

83 for standardizing research results Evaluative criteria generally accepted as standards in test construction but seldom employed in Health Belief Model investigations are also important elements of the study. Other studies using similar tests with instruments of like construction report results of similar magnitude. Wagner (1983) found that the Cronbach's alpha reliability coefficients for her Health Belief Model scales ranged from 0.65 up to 0.91, with all but one of the scales above 0.76. Her scales employed Health Belief Model components in ways specific to medical applications that assessed the psychosomatic elements of beliefs and behaviors. Using a factor analytic approach to evaluate Health Belief Model components yielded inconclusive results as to the unitary dimensionality of the usually robust barriers and benefits scales. This appears to be due to the inclusion of a wide range of items representing too many different activities and symptoms. Another study that also did an evaluative appraisal of an instrument development effort found distinct components and moderate reliabilities (Jette, Cummings, Brock, Phelps, & Naessens, 1981). These results were achieved despite the use of only a few items in each scale. A factor analysis yielded a concordance with existing Health Belief Model constructs of perceived barriers, perceived severity of disease, and concern for

PAGE 94

84 health. These results are of interest given that they were noted in a study that primarily investigated behavior associated with medical regimens. Even without items intended to measure the Health Belief Model components, the factors were identified. Examination of the methodology of instrument construction and use in Health Belief Model studies led to the conclusion that "moderately reliable indices of a wide spectrum of health beliefs can be constructed and replicated across samples." (Jette, Cummings, Brock, Phelps, & Naessens, 1981, p. 92). The authors further note that the obstacles to replicability of results stem from the lack of standardization of instruments and of sufficient rigor in the construction of beliefs measures. The previously cited study by Slenker, Price, Roberts, and Jurs (1984) (p. 82) also serves as a good foundation for the present study. It found the Health Belief Model to be both valid and reliable in the specific exercise application of jogging. As with the Champion (1984) study, measures of internal consistency and construct validity were applied. The five Health Belief Model scales of perceived benefits of jogging, perceived barriers to jogging, perceived susceptibility to disea se perceived severity of possible disease conditions, and general health motivation achieved KR-20 reliability coefficients of 0.83 or greater with the exception of general health motivation, which has a coefficient of

PAGE 95

0.57. Given that the KR-20 procedure is intended for dichotomous variables, it is unclear what a more appropriate analysis would yield in evaluating the components. Factor analysis employing an orthogonal varimax rotation revealed the five constructs to be distinct factors. 85 Of the possible survey/questionnaire mechanisms for obtaining an accurate measure of an individual's beliefs, the most frequently used method in Health Belief Model research has been the Likert scale. The seven-point Likert method of assessing beliefs has been found to have convergent validity with interview and multiple choice methods of assessment, two other popular assessment strategies. Of the methods compared, the Likert method showed the greatest validity, with zero method effects as opposed to 10 % for both multiple choice and interview methods (Cummings, Jette, & Rosenstock, 1978). Likert items have been the assessment method employed in most Health Belief Model studies, with consistent results. Lau, Hartman, and Ware (1986), performed a study examining the impact of health values on certain health protective behaviors. In a population of university students, those with higher valuing of health were more likely to report doing the behaviors, such as seat belt use, breast self examination, and exercise. Those with higher health-as-a-value scores were significantly more

PAGE 96

86 likely to have done the behaviors than those with low scores. Also, the authors cite an increase in health-as-a-value scores with increasing age within their subject pool. These values stabilized as students moved into adult age ranges, at a level still somewhat lower than in the adult population. The finding that age is an important factor to be taken into account in the measurement of beliefs and their impact on subsequent behavior corroborates a finding cited above (O'Connell et al., 1985). To avoid confounds in developing instruments and interventions utilizing the Health Belief Model, the use of adults of 25 years of age or older in preliminary studies appears warranted. Age 25 appears to clear by a safe margin the age range where detectable changes in beliefs are occurring. Hypotheses The Health Belief Model has shown a consistent capacity for prediction with many different kinds of health behaviors, as well as an ability to account for significant amounts of the variance in retrospective and prospective health behavior investigations (Janz & Becker, 1984). The pattern in the choice and application of instrumentation has been variable, with different methods adopted according to the particular perceived n eeds of each separate research effort. With the recent spate of research, particularly studies applying the model to the area of exercise behavior, study design has begun to

PAGE 97

stabilize into a well-documented and proven methodology with a foundation in accepted assessment practice. 87 Extending Health Belief Model research into the area of exercise behavior is a step indicated by many other sick role and preventive health behavior studies, and shows good potential utility for the specific example of general aerobic exercise. No empirically verified, replicable link has been established between the major Health Belief Model components and systematically defined, general aerobic exercise as a dependent variable. No prospective studies have examined the ongoing impact of beliefs in this area. The success of intervention investigations in the area of aerobic exercise depends on the existence of a reliable, repeatable measurement instrument that provides a basis for assessment of effects. Interaction between beliefs and behavior precludes absolute statements of causality, but repeated measures with a proven instrument provides a basis for first steps in defining the evolution of individual patterns of behavior. As was observed early in the first burst of research activity to use the Health Belief Model, "the hypothesis that behavior is determined by a particular constellation of beliefs can only be tested adequately where the beliefs are known to have existed prior to the behavior that they are supposed to determine" (Rosenstock,

PAGE 98

1974b, p. 362). Such findings hang on the existence of a tool with known characteristics and validity to as accurately as possible determine belief constellations specific to the area being researched. 88 The present study sought to develop and verify the attributes of just such an instrument, with regard to a carefully delineated yet general (many different kinds of exercise) definition of aerobic exercise. Following the evaluation of the validity, reliability, and exclusivity of the constructs, a test of the predictive power of the overall measure over a time period of one month was attempted. The following hypotheses were advanced: 1. The five Health Belief Model constructs of perceived benefits of aerobic exercise, perceived barriers to aerobic exercise, perceived susceptibility to disease, perceived severity of possible disease conditions, and general health motivation, as operationalized by subscale s of the Personal Beliefs Questionnaire, will be found to be reliable when evaluated by computed internal reliability coefficients and test-retest statistics. 2. The five Health Belief Model components, as measured by the Personal Beliefs Questionnaire, will be found to be valid, indepen dent constructs when assessed by measures of intercorrelation and factor analysis statistics. 3. The refined composite of health belief components, resulting from item analysis of the Personal Beliefs Questionnaire and utilizing the five belief components employed in this study, will account for a significant portion of the variance in classification of self-assigned aerobic exercise adoption categories. 4. The refined composite of health belief components from the Personal Beliefs Questionnaire will account for a significant portion of the variance determining class of actual activity/exercise participation as measured by subjects' self-report.

PAGE 99

89 5. The refined composite of health belief components from the Personal Beliefs Questionnaire will successfully predict membership in subjects' self-assigned categories of aerobic exercise adoption. 6. The refined composite of health belief components from the Personal Beliefs Questionnaire will successfully predict membership in classifications of actual activity/exercise participation as measured by subjects' self-report. 7. The refined composite of health belief components from the Personal Beliefs Questionnaire will successfully predict self-reported changes in subjects' exercise status occurring over the period of one month.

PAGE 100

CHAPTER 3 METHOD Subjects Participants in the study were male and female adult faculty at a mid-size college in central New Jersey. The sample was drawn from a pool of all full-time, on-campus faculty actively teaching during the Spring, 1990 semester. To avoid confounds from maturation effects noted in past investigations with younger populations, (O'Connell et al., 1985; Lau et al., 1986), only individuals over 25 years of age were utilized in the study. A convenience sample of 334 faculty at the college was selected from those in the available pool. Those faculty members selected were sent a mailing containing a letter describing the study and requesting participation, (Appendix A) a reply/consent form (Appendi x B), the "Personal Beliefs Questionnaire" (Gage, 1990; Appendix C), a survey return envelope, and a reply/consent form return envelope. One month following receipt of each subject's response to the first questionnaire, a second mailing was sent soliciting a second response. It contained a contact letter, the second form of the questionnaire (Appendix D), and reply materials. This response was sought to enable assessment of test-retest 90

PAGE 101

reliability for the beliefs measure and to identify any changes in aerobic exercise status. 91 Those faculty not responding to the first mailing of each version of the questionnaire received reminder letters, succeeded by a followup mailing with contents duplicating the first. A lottery reward system was employed to encourage participation, using a drawing for a 50 dollar cash incentive for the first response, and 100 dollars for the second. Receipt of the reply/consent forms qualified each respondent for the drawings. The use of separate reply/consent forms enabled responses to be anonymous to assure honest self report. Subjects were prompted for the last four digits of their home phone number on each questionnaire as an identifying code for matching the questionnaire protocols in the subsequent analysis. A total of 181 subjects sent in completed questionnaires and the reply/consent form, for a response rate of 54 percent. Two of the responses were unusable, one because the subject was under 25 years old, one because the questionnaire was only half completed. This left 179 usable first responses. The demographic characteristics of the subjects in this initial pool are summarized in Table 1 below. All these responding subjects received the second questionnaire within the specified time.

PAGE 102

92 Table 1: Summary of Demographic Descriptors For All Respondents to Questionnaire One. Observed Age Ranges Mean Age Mean Weight Gender Marital Status Education Ethnicity 3--28-29 Years 18--30-39 Years 58--40-49 Years 66--50-59 Years 31--60-69 Years 50.7 Years Old (n=l76) 167.8 Pounds (n=l76) Old Old Old Old Old 114--Male (n=l77) 63--Female 15--Single 4--Separated (n=l78) 136--Married 20--Divorced 5--Living Together 41--Masters 125--Doctorate 11--Postdoctorate (n=l78) 161--Caucasian/White 4--African-American/Black 2--Native American 9--Asian/Pacific Islander (n=l76) (Note: n varies due to omitted responses. ) For the second mailing, 165 completed returns were received, for an overall response rate of 49 percent and a 92 percent return rate by those completing the first questionnaire. As the responses were evaluated following data entry, it was discovered that number errors or omissions in subjects' recording of their phone nu mber identifier resulted in the loss of fifteen pairings for

PAGE 103

the responses, leaving a total of 150 usable pairs for those steps of the analysis requiring both replies. Instrumentation 93 A questionnaire was developed for the measurement of personal health beliefs about health, illness, the benefits from aerobic exercise, and the barriers to aerobic exercise. The measure was entitled the "Personal Beliefs Questionnaire" (words for the title were chosen to avoid communicating a "set" to subjects). The instrument is composed of a beliefs assessment section, a section to determine the subject's own classification of their aerobic exercise adoption stage according to specified stages of aerobic exercise adoption (after Prochaska & DiClemente, 1983), a section for subjects to detail specific aerobic exercise behavior, a section to detail significant activities that were not aerobic exercise, and demographic background data on each subject. Health Beliefs The part of the questionnaire assessing health beliefs was patterned after work done on the Health Belief Model in other related areas (Becker, Maiman, Kirscht, Haefner, & Drachman, 1977). In an effort to assure maximum validity and reliability, items selected for use in the instrument retained the exact structure of previously used and validated items. Studies that served as sources for the items included Maiman et al. (1977), Jette, Cummings,

PAGE 104

Brock, Phelps, & Naessens (1981), Champion {1984), with the employment of "aerobic exercise" as the criterion behavior for the items. 94 Additional items were selected from a pool drawn from instruments of known reliability and validity in assessing beliefs about exercise {Slenker, Price, Roberts, & Jurs, 1984; Cioffi, 1984; Sonstroem, 1987; instruments obtained from the authors). All items referring to beliefs about the criterion behavior of sustained aerobic exercise were accompanied in the instrument by the guidelines for aerobic exercise formulated by the American College of Sports Medicine {1978). These guidelines appear on every page where exercise-related questions are listed. Included in the beliefs portion of the questionnaire were questions assessing five belief components: General Health Motivation, Perceived Susceptibility to illness, Perceived Severity of illness threat, Perceived Benefits of aerobic exercise, and Perceived Barriers to aerobic exercise. All items used in the beliefs portion of the questionnaire were constructed as seven-point Likert responses. The Likert method of investigation has been found to have the greatest convergent validity in Health Belief Model investigations {Cummings, Jette, & Rosenstock, 1978). Consideration was also given to the number of items used to sample each component 's domain, to

PAGE 105

95 insure overall reliability for that subscale. The minimum number of items accepted for use on any one scale was nine. Following item construction, an evaluation of the items was conducted to insure content, or face validity for their respective subscales. This was accomplished through having the pool of items judged by experts familiar with exercise supervision and exercise programs. These experts were leaders in exercise and wellness programs in the Trenton area. Three such evaluators were asked to affirm the suitability of each item for measuring its intended component by assigning a randomized list of items to the hypothesized subscales after being familiarized with the conceptual intent of each. Those items not correctly assigned by at least two of the three judges were eliminated from further use. Feedback from the judges about the items was also considered in refining the final version of the questionnaire. Exercise Status The self-assessment of exercise participation was determined using a schema adapted from a study on smoking behaviors (Prochaska & DiClemente, 1983). Such a classification schema with exercise has been piloted by Sonstroem (1987). Subjects were requested to identify themselves using this framework. The actual labels were not employed in the instrument; categories were simply

PAGE 106

96 numbered one through seven. Exercise, for this investigation, was considered to be intentional, regular aerobic (vigorous, with elevated heart rate) exercise, at least 3 times/week, at least 20 minutes per session. This level represents the most universal guideline for the level of exercise needed to sustain fitness (American College of Sports Medicine, 1978). The framework labels as adapted for exercise were: Long term exerciser (Category #6)--Someone who has maintained a pattern of regular aerobic exercise for a period of not less than 18 months. Recent exerciser (Category #5)--Someone who has been engaging in aerobic exercise for any period up to nine months. Contemplator (Category #2)--Someone who has been seriously thinking about beginning an aerobic exercise program in the past year. Immotive (Category #1)--A person who is not an exerciser and not considering it. Relapser (Category #3)--A person who has tried but failed--for a period of less than six months--to begin an exercise program in th e past year (adapted from Prochaska & DiClemente, 1983). Additional categories were added for purposes of this study, Past Exerciser (Category #4)--A person who has maintained a program of regular exercise for at least six months sometime in the past five years, and Dedicat ed Long Term Exerciser (Category #7)--A person who is a long term exerciser, but who exercises daily and for periods of longer than 45 minutes. The categories were numbered in a

PAGE 107

hierarchical order of increasing investment in regular aerobic exercise. 97 The seven categories were presented in a section titled Your Exercise Pattern in the questionnaire. Accompanying the categories was a detailed description of the American College of Sports Medicine (1978) definition of the aerobic exercise parameters, for subjects to refer to in classifying their own exercise adoption stage. Subjects were prompted to choose the category that described their own current status. Following the category self-assignment section was a section for subjects to describe the type, duration and frequency of aerobic exercise they are pursuing for use in verifying the accuracy of their responses. Such a survey method of physical activity assessment has been noted as the method of choice for this kind of research (LaPorte, Montoye, & Caspersen, 1985; Washburn & Montoye, 1986). It has been demonstrated that machine-aided objective methods are now achievable, but such techniques are seen as cumbersome and too expensive for most research (Washburn, Cook, & LaPorte, 1989). Recall methods do have shortcomings in terms of accurate reporting (Perkins & Epstein, 1988), but the limitations were deemed acceptable given the need for a global view of exercise participation in this study. Thus, while considerably more comprehensive and detailed methods

PAGE 108

98 of assessment are possible, for purposes of this stud y a characterization of the typical weekly activity level was all that was sought. Since significant levels of other activity were expected to be a part of subjects' lifestyles, an additional section followed, paralleling the one on aerobic exercise but prompting subjects to note activity that "represents exercise, but is not described by the definition of regular aerobic exercise". Space for subjects to describe the type, duration, and frequency of their activity was provided. Such assessment is considered a necessary part of appraising a subject's overall level of activity (Washburn & Montoye, 1986). Information gleaned from the activity self-reports was used to construct a master variable denoting activity level. This was done to get a graded, purely behavioral assessment of involvement in activity and the level of that activity, as a contrast to the adoption stage self-assignment categories. The first category was labeled Sedentary, denoting those individuals who indicated no activity on either of the above self-report lists. The second category was named Minimal Activity, and was us ed for those who indicated no aerobic exercise, a duration of exercise less than twenty minutes per week and two or fewer times per week. The third category, Active, contained all respondents whose activity level exceeded the criteria for Minimal Activity but did not meet th e study guidelines for aerobic exercise. Aerobic was the

PAGE 109

99 final category, including all subjects whose activity was greater than three times a week, for longer than twenty minutes per session, and a verified aerobic exercise as defined by the American College of Sports Medicine (1978). For the second administration of the questionnaire, revisions were made to detect changes that had occurred in the intervening month (Appendix D). A block of questions duplicating those in the first was included for subjects to indicate the level of aerobic exercise they had been pursuing in the time since the first questionnaire was taken. Additionally, an item to allow subjects to indicate whether there report represented a change in exercise status from the first to the second response to the questionnaire was included; a supplementary question sought information as to the perceived reason for the change. Demographic Data For the final section of each survey, each subject was prompted to provide information on their age, weight, level of education, marital status, gender, and ethnicity. The first four questions appeared on the first survey, while the latter two were inserted on the second. Information about weight was obtained in both administrations for possible use in subsequent analysis. Results for all qualifying subjects who fully completed the first questionnaire are summarized in Table 1 above.

PAGE 110

100 Procedure The first steps in the analysis of the collected data were designed to verify the reliability and validity of items employed in the Personal Beliefs Questionnaire. The method for these evaluative steps was based on the literature addressing test construction and evaluation (Cronbach, 1970; Keppel, 1973; Anastasi, 1978) as well as past Health Belief Model investigations with similar purpose or parallel goals (Maiman et al., 1977; Jette et al., 1981; Champion, 1984, Slenker et al., 1984). The goal was to develop a procedure that would capture as its features a standard and widely accepted methodology for Health Belief Model instrument development and application. In this manner results of the current study would be more readily interpretable, and future research e s in this area would be provided with a greater foundation for replication or parallel investigation. Data entry from the completed questionnaires w as accomplished using a numeric keypad on an IBM terminal. Following entry of the data, the accuracy of the data capture was verified by comparing the data list with the original questionnaires. Analyses were run in a m a in frame environment using the Statistical Analysis System (S A S) The verified data set was used as the basis of all f u r th er analyses, with adjustments in the alignment of var i ab les continually saved as a master data s e t. Ma st e r var ia bl es

PAGE 111

101 that summarized some of the data were created to simplify the analysis process. Three steps were used to evaluate the reliability and validity of the items comprising each subscale. First, inter-item associations within each subscale were tested; internal consistency of each subscale was examined by calculating a coefficient alpha utilizing univariate statistics generated on the data (Cronbach, 1970). For the second step, inter-item correlation matrices were obtained for each subscale. Those variables showing unacceptably low correlation with their respective subscales were excluded from use in any further analysis. Finally, test-retest correlation coefficients were calculated for each subscale, utilizing items that were shown to be valid in the above tests. Item-by-item test-retest correlations were also examined to further identify items with unacceptably low reliability before instrument testing began. Using results of the above tests to identify items inconsistent with their scales or showing poor reliability, those items were eliminated from the variable lists used in the succeeding calculations. The scale tests of internal consistency, test-retest reliability, and intercorrelation were re-computed to assess the change that resulted from the new configuration and to insure continued scale function at optimal levels. When scale statistics were judged to have an acceptable alignment as

PAGE 112

102 indicated by the balance between reliability, independence as assessed by measures of intercorrelation, and internal consistency, that alignment was used for the analyses that followed. Factor analysis was used as the first measure of construct validity. The first step in the analysis was a principal component factor technique, employed to determine whether the theoretical constructs of General Health Motivation, Perceived Susceptibility to illness, Perceived Severity of illness threat, Perceived Benefits of aerobic exercise, and Perceived Barriers to aerobic exercise were sufficiently distinct to be considered discrete beliefs. The technique also provided for interpretation of the obtained clusters of items. The principal component analysis facilitated identification of independent components and assessment of the overall amount of variance explained within each of the beliefs as measured by the instrument. A varimax rotation was used to verify distinct factors through assuring orthogonality as the factors were extracted (Keppel, 1973; Ray, 1982; Afifi & Clark, 1984). Item by item correlations of questions with the different exercise adoption categories were computed to identify those items that alone showed significant correlation with each of the categories. Multiple regression was then applied using the five subscales, to

PAGE 113

103 determine the amount of variance each was able to account for in the overall regression equation. Multiple regression is the tool of choice for evaluating the variance accounted for by subscales of a complete measure (Afifi & Clark, 1984; Kerlinger & Pedhazur, 1973). Multiple regression allows the identification of those subscales accounting for the most variance in the various behavioral outcomes (Ray, 1982; Kerlinger & Pedhazur, 1973). It thus appears appropriate for successfully identifying the degree to which each belief component contributes to each dependent variable, the categories of exercise adoption or exercise behavior assessment measures. Another feature of multiple regression that suits it to the present study is that it can be used in stepwise fashion to identify the best equation for different independent variables (Kerlinger & Pedhazur, 1973; Isaac & Michael, 1983). To evaluate the utility of the overall beliefs scale for predicting exercise class outcomes, discriminant analyses were performed for the two measurements of the study population. Subjects were grouped according to their classification of their aerobic exercise adoption category. The percentage of subjects correctly classified within each category was noted, along with those subscales accounting for the greatest portion of the overall variance.

PAGE 114

CHAPTER 4 RESULTS Evaluation of the Personal Beliefs Questionnaire For item analysis and scale refinement purposes the items comprising the original questionnaire were assigned codes to be used in all succeeding steps (Appendix E). Data from the both administrations (First n=l79, second n=l65, matched pairs n=l50) were used in the instrument evaluation steps. For comparative purposes the performance of items in the second administration of the questionnaire was noted, in order to highlight any discrepancies or unusual changes in the performance of the instrument in its two uses. No such discrepancies were observed. The first evaluation of the scales, for internal consistency, yielded the results listed in Table 2. The coefficient alpha (Cronbach's Alpha) for each subscale was calculated utilizing univariate statistics generated on the data using the SAS routine UNIVARIATE. The five subscales--general health motivation (MOT), perceived susceptibility to disease (SUS), perceived severity of possible disease conditions (SEV), perceived benefits of aerobic exercise (BEN), and perceived barriers to aerobic 104

PAGE 115

exercise (BAR)--all had alpha coefficients of 0.773 or greater. Table 2: Coefficients Alpha for Initial Groupings of Questionnaire Items. Subscale Number of Items Coefficient Comprising Subscale Alpha MOT 9 0.833 SUS 15 0.829 SEV 15 0.773 BEN 12 0.931 BAR 11 0.888 105 For the next step, inter-item correlation matrices were generated for all variables and subscale scores, to allow each item's correlation with its subscale to be extracted. The items showing nonsignificant correlations with their respective subscales were then noted for removal from any further analysis (see Appendix E). An additional criterion for screening items was the pattern within the correlation matrix for each subscale. Items standing markedly apart from the general level or pattern of correlations achieved by companion items were also eliminated. Items excluded on the basis of poor relative intercorrelation with their subscales are listed in Table 3 along with the correlation each item had with its scale.

PAGE 116

106 Finally, test-retest correlation coefficients were calculated for the subscales to assess further the reliability of the scales. Results appear in Table 4. Item-by-item test-retest correlations were also examined to further identify items with unacceptably low reliability before instrument testing began. Only one item not already screened was eliminated on the basis of poor test-retest performance, a BAR item (Rl). The item removed addressed worry about regular exercise causing health problems. Table 3: Questionnaire Items Screened Due to Poor Relative Association With Subscale. Item Code and Content Product-Moment Correlation with Subscale Total Score (Pearson's r) H8 (MOT) Other things more important than health. Sl (SUS) Chance of getting asthma. V8 (SEV) Seriousness of cold V9 (SEV) Seriousness of a cavity Vll (SEV) Seriousness of sickness Vl2 (SEV) Seriousness of lacking energy Vl5 (SEV) Seriousness of overweight Bl2 (BEN) Regular exercise provide friendship/socializing ***=p .001 **=p < .01 *=p .05 0.468*** 0.395*** 0.168ns 0.083ns 0.232* 0.368*** 0.311** 0.580*** Note: See Appendix E for complete item texts. Removal of items from the subscales was expected to produce changes in their performance. To assess the change, the scale tests of internal consistency,

PAGE 117

107 test-retest reliability, and intercorrelation were re-computed to assess the change brought about by revision of the subscales. Table 5 displays the resulting reliability coefficients. For all scales but BAR, internal consistency was improved; consistency coefficients were all 0.832 or greater. Table 4: Questionnaire Subscale Test-Retest Correlation Coefficients. MOT Subscale SUS Subscale SEV Subscale BEN Subscale BAR Subscale n=l50 ***=p s .001 Correlation Coefficient 0.829*** 0.819*** 0.575*** 0.841*** 0.880*** Test-retest reliability as gauged by computed correlations retained the levels of association displayed in the first administration of the questionnaire (Table 6). A diminution in the observed correlations was seen in four of the five subscales, as was expected when items were removed from the test measure. The common pattern is that a scale with fewer items generally is less r e liable than a longer one. All scale correlations remained significant at the p s 0.001 level. Additionally, only on e

PAGE 118

108 item was found to be below the p < 0.001 level of significance for correlation with its subscale, achieving a p 0.01 level of significance. All retained items were significant at the p 0.001 level on test-retest correlations. Based on the results of the test-retest correlations and the measures of internal consistency, the scales were judged to be reliable and Hypothesis 1 was Table 5: Coefficients Alpha for Final Groupings of Questionnaire Items. Subscale Number of Items Coefficient Comprising Subscale Alpha MOT 8 0.850 SUS 14 0.832 SEV 10 0.859 BEN 11 0.935 BAR 10 0.859 accepted. Scale alignments were judged to have an acceptable structure as indicated by sampling, reliability, and internal consistency. The revised structure was thus adopted for the ensuing analyses. Toward assessing the construct validity of the subscales, a determination of the degree to which the subscales were mutually independent was sought Prior to the application of factor analysis, the correlation matrix

PAGE 119

Table 6: Questionnaire Subscale Test-Retest Correlation Coefficients for Final Item Groupings. MOT Subscale SUS Subscale SEV Subscale BEN Subscale BAR Subscale Total Scale n=150 ***=p < .001 Correlation Coefficient 0.811*** 0.818*** 0.531*** 0.840*** 0.883*** 0.813*** 109 obtained in the above determinations of reliability for the revised instrument was used as the source of a matrix of subscale intercorrelations, reproduced in Table 7. Each scale correlation with each other scale below the diagonal was recorded. A number of the intercorrelations between subscale scores were significant. To appraise the degree to which subscales function as independent constructs, factor analysis was used as the second overall measure of construct validity. The first step in the analysis was a principal component factor technique, employed to determine whether the theoretical constructs represented by the MOT, SUS, SEV, BEN, and BAR subscales were in fact sufficiently distinct to be considered separate beliefs. Using the SAS routine FACTOR,

PAGE 120

110 a factor analysis with principal components specified in an orthogonal varimax rotation yielded a factor matrix. Scree plotting of the eigenvalues obtained served to identify five factors as the most meaningful (accounting for the most variance proportionately) of those obtained. Table 7: Subscale Correlation Matrix for Final Item Groupings. MOT SUS SEV BEN MOT Subscale (1) SUS Subscale -0.064ns (1) SEV Subscale 0.14lns 0.267** (1) BEN Subscale 0.361*** -0.028ns 0.206* (1) BAR Subscale 0.422*** -0.229* -0.ll0ns 0.395*** n=l79 ***=p .s .001 **=p .s .01 *=p < .05 BAR ( 1) The five factors together accounted for 51% of the overall variance. The factor groupings are shown in Table 8 with their eigenvalues. Groupings were obtained by noting factor loadings occurring in clusters, bounded by drops in the loadings of other variables. Only items with factor loadings of 0.35 or greater were used in the groupings. The examination of factor grouping content will be discussed in Chapter 5. Based on factor analysis results and an examination of the subscale intercorrelations, Hypothesis 2 was only weakly supported and must be

PAGE 121

111 rejected. There appears to be significant association between certain subscales, most notably MOT, BEN, and BAR. Table 8: Summary of Factor Analysis Results. Factor 1 2 3 4 5 Subscale and Eigenvalue Items BEN ( 9 ) 10 3 81 BAR (5) MOT ( 1) SEV (6) 6.806 SUS ( 1) MOT ( 5) 3 7 9 5 SUS (6) 3.476 BAR (2) 2.690 Percent of Cumulative Variance Percentage 19.59 19.59 12.84 32.43 7.16 39.59 6.56 46.15 5.07 51.22 A Test of the Predictive Validity of the Personal Beliefs Questionnaire An examination of the distribution of subjects in the exercise adoption categories found a good distribution with the exception of category 7, as may be seen in Table 10. Given so few subjects in the top category, the decision was made to combine Categories 6 and 7 for the tests of predictive validity. Prior to the regression analysis, item correlations with the dependent exercise adoption category variable obtained in the second administration were computed. The matrix was examined to identify those items showing greatest correlation with each of the categories. Items

PAGE 122

correlating at a level 0.35 or higher are reported in Table 10. Items were keyed so that all responses in the Table 9: Distribution of Subjects in Exercise Adoption Categories. Exercise Adoption First Second Category Administration Administration 1 22 (12.3%) 21 (12.8%) 2 20 (11.2%) 24 (14.6%) 3 27 (15.1%) 23 (14.0%) 4 39 (21.8%) 38 (23.2%) 5 18 (10.1%) 15 (9.1%) 6 49 (27.4%) 39 (23.8%) 7 4 (2.2%) 4 (2.4%) Total 179 Total 164 112 direction that would be expected to result in health protective behavior scored low values, thus all correlations with the increasing value scale of exercise adoption are negative. The results of this analysis showed the strongest and greatest number of correlations to be occurring in the Barriers scale. Stepwise multiple regression was then applied usin g a composite of all independent variables m eas ured, including the five health belief subscales, to determine the strength of each in the overall regression equation The analysis was implemented through the SAS routine STEPWISE

PAGE 123

113 (Ray, 1982) to identify the subscales accounting for the variance in the criterion categories and exercise levels Table 10: Individual Items Correlating Most Highly With Adoption Stage Criterion. Item Number and Subscale H3 (MOT) HS (MOT) H6 (MOT) H9 {MOT) Bl (BEN) RS (BAR) R6 (BAR) R7 (BAR) R8 {BAR) R9 {BAR) Rl0 (BAR) Content (For complete text see Appendix E) How good job caring for health right now Best description number behaviors to insure health Frequently do things to improve my health Health habits I practice Pearson Product Moment Correlation Coefficient -0.37*** -0.36*** -0.40*** -0.36*** How much regular exercise improve your overall health -0.36*** How difficult maintain -0.64*** program regular exercise How much regular exercise -0.51*** interferes with activity Regular program exercise -0.57*** difficult to schedule Maintaining program regular -0.57*** exercise more trouble than worth Preference to do other -0.45*** things could keep me from exercise Lack of desire could keep -0.57*** me from regular exercise Matched pairs n = l50 ***=p .001 observed in the second administration of the questionnaire. The end product is shown in Table 11, in which all variables were included to exhibit the role each plays in the prediction of adoption category. Only two of

PAGE 124

114 the var i ables were significant at the p < 0.05 level, BAR and SEV. The barriers subscale appears to play the dominant role in prediction, accounting for almost 40 % of the overall variance. About 44% of the variance was accounted for overall with all the predictor variables entered; results were significant at the p 0.001 level. The only other predictor variable achieving significance was the severity subscale, accounting for about 2% of the v a riance. The only non-belief variable contributing meaningfully to the prediction of the dependent vari a bl e of exercise adoption category was age, with over one percent of the variance attributable to this feature. Table 11: Sou r ces of Variance in Exercise Adoption Categories. Variable BAR S E V Age B EN MOT SU S Wei ght E t hnici t y Mar S t a tu s Term. D e g ree Ge n der R Squa r e 0. 3 96 0. 4 12 0.4 2 6 0. 43 2 0 434 0. 435 0. 43 7 0 .438 0 438 0 438 0.438 Change in R Square 0.396* 0.016* 0.014 0.006 0 002 0.001 0.002 0 001 0.000 0 000 0.000 =p 0 05 **=p 0 001 (Final OF 11,133 ) O verall F V alue 93 59*** 49 74*** 34 85* * 26.58* ** 21 28 ** 17 73*** 1 5 .1 6*** 13 23*** 11 70*** 10.46*** 9 44 *** The analysis was repea t ed u t ilizing only the five belief variables, to differentiate th e in s trum e nt

PAGE 125

115 variables from those comprising the makeup of the subjects taking the measure, and provide full focus onto the test of the Health Belief Model belief variables. Such a prediction analysis shows which variables would be of interest to future intervention studies in the area of aerobic exercise adoption. Table 12 exhibits the findings from the stepwise procedure, showing a change in the order in which the belief variables entered into the prediction equation. Only the BAR subscale was significant, accounting for almost 37% of the overall variance. Overall, the combination of the five subscale variables accounted for 40% of the variance in predicting aerobic exercise adoption stage. Based on the results of the analysis for the five belief variables, Hypothesis 3 was accepted. Table 12: Sources of Variance From Health Belief Model Belief Subscales in Exercise Adoption Categories. Variable R Square Change in Overall R Square Value F BAR 0. 368 0.368* 85.60*** MOT 0.383 0.015 45.29*** BEN 0.390 0.007 30.94*** SUS 0.395 0.005 23.54*** SEV 0.397 0.002 18.86*** *=p < 0.05 ***=p < 0.001 (Final DF 5, 143) A similar brace of stepwise regressions was performed utilizing the two variable groupings, with the criterion behavior the exercise assessment category synthesized from

PAGE 126

116 subjects' self-report about activity. For the first regression, it became apparent that the overall composite of variables was accounting for less of the variance with the behavior based category than with the categories based on self-assignment to exercise adoption categories (Table 13). Approximately 36% of the overall variance in exercise assessment category was explained by the eleven independent variables. The most meaningful variable in the exercise assessment analysis was, again, the barriers subscale, accounting for 31% of the variance. For the analysis of the composite variable set, it was the only variable to achieve significance at the p~0.05 level. The order in which the variables entered into the equation in this case Table 13: Sources of Variance in Exercise Behavior Categories. Variable BAR SEV MOT Ethnicity Term. Degree BEN Age SUS Mar. Status Gender Weight R Square 0.310 0.343 0.350 0.353 0.355 0.357 0.358 0.358 0.358 0.358 0.358 Change in R Square 0.310* 0.033 0.007 0.003 0.002 0.002 0.001 0.000 0.000 0.000 0.000 = p 0.05 *** = p 0.001 (Final DF 11,134) Overall F Value 64.74*** 37.35*** 25.55*** 19.26*** 15.43*** 12.84*** 10.98*** 9.56*** 8.44*** 7. 5 4*** 6.80***

PAGE 127

11 7 changed as well, with the demographic variables Ethnicity Terminal Degree, and Age entering in ahead of the last beliefs variable. The amount of variance accounted for by these variables, however, was negligible. For the second run on exercise assessment with only the belief subscale variables included, the drop in overall variance accounted for was similar to that noted in the first comparison with the adoption category regressions (Table 14). Two variables achieved significance, with BAR and SEV together accounting for 34 % of the overall 35 % of the variance that could be explained by the belief subscales. As with all t h e other regressions performed, the F value showed the overa l l R square of the variable set to be significant at the p 0.00 1 level. Hypothesis 4 was thus accepted. Table 14: Sources of Variance From Health Belie f Model Belief Subscales in Exercise Behavior Categori e s. Variable R Square Change in Overall F R Square Value BAR 0.310 0.3 1 0* 65.9 6 *** SEV 0.342 0.03 2 3 7 .8 7 *** MOT 0.350 0.008 2 6 .11*** BEN 0.352 0.002 19.53*** SUS 0.352 0.000 15.55*** = p < 0.05 *** = p 0.001 (Final OF 5, 143) A variant of the analysis on exercise behavior categories was performed to identify the degree to w hich the beliefs variables could predict exercise behavior w hen

PAGE 128

118 only the two poles of behavior were used. The exercise assessment categories of Sedentary and Aerobic were selected to provide an example with more pronounced differences in the criterion categories. The outcome of this analysis is shown in Table 15. The proportion of variance explained was the highest of any of the observed R-squares obtained in the stepwise regressions performed. A pattern observed in the documenting of the predictive strengths of the independent variables in the study--that continued in this analysis--was the emergence of the barriers subscale as the principal predictor. Not much distribution of assessed variance among the other variables was noted in any of the regression analyses, with the severity subscale entering in as the second most significant variable in all but the instance of exercise adoption using the five subscales. Age appears to play a small role in prediction as one of the composite array of variables in the exercise adoption category. Table 15: Sources of Variance From Health Belief Model Belief Subscales in Antipodal Exercise Behavior Categories. Variable R Square Change in Overall R Square Value F BAR 0.468 0.468* 61.47*** SEV 0.500 0.032* 34.43*** MOT 0.509 0.009 23.51*** SUS 0.511 0.002 17.48*** BEN 0.511 0.000 13.78*** *=p < 0.05 *** = p 0.001 (Final OF 5 66)

PAGE 129

119 To note any change occurring in the regression results from the first administration to the second, the maximum R-squares obtained from the stepwise regressions were compared (Table 16). For the analyses, the beliefs variables obtained on the first administration of the questionnaire were regressed on the criterion variables obtained in both the first and second administrations. While the performance of the predictor variables dropped for the exercise adoption categories, it rose for both exercise behavior analyses performed. No change was noted in the order in which the predictor variables entered into the regression equation. Table 16: Change in Total R-square For Criterion Variables From First to Second Administrations Using Independent Belief Variables Obtained in First Administration. criterion Variable Adoption Categories Exercise Behavior Categories Antipodal Exercise Behavior Categories ***=p ,:S. 0.001 First Administration R-square 0.439*** OF 5, 172 0.345*** OF 5, 172 0.440*** OF 5, 90 Second Administration R-square 0.397*** OF 5, 143 0.352*** OF 5, 143 0.511*** OF 5, 66

PAGE 130

120 To test the hypotheses regarding the utility of the overall beliefs scale for predicting exercise category outcomes, discriminant analyses were performed using the measurements of the belief variables obtained on the study population. The performance of the beliefs variables in assigning subjects to the categorical dependent variables of exercise adoption category, exercise behavior category, and change category noted in the second administration of the questionnaire. The six adoption categories Immotive, Contemplator, Relapser, Past Exerciser, Recent exerciser, and (Dedicated) Long term exerciser were used for the first discriminant analysis. The four exercise behavior classifications of Sedentary, Minimal Activity, Active, and Aerobic were used for the second, and the third was conducted using the extremes of the exercise behavior classifications Sedentary and Aerobic as was noted above in the results of the regression analysis. The test of the predictive efficacy of the beliefs was conducted, but the number of subjects indicating a change in their exercise behavior (24, or 16% of all matched pairs) was judged to be low for a meaningful interpretation. All analyses were conducted using the DISCRIM package of SAS (Ray, 1982). The results of the discriminant analysis for the exercise adoption categories appear in Table 17, a matrix allowing "hit" and "miss" totals to be shown for each c e l l of predicted versus actual status for the 149 subjects

PAGE 131

A C T u A L 121 examined. The overall classification rate was 45 percent correct assignment for all subjects. This performance of the measure well exceeded the chance rate of 16.7 percent for each cell. The adoption categories where prediction was least accurate were Past Exerciser and Relapser. The results offer low but sufficient support for accepting Hypothesis 5. Table 17: Number of Observations and Percents Classified Into Exercise Adoption Categories. Predicted Category Total N 1 2 3 4 5 6 Immotive 10 2 2 2 2 0 (1) 55.6% 11.1% 11.1% 11.1% 11.1% (0) 18 Contemplator 5 12 2 1 0 2 (2) 22.7% 54.5% 9.1% 4.5% (0) 9.1% 22 Relapser 3 4 7 4 2 0 ( 3) 15.0% 20.0% 35.0% 20.0% 10.0% (0) 20 Past 6 8 3 7 8 4 Exerciser 16.7% 22.2% 8.3% 19.4% 22.2% 11.1% 36 (4) Recent 0 1 1 2 5 4 Exerciser (0) 7.7% 7.7% 15.4% 38.5% 30.8% 13 (5) Long Term 0 2 0 0 12 26 Exerciser (0) 5.0% (0) (0) 30.0% 65.0% 40 (6) n = l49 Prior probability, each cell=16.7% In the application of the discriminant analysis technique to the example of behavior, an improvement in the performance of the beliefs variables was noted, as shown in Table 18. The percentage of correct

PAGE 132

122 classification overall was 52 percent, compared to a chance rate of 25 percent. While all cells again exceeded chance, the Sedentary category showed distribution of actual subjects across all predicted categories. Table 18: Number of Observations and Percents Classified Into Exercise Behavior Categories. Predicted Category Total N 1 2 3 4 A Sedentary 8 9 6 4 (1) 29.6% 33.3% 22.2% 14.8% 27 C Minimal 6 21 4 1 T (2) 18.8% 65.6% 12.5% 3.1% 32 u Active 6 11 16 12 ( 3) 13.3% 24.4% 35.6% 26.7% 45 A Aerobic 5 1 6 33 L (4) 11.1% 2.2% 13.3% 73.3% 45 n = 149 Prior probability, each cell=25% As before when examining the ability of the model to account for overall variance, an analysis of the combin ed strength of the model variables in predicting two behavioral categories with little overlap was done to te st the accuracy with dichotomous classification. In Table 19, the correct classification rate overall was 82 percent, with the chance rate for two categories at 50 percent. Presented with these results and those of the preceding analysis, Hypothesis 6 was accepted.

PAGE 133

123 Table 19: Number of Observations and Percents Classified Into Antipodal Exercise Behavior Categories. Predicted Category Total N 1 2 A Sedentary 23 4 C (1) 85.2% 14.8% 27 T u Aerobic 9 36 A (2) 20.0% 80.0% 45 L n=72 Prior probability, each cell=50% As was noted above, the planned test of the model in predicting a change in exercise activity was hampered by a low number of subjects indicating a change. Examination of the data showed that, for those indicating a change, their report of activity level did not support the assertion. The data review also revealed that there were twenty-five additional changes in exercise activity, as assessed on the exercise/activity responses, that subjects did not record. It was decided to examine the capability of the model variables to predict the changes, producimg the results in Table 20. In this case, again with the rate of chance being 50 percent, the belief variables allowed a correct prediction rate of 75 percent. While this was seen as support for Hypothesis 7, the results must be heavily qualified.

PAGE 134

124 Table 20: Number of Observations and Percents Classified Into Self-Reported Change in Exercise Behavior Categories. Predicted Category Total N 1 2 A Decrease 3 2 C (1) 60.0% 40.0% 5 T u Increase 4 15 A (2) 21.1% 78.9% 19 L n = 24 Prior probability, each cell=50%

PAGE 135

CHAPTER 5 DISCUSSION The results of this study show support for the Personal Beliefs Questionnaire as an effective instrument for beliefs assessment and for the ability of the beliefs so measured to distinguish between different categories of subjects. Relative to other studies of this type (Jette et al., 1981; Wagner, 1983), the current effort resulted in an acceptable instrument that shows potential for application in exercise adoption progra m s. For research, at least, the findings warrant further employment and evaluation of the Personal Beliefs Questionnaire with populations of potential exercisers. The Comparative Efficacy of the Personal Beliefs Questionnaire Using other carefully conducted Health Belief Model instrument development projects as benchmarks (e.g. Champion, 1984), the Personal Beliefs Questionnaire had comparatively excellent internal consistency for its scales. Other studies of this type (Jette et al., 1981; Slenker et al., 1984) have recorded alpha coefficients from 0.30 to 0.92, with most values occurring in the 0.65--0.80 range. With the range for the current study from 0.83--0.94 for the final item groupings, the scale 125

PAGE 136

from 0.83--0.94 for the final item groupings, the scale groupings of items show a high degree of internal consistency. 126 Observed test-retest correlations evidenced good reliability as measured by test-retest correlations. Reported correlations from the above-cited studies occur in a range from 0.35--0.87, with most values observed between 0.65 and 0.83. The SEV subscale, with the most items removed in the development of the survey, showed the loss in a correlation that was substantially lower than its companion scales. These favorable findings are believed to result in part from the use of previously validated items or item structures wherever possible in instrument construction. The result achieved was considered impressive given that beliefs were being assessed. Beliefs have been demonstrated to be a dynamic psychological constructs that are subject to change from many different influences. One way in which the benefits of using previously validated items did not extend to the new alignment of items was in the subscales' achievement of independence from each other as assessed by intercorrelation. The failure of this hypothesis to be supported is based on the number of intercorrelations found to be significant in the analysis. Oth er studies where scales have been developed have observed higher scale intercorrelations then the

PAGE 137

127 current study, and those correlations have been noted without comment about their threat to the ability of the belief components to measure what they purport to measure (Maiman et al., 1977; Given, Given, Gallin, & Condon, 1983). Many other studies consulted report no computations of scale intercorrelations other than what may be inferred from factor analyses. This lack of attentiveness to scale independence appears to be a weakness in the method generally utilized in Health Belief Model research for the evaluation of instruments .. Assessing the Criterion Validity of the Personal Beliefs Questionnaire Factor analysis, while providing some support for the five beliefs constructs as independent entities, also detects and expresses some subscale overlap, in a manner different from the above test of intercorrelation. The five most significant scales, as determined by scree plotting and judgement of the overall eigenvalues contributed by the factors, clustered around content largely supportive of the Health Belief Model (see page 111). Factor 1 included items addressing what was summarized as "aerobic exercise approach/avoidance". The items in this cluster were from opposite poles of a pair of related constructs--perceived benefits and perceived barriers--and addressed expectations resulting from personal action/exercise. They are a good operational illustration of the personal fields referred to by

PAGE 138

128 Rosenstock (1974a), in which a person moves towards positive fields and away from negative ones. The expectations specified were notable for the positive and negative features associated with the stimulus of aerobic exercise. Their co-occurrence in one factor (with a like item from the health motivation subscale) appears to illustrate a continuum being assessed by the two constructs, with each construct making up one of the poles. The unitary association of these separate constructs has been noted in other Health Belief Model studies (Cummings, Jette, and Rosenstock, 1978). Factor 2 was labeled simply "health threat" and was comprised of items from the severity subscale (with one susceptibility item) addressing the seriousness of specific conditions such as cancer, heart trouble, and the flu. Again, the correspondence of this factor to the field theory roots of the Health Belief Model is not hard to discern. Factor 3, "behavioral self-assessment of health protection" contained items from the health motivation subscale that assessed specific behavioral evidence of health motivation, as opposed to attitudinal expression. The items' content reflected a personal stock-taking of behavioral commitment to health. The fourth factor was titled "likelihood of everyday illness" and was composed of items from the susceptibility subscale. This was the factor corresponding most closely to the hypothesized nature of the construct

PAGE 139

129 it represents. The illnesses represented in the items were the less serious ones, such as flu, strep throat or a cold. Factor 5 clustered around "specific fears about injury while exercising", and was made up of items from the barriers subscale. These questions both addressed injuries that could occur while exercising, fears about which could serve as barriers to performing the exercise. While the results of the factor analysis were not a pure, item-by-item confirmation of the model components, the nature and configuration of the clusters was seen as highly supportive of the five beliefs subscales as independent constructs. When used to evaluate the scales in the Health Belief Model for their independence and ability to measure separate, discrete constructs, factor analysis findings have been quite variable. Most studies report the model to be supported by the findings of factor/principal component analysis, with occasional indications of revision suggested (Wagner, 1983; Slenker et al., 1984). Model performance is very dependent on the behavior being assessed, with a much different configuration likely for sick-role or illness behaviors than for preventive behavior like aerobic exercise. Construct validity gained a second confirmation from the regression analyses performed. An additional benefit from the prior-validation strategy of item selection may

PAGE 140

130 have been realized in the figures achieved accounting for the variance in the criterion variables. The range as determined by R-squares was from 35% to 51% of the overall variance, comparing favorably to other Health Belief Model studies (Jette et al., 1981; Wagner, 1983). The major portion of variance remains unaccounted for, however, even with fairly good comparative efficacy by the Personal Beliefs Questionnaire. It was beyond the scope of this study to successfully assess and control for the number of factors associated with exercise outcomes that are listed on page 50 (Powell, 1988); what may be safely concluded is that beliefs play a significant role in differentiating aerobic exercisers from nonexercisers. In observing the overall variance that the instrument was able to account for in regressions on the kinds of exercise categories posited for this study, it is clear that some major sources of variance were not included in the universe of variables sampled. The test of the demographic variables likely to affect exercise outcomes revealed no significant relationship with the outcome variables. With the exercise adoption categories age did enter third into the regression model, accounting for approximately one and a half percent of the overall variance. Examination of the correlation matrix of subscale total scores, demographic variables and dependent variables showed no significant association for age with the second reporting of exercise

PAGE 141

131 adoption category. The relationship between age and adoption category that was noted was a negative correlation, showing that younger respondents tended to be in higher categories. It should be emphasized that the relationship was slight. Weight, a variable often associated with peoples' exercise habits, did not correlate significantly with outcome variables assessed in this study. The observed change in total variance accounted for in the criterion variables from the first administration to the second was found to be as expected for the exercise adoption categories, with a slight decrement noted. Similar findings were obtained in one of the few repeated measures reported in Health Belief Model research (Becker et al., 1977). The unexpected result was the change recorded in the amount of variance accounted for in the behavioral outcome variables. Especially for the antipodal categories of Sedentary and Aerobic, the belief variables showed a marked upward difference in predictive potency when the retest categories were used in a regression that included initially obtained beliefs. Whether this is evidence of causality on the part of observed beliefs or the result of attrition of subjects from the sample between the administrations of the questionnaire is not clear.

PAGE 142

132 In applications of discriminant analysis Afifi & Clark (1984) recommend that the number of variables used in predicting class be greater by two than the number of class-cases being predicted. While all of the analyses performed met this qualification, the result for the category where respondents noted an increase or decrease in activity must be considered spurious. A regression analysis performed on the change category showed no predictor variables accounting for a significant portion of the variance, with belief variables accounting for only eighteen percent of the total variance. With documentation of a relationship lacking, the seemingly good prediction rate is more likely "noise". Slenker (1984), when applying her measure that assessed beliefs about jogging, found a "hit" rate of 92 percent when distinguishing between joggers and nonexercisers. Using the separated categories of Aerobic and Sedentary as the classification variables, the beliefs assessed in the Personal Beliefs Questionnaire were able to correctly classify 82 percent. Given the broader definition of exercise chosen for this study, this was regarded as a good performance of the instrument in distinguishing the categories. Similarly, for the full range of exercise behavior and for the exercise adoption categories, the instrument performed well when compared to the chance rate.

PAGE 143

133 A less sophisticated assessment of the instrument's performance was apparent from a review of the overall total scores obtained when plotted against the antipodal categories of Aerobic and Sedentary. With all items keyed to produce a low score if expected to be associated with an "exercise" outcome, high for non-exercisers, the range for the total scores was from 102 to 312. The median score for the outcomes observed was 171. For the sedentary group, 7 scored below, 20 above the 171 score. Within the aerobic exerciser group, 28 scored below, 17 above 171. These observations demonstrate the nature and meaning of the distribution, showing that a higher score is more likely to be associated with a non-exerciser, a lower score with an exerciser. The distribution of scores was conspicuously more spread out for the aerobic category than for the sedentary group, when one outlier was discarded. Range for the aerobic group was 126, from 102 to 228, while the sedentary group range was 65, from 158 to 223. All 20 of the observations with scores of 153 or lower were in the aerobic category. Thus it appears that threshold level in overall scores of 150 or below could predict an aerobic exerciser very reliably, while interpreting scores at the middle and upper parts of the range would be far more ambiguous.

PAGE 144

Limitations Affecting the Results of This Study 134 In applying the learnings from this study to other situations, the limitations inherent in the sample and execution of the study will need to be taken into account. The sample population was chosen to assure members of the sample would be over the age of 25, and the goal was certainly realized. Only one of the questionnaires had to be discarded due to the age of the respondent. Those sampled, as can be readily seen on page 92, were nearly all Caucasian/White, very well educated, and older than typical social science research samples. The sample was also not random, as almost all of the available pool was utilized. These characteristics make it representative of most college and university faculties, but limit generalizability to other populations. While this may control for some factors and allow for firmer conclusions about the use of such an instrument with a sample characterized by such an advanced level of education, older average age, and limited diversity, the similar performance of the instrument with a more general population is not assured. The sample was also likely skewed toward a higher average income, although this information was not gathered. Future work would be best served by a more ethnically diverse, younger adult sample with a more varied educational background. This would be

PAGE 145

135 keeping with the practice of most previous Health Belief Model investigations. Response rate was less than expected; while deemed acceptable for purposes of this study, it made for a smaller than desired pool of subjects with which to evaluate an instrument. Data were entered in the order received to enable response delay effects to be isolated if they occurred; none were identified. Item attrition that occurred during instrument evaluation was also less than expected, such that the desired ratio for doing factor analysis was not achieved in the current study. A recommended factor analysis ratio is an n of at least four to five times the number of variables being analyzed (Gorsuch, 1983). The ratio achieved in this study was 3.38, which makes the results of the factor analysis employed in the study meaningful, but possibly suspect. The sample size was considered adequate for the other analyses. An additional factor affecting the soundness of the study is introduced by virtue of the fact that there is some association between the scales. This makes for multicollinearity; when variables are associated they covary in ways that make analysis--especially regression analysis--harder to interpret. The level of association between the independent variables in this study was sufficiently low that multicollinearity was not considered

PAGE 146

136 a major factor, but neither is there complete freedom from its influence. Replication of this study would identify whether this was a sample or instrument effect. The component of exercise assessment encountered some problems in the execution of this study; in the literature on exercise assessment there are indicators that such problems have not yet been resolved even in very sophisticated studies (Perkins & Epstein, 1988). One of the factors noted in reviewing the data was what appeared to be a social desirability effect. Kirscht (1983) comments that self-reports of socially desirable behavior are over and socially undesirable behavior are under actual levels. In the present study, it was noted that fifteen of the subjects reported they were in Category 6 (Long term exerciser) on the first administration of the questionnaire when their exercise self-report did not concur, nine did so on the second. The converse situation, not reporting aerobic exercise when the exercise questions indicated the level was indeed satisfying the American College of Sports Medicine (1978) minimums, occurred seven times in the first administration, six in the second. In accounting for discrepancies observed between the exercise adoption and the actual behavior reported on the measures, there appears to be an effect from the perceived desirability of being an aerobic exerciser.

PAGE 147

137 Participation in exercise was high as reported by the sample population. Where Powell (1988) reports that nine to twenty percent of adults exercise at an aerobic level, thirty percent of the subjects returning both questionnaires indicated exercise activity that fit aerobic exercise criteria. When exercise levels that could be considered active (greater than twenty minutes per session and three or more times per week) were considered, the level for the sample population rose to sixty percent, as opposed to a range of twenty to fifty-five percent reported in the Powell (1988) chapter. Powell reports that males tend to exercise more than females, persons of higher socioeconomic status exercise more than those of lower socioeconomic status, and younger persons exercise more than do older persons. Given the impact of these contributing factors, the knowledge that two of the three factors characterize this sample leads to the likelihood that they contribute to the higher reported rates, as does the availability of more leisure time for this population. While over-reporting of exercise participation has been observed as a drawback of the self-report method (Washburn & Montoye, 1986), the higher levels of participation reported by this sample are believed to be accurate for a study of this type. There was a high degr e e of correspondence between the adoption categories reported and the exercise participation as noted in the type,

PAGE 148

138 duration, and frequency blanks on the survey. The campus surveyed has an extensive, well-attended program for exercise and wellness that may contribute to the abnormally high participation rates observed. Implications for Future Research on the Precursors to Aerobic Exercise Behavior Peoples' perceptions about barriers to exercising appear to be emerging as one of the major factors contributing to their decision-making and behavior in the realm of aerobic exercise. Slenker et al. (1984) noted similar findings in differentiating between joggers and nonexercisers. In the case of exercise, then, it seems that beliefs about perceived barriers play a central role in whether someone ultimately adopts an exercise program, whether in the specific area of jogging or the more general, multiple exercise criterion examined in the present study. It was believed that a study of this sort might perform less well given the much broader definition of exercise employed. While the scores found were less impressive than the Slenker et al. (1984) study, they compare favorably to most other applications of this methodology to health-protective behavior (Silver, 1983; Champion, 1984; Chen & Tatsuoka, 1984; O'Connell et al., 1985). Unlike some other recent studies investigating the role of Health Belief Model beliefs in the determination of exercise status (e.g. Slenker et al. 1984), the present

PAGE 149

139 study found very little predictive power inherent in the belief components other than perceived barriers. Perceived severity played a minor role in accounting for the variance in the criterion categories or as weight in prediction equations. This confirms earlier observations by Rosenstock (1974a) that one of the driving forces causing people to act--that the Health Belief Model approach to assessment identifies--is an avoidance orientation in response to a perceived threat. At least in regard to exercise, a way of weighting the belief variables so that more information from the other areas may be contributed is recommended to overcome what appears to be an almost unidimensional way of functioning. Perceived barriers has been shown to be the dominant belief affecting prediction and accounting for the most variance in studies o f exercise / preventive behavior. While perceived susceptibility and perceived severity have contributed much in past investigations of illness-related and sick-role behavior, their role becomes limited in preventive behavior investigations. Either modification of the model or its employment in the area of exercise behavior is indicated. Scale refinement would be possible if this method were to be used in conjunction with programs that deal with exercise adoption. It may be possible to obtain needed information about critical

PAGE 150

beliefs such as perceived barriers without using the entire instrument to assess all beliefs. 140 Along with the effort to apply the Health Belief Model to the area of general aerobic exercise, this study was also an attempt to develop and employ a standard method for use in studies of health-related behavior. It is believed that the employment of items from the pool already tested and in use helped to deliver a more reliable and valid instrument, as well as results that may be more easily compared to other studies. One of the originators of the model has repeatedly called for such an effort so that the value of succeeding studies would be increased (Janz & Becker, 1984). For future studies addressing exercise behavior, whether or not they are in the realm of attitude/belief research, a standard, usable, and universal protocol for assessing exercise is needed. While this study used self-report around a widely-recognized standard (American College of Sports Medicine, 1978), the comparative value is compromised by the unique nature of the scales devised. Just as many different instruments have been employed in Health Belief Model research, there has also been a plethora of different kinds of exercise measures, each slightly different (Perkins & Epstein, 1988). Until practical approaches are made feasible and reliable for

PAGE 151

objective measurement, a standard for self-report is indicated. 141 The adoption-stage schema for exercise based on the Prochaska and DiClemente (1983) behavior change theory described on page 38 seemed to gain support in this study, in that subjects presented with the descriptors for the categories self-assigned to an almost uniform distribution across the stages, a finding duplicating the other known instance of this schema to exercise (Sonstroem, 1987). Additionally, the predictor variables did moderately well at differentiating between the stages, implying progressively greater investment in aerobic exercise. This could mean a greater level of either behavioral or cognitive activity separates subjects in the category levels. The greater amount of variance accounted for with the six adoption stages as compared to the four exercise behavior categories may be due, in part, to the additional psychological mechanisms or elements of change assessed by the instrument. The hypothesized process of exercise adoption is one that is psychosocial in nature, so an instrument that determines beliefs about health and exercise would be expected to measure the gradations of change occurring in that process. In attempting to distinguish categories that entail the assessment of mental states or beliefs, it appears that there are too many unmeasured variables involved to

PAGE 152

142 make any authoritative, integrative statements. Available statistical tools are still limited when looking to them for results that can provide the foundation for models that take into account all variables in an interactive fashion. In the study of behavior, whatever the behavior or the context, valuable information is lost without a method that can at least crudely assess representative variables across a wide spectrum. Such an undertaking will be massive in scope, and will demand considerable commitment on the part of its research subjects. Such efforts are necessary, however, to finally dispense with the qualifications that must be placed on efforts like the present one. Again, a reference to the list on page 50 (Powell, 1988) will show that about one third were assessed in some way in this study. An illustration of the variables missed unexpectedly encountered in the survey. On the response blank provided subjects to cite reasons for their change in exercise behavior, three specified completing the first survey as the reason. Eight mentioned more favorable weather (the project was conducted in the spring). This serendipity illustrates the sorts of effects that were not anticipated or controlled for in the study, but were identified as having an impact. An issue that received no definitive answers from this study was that of causality, or whether the beliefs measured caused the subsequently measured behavior. In this study, an improvement in the performance of the

PAGE 153

143 instrument in predicting exercise behavior was noted in the time that elapsed from the first administration to the second. In drawing conclusions about the impact of the variables on the subsequent behavior, while there are some positive indicators, the time period spanned was still too short. Future efforts will need to adopt a longitudinal or prospective format that covers a greater time period than did the current study. It does seem that the simpler the criterion question--e.g. "exercise or no exercise?", as opposed to "what level of exercise?"--the more definitive the answer. Conclusion This investigation has served to affirm the beliefs-behavior link for an expanded definition of aerobic exercise, and to provide tentative evidence for a causal relationship of beliefs to exercise behavior. Limitations in the study preclude generalizability to a wide range of populations, but the results do parallel other investigations into the contribution of health beliefs to preventive behavior. Given the body of work showing the efficacy of intervention studies in changing behavior (Becker et al., 1977; Beck & Lund, 1981), an intervention study involving a population of would-be exercisers seems both more possible and more desirable as a consequence of this study. For the researcher, the perceived barriers are the

PAGE 154

144 time such a study would take and the work of keeping subjects involved whatever their exercise outcome. There have been recommendations for such research on the part of model proponents (Kirscht, 1974; Janz & Becker, 1984), but few investigations attempting such a methodology. Researchers outside the traditional bailiwick of Health Belief Model research have published research supporting the beliefs-attitude-behavior link, adding more evidence for the validity of the model and its mechanisms {Turk, Rudy, & Salovey, 1984). With the amount of research the Health Belief Model has generated, most of it supportive of the relationship between beliefs and behavior, the step clearly indicated now is identifying ways the learnings may be applied. Psychologists, health educators, and other change agents need to find or generate ways to exploit the clear impact that peoples' perceived barriers to behavior change have on their ability to succeed at lifestyle change involving the adoption of preventive behavior. It seems that such learnings may also be applicable to cessation of behavior, as in the addictions, in the same way that they may apply in the case of exercise. Given the gains possible in public health with even incremental changes in exercise norms and behavior, the need is transparent. Kiesler and Morton (1988) delineate palpably the ways in which psychology can play a significant role in the public interest, in affecting the health of people and

PAGE 155

145 expanding the role of psychology. Psychologists and health professionals who work with people seeking help with lifestyle change have a potentially powerful tool whose mechanisms will become more apparent with its increased use. Experimentation and commitment are needed to validate the tool and to identify the best forms. Given that groundwork, another avenue for change w ill be made available.

PAGE 156

APPENDIX A CONTACT LETTER FOR QUESTIONNAIRE Date Dear My name is Larry Gage, and I am a counselor on the staff of the Psychological Counseling Services at Trenton State College. I am also a doctoral candidate in counseling psychology at the University of Florida, and as such I am contacting you to request your participation in my dissertation project, a study of peoples' thoughts about health and exercise. Your participation will help a colleague with an important (to him!) project, and afford you two chances to win substantial cash prizes. What I am requesting are two completions of the enclosed survey, one now and one a month after I receive the first from you. Each survey takes ten to fifteen minutes to complete, with the first reply entering you in a drawing that awards the winner fifty dollars, the second one hundred dollars. Completing both surveys, therefore, makes it possible to win up to $150. An addressed envelope is enclosed for your first survey, with another envelope for a reply-consent form that allows your survey responses to be completely anonymous. Your surveys will be identified by the last four digits of your home telephone number. Please, take a few moments to send in your first reply. The reply-consent form enters you in the drawing and lets me know who is to receive the follow-up. Yes, it is possible to send the reply form back and not complete the survey; I am using an honor system. That is the risk I take to insure anonymity for your responses. If anything seems unclear or you have questions about the procedure, please feel free to contact me at 771-2247. The enclosed items are self explanatory, returnable in the attached envelopes. Simply complete each item and drop in campus mail. Please, do it soon, so it won't get misplaced. I will be grateful for your participation. Thanks! Sincerely, Larry Gage 146

PAGE 157

APPENDIX B REPLY-CONSENT FORM INFORMED CONSENT You are being asked to provide some information about yourself on the enclosed survey, and to complete a similar survey again in one month. The surveys contain questions about your opinions of health and exercise. The purpose of the surveys is to learn about peoples' thoughts about exercise and health, and to determine their pattern of exercise. Once obtained, this information will be valuable for planning exercise programs. Your return of each of the surveys qualifies you for cash prizes of fifty dollars for the first and one hundred dollars for the second. All participants are eligible for both drawings. Odds of winning are approximately one in two hundred and fifty. Information you give will be strictly confidential and anonymous. You are asked on the survey ans w er sheet to provide the last four digits of your home telephone number to enable your responses to be paired. At no time will your name become paired with the data sought in this process, as this certification of participation is under separate cover from your response to the surveys. Voluntary Consent I have read and understand the procedure described above. I freely consent to participate. I understand that I am not required to participate and that I may stop participation at any time. Name (please print) ---------------------Address ----------------Office Telephone SIGNATURE Larry Gage Psychological Counseling Services 1926 Pennington Rd. Hillwood Lakes CN 4700 Trenton, New Jersey 08650-4700 Telephone (609) 771-2247 147 Date

PAGE 158

APPENDIX C PERSONAL BELIEFS QUESTIONNAIRE Important! Please enter the last four digits of your home telephone number here: _______ PERSONAL BELIEFS QUESTIONNAIRE The following brief questionnaire is designed to assess beliefs you have about health, sickness and exercise. Everyone has their own set of beliefs; there are no right or wrong answers. Please respond quickly to each item, with the response that first comes to mind. Your answers to the questionnaire will be anonymous, thus completely confidential. Please answer as honestly as possible, answering in the way that best describes your current beliefs about each question. For each item circle the number that comes closest to capturing your belief for that item. Notice that you may respond using the answer at either end of each scale provided (e.g. Quite concerned; Unconcerned) or any of the five responses in between. An answer of "4" is a balance point between the two ends; in the example above it would mean you were midway between concerned or unconcerned about the issue in question. Sometimes the choice of responses change from one question to the next, so please take care to use the scale paired with the particular question you are answering. Section 1. 1. Some people are more concerned about their health while others are not as concerned. How concerned are you about your own health? Quite concerned<----------------------->Unconcerned 1 2 3 4 5 6 7 2. Some people are more concerned about the chance of getting sick, while others are not as concerned. How concerned are you about the chance of getting sick? Quite concerned<------------------------>Unconcerned 1 2 3 4 5 6 7 3 How good a job are you doing in taking care of your health right now? Excellent<---------------------------------->Poor 1 2 3 4 5 6 7 148

PAGE 159

4. How likely is it that you will try to do a better job of taking care of your health in the future? Very likely<------------------------>Very unlikely 1 2 3 4 5 6 7 5. What best describes the number of individual behaviors or precautions that you engage in to insure your good health? Many<----------------------------------->None 1 2 3 4 5 6 7 6. I frequently do things to improve my health. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 7. I search for new information related to my health. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 8. Although I am concerned about my health, there are other things that are more important to me. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 9. In general, the health habits I practice are: Good<--------------------------------->Bad 1 2 3 4 5 6 7 Section 2. For the following ten questions, please respond according to the scale below. Very likely < -------------------------> Very unlikely 1 2 3 4 5 6 7 Record the number indicating your response in the blank to the right of each condition listed below. How much chance is there that you could ever get: 10. asthma 11. pneumonia 12. cancer 13. anemia 14. the flu 20. Compared easily. Strongly 1 Response Response 15. heart trouble 16. strep throat 17. a cold 18. a cavity 19. a stroke to other people my age, I get sick more agree < ------------------> Strongly disagree 2 3 4 5 6 7 149

PAGE 160

21. How likely do you think it is, in general, that you will get some sort of illness? Very likely<------------------------->Very unlikely 1 2 3 4 5 6 7 22. Compared to other people my age, I am more likely to have to spend time in bed due to my condition. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 23. I worry a lot about contracting some disease that will prevent me from pursuing normal activities. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 24. My physical health makes it more likely that I will be troubled by disease. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 Section 3. For the following ten questions, please respond according to the scale below. Very serious<-------------------->Not at all serious 1 2 3 4 5 6 7 Record the number indicating your response in the blank to the right of each condition listed below. How serious would it be to get the following condition within the next twelve months? Response Response 25. asthma 30. heart trouble 26. pneumonia 31. strep throat 27. cancer 32. a cold 28. anemia 33. a cavity 29. the flu 34. a stroke 35. Whenever I get sick, it seems to be very serious. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 36. Lacking energy often, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 37. Developing high blood pressure or cardiovascular problems, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 150

PAGE 161

38. Developing nervousness or stress problems, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 39. Becoming overweight, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 Section 4. For the following sections, the term "regular exercise" refers to aerobic exercise of at least 20 minutes duration, at least 3 times a week. Aerobic exercise is a rhythmic, continuous, sustained exercise such as running, rowing, rope-skipping, bicycling, or swimming that produces an elevated heart rate. 40. How much do you feel maintaining a program of regular exercise will help improve your overall health? Help a lot<-------------------------->Not help at all 1 2 3 4 5 6 7 41. How well do you feel a program of regular exercise will help prevent illness? Help a lot<-------------------------->Not help at all 1 2 3 4 5 6 7 42. A program better. Strongly 1 of regular exercise would make my heart work agree<------------------->Strongly disagree 2 3 4 5 6 7 43. Regular exercise will help me to tone up my muscles. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 44. A program of regular exercise would help me gain better posture. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 45. A program weight. Strongly 1 of regular exercise would help me lose agree<------------------->Strongly disagree 2 3 4 5 6 7 46. Regular exercise would really help my appearance. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 151

PAGE 162

Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more/week. 47. Regular exercise would really help increase my energy level. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 48. Participating in a regular exercise program would increase my self respect. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 49. Regular exercise would really help me keep a good outlook on life. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 50. Participating in regular exercise would help me do better thinking. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 51. A program of regular exercise could benefit me by providing friendship and socialization with other exercisers. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 Section 5. Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more/week. 52. Sometimes I worry that regular exercising can cause health problems. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 53. Sometimes I am concerned that regular vigorous exercising would cause me to have a heart attack. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 54. I am concerned about the possibility of being injured while maintaining a regular exercise program. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 Regular Exercise = Aerobic, 20 minutes (or more) per session, 3 times or more/week. 152

PAGE 163

55. A regular program of exercise would likely make me stiff and sore. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 56. How difficult would you say it is (would be) for you to maintain a regular program of exercise? Very difficult<------------------------->Very simple 1 2 3 4 5 6 7 57. How much do you feel maintaining a regular program of exercise interferes (would interfere) with your normal activities? Interfere a lot<-------------------->No interference 1 2 3 4 5 6 7 58. A regular schedule. Strongly 1 program of exercise is difficult for me to agree<------------------->Strongly disagree 2 3 4 5 6 7 59. Sometimes it seems to me that maintaining a regular program of exercise is more trouble than is worth. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 60. A preference to do other things with my time could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 61. Lack of desire or interest could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more/week. 62. Unsuitable weather conditions could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 153

PAGE 164

Section 6 Your Exercise Pattern In response to this question we would like to determine as specifically as possible what category you would place yourself in according to your current status regarding exercising. Please note that it is important for you to respond with your actual level of participation in exercise, as opposed to what you may wish it to be. "Exercise" is defined for purposes of this question as aerobic exercise of at least 20 minutes duration, at least 3 times a week. Aerobic exercise is a rhythmic, continuous, sustained exercise such as running, rowing, rope-skipping, bicycling, or swimming that produces an elevated heart rate. Please place a check next to the category that describes your current status. Important: If your activity level is significant but does not conform to the definition of regular exercise, be sure to answer Section 7. Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more/week. Category i You are in this category if you are not exercising and have not thought about beginning an exercise program at all in the previous year. Cateqory You are in this category if you have been seriously thinking about how to begin a regular exercise program in the past year. Category~ You are in this category if have actively tried but not succeeded (started briefly, for less than six months, then stopped) in beginning an exercise program during the past year. Category~ You are in this category if you are not currently in a program of regular exercise, but you have maintained a pattern of regular exercise for over six months at any time in the past five years. Category 2 You are in this category if you have begun exercising regularly recently and have continued for a period of !!IL to nine months. Category You are in this category if you have maintained a pattern of regular exercise for a period of more than eighteen months. Category 7 You are in this category if you are a Category 6 person who exercises daily and for periods of longer than 45 minutes. 154

PAGE 165

If you indicated you are now pursuing a program of regular exercise, please describe the type of exercise; also indicate your usual duration and frequency: Type _____________________________ Duration Minutes/Session. --------Frequency Times/Week. -------Section 7 Other Activity Many people who are not aerobic exercisers may do other forms, such as walking, as part of their daily lifestyle. Such activity may be intentional or part of a job; it may be of short or long duration. If you engage in activity that represents exercise, but it is not described by the definition of regular aerobic exercise in Section 6, please detail the activity here: Type ____________________________ Duration Minutes/Session. --------Frequency ________ Times/Week. Section 8 About You 1. Age: _____ 2. Marital Status (circle one): Single Married Divorced Separated Living Together 3. Terminal Degree (circle one): Bachelors Masters Doctorate Postdoctorate 4. Body Weight: lbs. ------This concludes the questionnaire. Thank you for taking the time to complete it. Simply place it in the attached envelope marked "Survey Return" and drop in Campus Mail. Be sure you also return the Reply/Consent form in its envelope so you will be eligible for the drawing. And thank you for participating!! 155

PAGE 166

APPENDIX D SECOND ADMINISTRATION OF QUESTIONNAIRE Important! Please enter the last four digits of your home telephone number here: _______ PERSONAL BELIEFS QUESTIONNAIRE The following brief questionnaire is designed to assess beliefs you have about health, sickness and exercise. Everyone has their own set of beliefs; there are no right or wrong answers. Please respond quickly to each item, with the response that first comes to mind. Your answers to the questionnaire will be anonymous, thus completely confidential. Please answer as honestly as possible, answering in the way that best describes your current beliefs about each question. For each item circle the number that comes closest to capturing your belief for that item. Notice that you may respond using the answer at either end of each scale provided (e.g. Quite concerned; Unconcerned) or any of the five responses in between. An answer of "4" is a balance point between the two ends; in the example above it would mean you were midway between concerned or unconcerned about the issue in question. Sometimes the choice of responses change from one question to the next, so please take care to use the scale paired with the particular question you are answering. Section 1. 1. Some people are more concerned about their health while others are not as concerned. How concerned are you about your own health? Quite concerned<----------------------->Unconcerned 1 2 3 4 5 6 7 2. Some people are more concerned about the chance of getting sick, while others are not as concerned. How concerned are you about the chance of getting sick? Quite concerned<------------------------>Unconcerned 1 2 3 4 5 6 7 3. How good a job are you doing in taking care of your health right now? Excellent<---------------------------------->Poor 1 2 3 4 5 6 7 156

PAGE 167

4. How likely is it that you will try to do a better job of taking care of your health in the future? Very likely<------------------------>Very unlikely 1 2 3 4 5 6 7 5. What best describes the number of individual behaviors or precautions that you engage in to insure your good health? Many<----------------------------------->None 1 2 3 4 5 6 7 6. I frequently do things to improve my health. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 7. I search for new information related to my health. Strongly agree<------------------> Strongly disagree 1 2 3 4 5 6 7 8. Although I am concerned about my health, there are other things that are more important to me. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 9. In general, the health habits I practice are: Good<--------------------------------->Bad 1 2 3 4 5 6 7 Section 2. For the following ten questions, please respond according to the scale below. Very likely < -------------------------->Very unlikely 1 2 3 4 5 6 7 Record the number indicating your response in the blank the right of How much 10. asthma 11. pneumonia 12. cancer 13. anemia 14. the flu 20. Compared easily. each condition listed below. chance is there that you could ever get: Response Response 15. heart trouble 16. strep throat 17. a cold 18. a cavity 19. a stroke to other people my age, I get sick more to Strongly 1 agree<------------------->Strongly disagree 2 3 4 5 6 7 157

PAGE 168

21. How likely do you think it is, in g e n e ral, th a t y o u will get some sort of illness? Very likely<----------------------->Very unlik el y 1 2 3 4 5 6 7 22. Compared to other people my age, I am m o re l ikely to have to spend time in bed due to my condi t i o n. Strongly agree<-------------------> St r o ngly d isagree 1 2 3 4 5 6 7 23. I worry a lot about contracting s o m e disea se that will prevent me from pursuing normal activitie s Strongly agree<------------------->Str o ngly disagree 1 2 3 4 5 6 7 24. My physical health makes it more likely that I w i ll be troubled by disease. Strongly agree<------------------->Str o ngly disagree 1 2 3 4 5 6 7 Section 3. For the following ten questions, please respond according to the scale below. Very serious<-------------------->Not at all serious 1 2 3 4 5 6 7 Record the number indicating your response in the blank to the right of each condition listed below. How serious would it be to get the following condition within the next twelve months? 25. asthma 26. pneumonia 27. cancer 28. anemia 29. the flu 35. Whenever Strongly 1 Response Response 30. heart trouble 31. strep throat 32. a cold 33. a cavity 34. a stroke I get sick, it seems to be very serious. agree<------------------->Strongly disagree 2 3 4 5 6 7 36. Lacking energy often, for me, would be: Very serious < ------------------->Not at all serious 1 2 3 4 5 6 7 37. Developing high blood pressure or cardiovascular problems, for me, would be: Very serious < ------------------> Not at all serious 1 2 3 4 5 6 7 158

PAGE 169

38. Developing nervousness or stress problems, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 39. Becoming overweight, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 Section 4. For the following sections, the term "regular exercise" refers to aerobic exercise of at least 20 minutes duration, at least 3 times a week. Aerobic exercise is a rhythmic, continuous, sustained exercise such as running, rowing, rope-skipping, bicycling, or swimming that produces an elevated heart rate. 40. How much do you feel maintaining a program of regular exercise will help improve your overall health? Help a lot<-------------------------->Not help at all 1 2 3 4 5 6 7 41. How well do you feel a program of regular exercise will help prevent illness? Help a lot<-------------------------->Not help at all 1 2 3 4 5 6 7 42. A program better. Strongly 1 of regular exercise would make my heart work agree<------------------->Strongly disagree 2 3 4 5 6 7 43. Regular exercise will help me to tone up my muscles. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 44. A program of regular exercise would help me gain better posture. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 45. A program weight. Strongly 1 of regular exercise would help me lose agree<------------------->Strongly disagree 2 3 4 5 6 7 46. Regular exercise would really help my appearance. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 159

PAGE 170

Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more/week. 47. Regular exercise would really help increase my energy level. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 48. Participating in a regular exercise program would increase my self respect. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 49. Regular exercise would really help me keep a good outlook on life. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 50. Participating in regular exercise would help me do better thinking. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 51. A program of regular exercise could benefit me by providing friendship and socialization with other exercisers. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 Section 5. Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more / wee k 52. Sometimes I worry that regular exercising can cause health problems. Strongly agree<------------------->Strongly d i sag ree 1 2 3 4 5 6 7 53. S ometimes I am concerned that regular vigorous exe r ci sing would c a u s e me to have a h e art atta ck St rongly agr ee< ------------------> Strongly di sagree 1 2 3 4 5 6 7 54. I am c o nce rn e d a bout the possibi l it y o f be i ng i n jured w h ile mai n tai n i ng a r e gular exercis e pro g r am S t r o ng ly agree< --------------> St r ongly d isagree 1 2 3 4 5 6 7 Regular Exercise= Aerob ic 20 m i nut es ( or mo r e) per sessi on, 3 times or more/week 160

PAGE 171

55. A regular program of exercise would likely make me stiff and sore. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 56. How difficult would you say it is (would be) for you to maintain a regular program of exercise? Very difficult<------------------------->Very simple 1 2 3 4 5 6 7 57. How much do you feel maintaining a regular program of exercise interferes (would interfere) with your normal activities? Interfere a lot<-------------------->No interference 1 2 3 4 5 6 7 58. A regular schedule. Strongly 1 program of exercise is difficult for me to agree<------------------->Strongly disagree 2 3 4 5 6 7 59. Sometimes it seems to me that maintaining a regular program of exercise is more trouble than is worth. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 60. A preference to do other things with my time could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 61. Lack of desire or interest could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more/week. 62. Unsuitable weather conditions could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 161

PAGE 172

Section 6 Your Exercise Pattern In response to this question we would like to determine as specifically as possible what category you would place yourself in according to your current status regarding exercising. Please note that it is important for you to respond with your actual level of participation in exercise, as opposed to what you may wish it to be. "Exercise" is defined for purposes of this question as aerobic exercise of at least 20 minutes duration, at least 3 times a week. Aerobic exercise is a rhythmic, continuous, sustained exercise such as running, rowing, rope-skipping, bicycling, or swimming that produces an elevated heart rate. Please place a check next to the category that describes your current status. Important: If your activity level is significant but does not conform to the definition of regular exercise, be sure to answer Section 7. Regular Exercise= Aerobic, 20 minutes (or more) per session, 3 times or more/week. Category i You are in this category if you are not exercising and have not thought about beginning an exercise program at all in the previous year. Cateoory You are in this category if you have been seriously thinking about how to begin a regular exercise program in the past year. Category~ You are in this category if have actively tried but not succeeded (started briefly, for less than six months, then stopped) in beginning an exercise program during the past year. Category~ You are in this category if you are not currently in a program of regular exercise, but you have maintained a pattern of regular exercise for over six months at any time in the past five years. Category 2 You are in this category if you have begun exercising regularly recently and have continued for a period of 1liL to nine months. Category Q You are in this category if you have maintained a pattern of regular exercise for a period o f more than eighteen months. category 2 You are in this category if you ar e a Category 6 person who exercises daily and for periods o f longer than 45 minutes. 162

PAGE 173

If you indicated you are now pursuing a program of regular exercise, please describe the type of exercise; also indicate your usual duration and frequency: Type ____________________________ Duration Minutes/Session. --------Frequency ________ Times/Week. Section 7 Other Activity Many people who are not aerobic exercisers may do other forms, such as walking, as part of their daily lifestyle. Such activity may be intentional or part of a job; it may be of short or long duration. If you engage in activity that represents exercise, but it is not described by the definition of regular aerobic exercise in Section 6, please detail the activity here: Type ____ ______ ___________ ___ ____ Duration _________ Minutes/Session. Frequency Times/Week. -----Section 8 Changes in Exercise Status If the information record e d above represents a chang e in your activity or exercise status since your first response to this survey, please detail the change below. Add/Discontinue what activity? ------------------------To what do you attribute the change? ___________________________ 163

PAGE 174

Section 9 About You ****Note: Different Questions***** 1. Sex (circle one): Male Female 2. Race/Ethnicity (circle one): White/Caucasian Black/African American Hispanic/Latino Native American Asian/Pacific Islander Other 3. Body Weight: _______ lbs. This concludes the questionnaire. Thank you for taking the time to complete it again! Simply place it in the attached envelope marked "Survey Return" and drop in Campus Mail. Be sure you also return the Reply/Consent form in its envelope so you will be eligible for the second drawing. 164

PAGE 175

APPENDIX E LIST OF ORIGINAL ITEMS AND ASSIGNED CODES ###-Denotes items removed from scale in instrument tests. Health Motivation 1. Some people are more concerned about their health while others are not as concerned. Ho w concerned are you about your own health? Quite concerned<----------------------->Unconcerned 1 2 3 4 5 6 7

2. Some people are more concerned about the chance of getting sick, while others are not as concerned. How concerned are you about the chance of getting sick? Quite concerned<----------------------->U nconcerned 1 2 3 4 5 6 7

3. How good a job are you doing in taking care of your health right now? Excellent<---------------------------------->Poor 1 2 3 4 5 6 7 < H4> 4. How likely is it that you will try to do a b e tter job of taking care of your health in th e future? Very likely<----------------------->V er y un l ikel y 1 2 3 4 5 6 7 < H5 > 5. What best describes the number of ind i vidual behaviors or precautions that you engage in to insure your good health? Many < ----------------------------------> None 1 2 3 4 5 6 7 < H6 > 6. I frequently do things to improve my h e alth. Strongly agree < ------------------> Strong l y disagr ee 1 2 3 4 5 6 7 7. I search for new information related to m y he a lth. Strongly agree<------------------> Strong l y disagre e 1 2 3 4 5 6 7 ### 8. Although I am concerned about my hea l th there are other things that are more important to me. Strongly agree<------------------> Strong l y d i s a g ree 1 2 3 4 5 6 7 9. In general, the health habits I practice are: Good<--------------------------------->Bad 1 2 3 4 5 6 7 165

PAGE 176

Susceptibility For the following ten questions, please respond according to the scale below. Very likely<-------------------------->Very unlikely 1 2 3 4 5 6 7 Record the number indicating your response in the blank to the right of each condition listed below. How much chance is there that you could ever get: ### 10. asthma 11. pneumonia 12. cancer 13. anemia 14. the flu 15. heart trouble 16. strep throat 17. a cold 18. a cavity 19. a stroke 20. Compared to other people my age, I get sick more easily. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 21. How likely do you think it is, in general, that you will get some sort of illness? Very likely<------------------------->Very unlikely 1 2 3 4 5 6 7 22. Compared to other people my age, I am more likely to have to spend time in bed due to my condition. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 23. I worry a lot about contracting some disease that will prevent me from pursuing normal activities. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 < S15 > 24. My physical health makes it more likely that I will be troubled by disease. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 166

PAGE 177

Severity For the following ten questions, please respond according to the scale below. Very serious<-------------------->Not at all serious 1 2 3 4 5 6 7 Record the number indicating your response in the blank to the right of each condition listed below. How serious would it be to get the following condition within the next twelve months? 10. asthma 15. heart trouble 11. pneumonia 16. strep throat 12. cancer ### 17. a cold 13. anemia ### 18. a cavity 14. the flu 19. a stroke ### 35. Whenever I get sick, it seems to be very serious. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 ### 36. Lacking energy often, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 37. Developing high blood pressure or cardiovascular problems, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 38. Developing nervousness or stress problems, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 ### 39. Becoming overweight, for me, would be: Very serious<------------------->Not at all serious 1 2 3 4 5 6 7 167

PAGE 178

Benefits of Aerobic Exercise For the following sections, the term "regular exercise" refers to aerobic exercise of at least 20 minutes duration, at least 3 times a week. Aerobic exercise is a rhythmic, continuous, sustained exercise such as running, rowing, rope-skipping, bicycling, or swimming that produces an elevated heart rate. 40. How much do you feel maintaining a program of regular exercise will help improve your overall health? Help a lot<-------------------------->Not help at all 1 2 3 4 5 6 7 41. How well do you feel a program of regular exercise will help prevent illness? Help a lot<-------------------------->Not help at all 1 2 3 4 5 6 7 42. A program of regular exercise would make my heart work better. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 43. Regular exercise will help me to tone up my muscles. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 44. A program of regular exercise would help me gain better posture. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 45. A program of regular exercise would help me lose weight. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 < B7> 46. Regular exercise would really help my appearance. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 47. Regular exercise would really help increase my energy level. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 48. Participating in a regular exercise program would increase my self respect. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 168

PAGE 179

49. Regular exercise would really help me keep a good outlook on life. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 50. Participating in regular exercise would help me do better thinking. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 ### 51. A program of regular exercise could benefit me by providing friendship and socialization with other exercisers. Strongly agree<------------------> Strongly disagree 1 2 3 4 5 6 7 Barriers to Aerobic Exercise ### 52. Sometimes I worry that regular exercising can cause health problems. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 53. Sometimes I am concerned that regular vigorous exercising would cause me to have a he a rt attack. Strongly agree<------------------> Strongly disagree 1 2 3 4 5 6 7 < RJ > 54. I am injured while Strongly 1 concerned about the possibility of being maintaining a regular exercise program. agree<------------------> Strongly disagree 2 3 4 5 6 7 < R4 > 55. A regular program of exercise would l ikely make me stiff and sore. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 56. How difficult would you say it is (would be) for you to maintain a regular program of exercise? Very difficult < ------------------------> Very simple 1 2 3 4 5 6 7 < R6> 57. How much do you feel maintaining a regular program of exercise interferes (would interfere) with your normal activities? Interfere a lot<-------------------->No interference 1 2 3 4 5 6 7 1 69

PAGE 180

58. A regular program of exercise is difficult for me to schedule. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 59. Sometimes it seems to me that maintaining a regular program of exercise is more trouble than is worth. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 60. A preference to do other things with my time could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 61. Lack of desire or interest could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 62. Unsuitable weather conditions could keep me from maintaining a regular program of exercise. Strongly agree<------------------->Strongly disagree 1 2 3 4 5 6 7 170

PAGE 181

REFERENCES Afifi, A.A., & Clark, V. (1984). Computer-Aided Multivariate Analysis. Lifetime Learning Publications: Belmont, California. Ajzen, I., & Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and review of empirical research. Psychological Bulletin, 84, 888-918. American College of Sports Medicine. (1978). The recommended quantity and quality of exercise for developing and maintaining fitness in healthy adults. Medicine and Science in Sports, 10(3), vii-x. Anastasi, A. (1988). Psychological Testing (Sixth edition). New York: Macmillan. Bahrke, M.S., & Morgan, W.P. (1978). Anxiety reduction following exercise and meditation. Cognitive Therapy and Research, ~(4), 323-333. Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37(2), 122-147. Beck, K.H., & Lund, A.K. (1981). The effects of health threat seriousness and personal efficacy upon intentions and behavior. Journal of Applied Social Psychology, 11, 401-415. Becker, M.H., Drachman, R.H., & Kirscht, J.P. (1974). A new approach to explaining sick-role behavior in low-income populations. American Journal of Public Health, 64(3), 205-216. Becker, M.H., Haefner, D.P., Kasl, S.V., Kirscht, J.P., Maiman, L.A., & Rosenstock, I.M. (1977). Selected psychosocial models and correlates of individual health related behaviors. Medical Care, 15(5), Supplement, 27-46. Becker, M.H., & Maiman, L.A. (1975). Sociobehavioral determinants of compliance with health and medical care recommendations. Medical Care, 13, 10-24 171

PAGE 182

172 Becker, M.H., Maiman, L.A., Kirscht, J.P. Haefner, D.P., & Drachman, R.H. (1977). The health belief model and prediction of dietary compliance: A field experiment. Journal of Health and Social Behavior, 18, 348-366. Becker, M.H., & Maiman, L.A. (1980). Strategies for enhancing patient compliance. Journal of Community Health, Q(2), 113-135. Belloc, N.B., & Breslow, L. (1972). The relation of physical health status and health practices. Preventive Medicine, i, 409-421. Blair, S.N., Kohl, H.W., Paffenbarger, R.S., Clark, D.G., Cooper, K.H. & Gibbons, L.W. (1989). Physical fitness and all-cause mortality: A prospective study of healthy men and women. Journal of the American Medical Association, 262(17), 2395-2401. Carmody, T.P., Senner, J.W., Malinow, M.R., & Matarazzo, J.D. (1980). Physical exercise rehabilitation: Long-term dropout rate in cardiac patients. Journal of Behavioral Medicine,~' 163-168. Champion, V.L. (1984). Instrument development for health belief model constructs. Advances in Nursing Science, (April), 73-85. Chen, M., & Tatsuoka, M. (1984). The relationship between American women's preventive dental behavior and dental health beliefs. Social Science in Medicine, 19(9), 971-978. Cioffi, J.P. (1980). The effect of health status feedback on health beliefs: An inquiry into the prebehavioral outcomes of a health hazard appraisal. Medical Care, 18, 41-47. Cook, T.D., & Campbell, D.T. (1979). Quasi experimentation: Design and analysis issues for field settings. Chicago: Rand McNally College Publishing Co. Cooper, J., & Croyle, R.T. (1984). Attitudes and attitude change. Annual Review of Psychology, 35, 395-426. Cronbach, L.J. (1970). Essentials of psychological testing. New York: Harper and Row. Cummings, K.M., Jette, A.M., & Rosenstock, I.M. (1978). Construct validation of the Health Belief Model. Health Education Monographs, 6(4), 394-405.

PAGE 183

173 Cummings, K.M., Becker, M.H., & Maile, M.C. (1980). Bringing the models together: An empirical approach to combining variables used to explain health actions. Journal of Behavioral Medicine, 3(2), 123-145. Department of Health and Human Services. (1980). Promoting health/preventing disease: Objectives for the nation. Washington, DC: U.S. Government Printing Office. Department of Health Education and Welfare. (1979). Healthy people: The surgeon general's report on health promotion and disease prevention. (DHEW [PHS] Publication No. 79-55071). Washington, DC: U.S. Government Printing Office. Diekstra, R.F., & Jansen, M.A. (1988). Psychology's role in the new health care systems: The importance of psychological interventions in primary health care. Psychotherapy, 25(3), 344-351. Dishman, R.K. (1981). Prediction of adherence to habitual physical activity. In F.J. Nagle & H.J. Montoye (Eds.) Exercise in health and disease (pp. 259-275). Springfield, IL: Charles C. Thomas. Dishman, R.K. (1982). Compliance/adherence in health-related exercise. Health Psychology, 1(3), 237-267. Dishman, R.K. (Ed.). (1988). Exercise adherence: Its impact on public health. Champaign, IL: Human Kinetics Books. Dishman, R.K., & Gettman, L.R. (1980). Psychobiologic influences on exercise adherence. Journal of Sports Psychology,~, 295-310. Dishman, R.K., Ickes, w., & Morgan W.P. (1980). Self-motivation and adherence to habitual physical activity. Journal of Applied Social Psychology, 10, 115-132. Epstein, L.H., & Wing, R.R. (1980). Aerobic exercise and weight. Addictive Behaviors, 2, 371-388. Folkins, C.H., & Sime, W.E. (1981). Physical fitness training and mental health. American Psychologist, 36, 373-389. Franklin, B.A. (1978). Motivating and educating adults to exercise. Journal of Physical Education and Recreation, (June), 13-17.

PAGE 184

Gale, J.E., Eckhoff, W.T., Mogel, S.F., & Rodnick, J.E. (1984). Factors related to adherence to an exercise program for healthy adults Medicine and Science in Sports and Exercise, 16(6), 544-549. Given, C.W ., Given, B.A., Gallin, R.S., & Condon, J.W. (1983). Development of scales to measure beliefs of diabetic patients. Research in Nursing and Health., 127-141. Godin, G., & Shepard, R.J. (1986). Importance of type of attitude to the study of exercise-behavior. Psychological Reports, 58, 991-1000. Gorsuch, R. (1983). Factor analysis (Second edition). Hillsdale, NJ: Lawrence Erlbaum Associates. 174 Greist, J.H., Klein, M.H., Eischens, R.R., Faris, J., Gurman, A.S., & Morgan, W.P. (1979). Running as treatment for depression. Comprehensive Psychiatry, Q, 41-54. Harris, D.M., & Guten, S. (1979). Health protective behavior: An exploratory study. Journal of Health and Social Behavior, 20, 17-29. Horn, D. (1976). A model for the study of personal choice health behaviour. International Journal of Health Education, 19, 89-98. Hughes, J.R. (1984). Psychological effects of habitual aerobic exercise: A critical review. Preventive Medicine, .ll, 66 -78. Isaac, s., & Michael, W.B. (1983). Handbook in Research and Evaluation (Second Edition). San Diego, CA: EdITS Publishers. Janis, I.L. (1983). The role of social support in adherence to stressful decisions. American Psychologist, 38(2), 143-160. Janz, N.K., & Becker, M.H. (1984). The Health Belief Model: A decade later. Health Education Quarterly, 11(1), 1-47. Jette, A.M., Cummings, M., Brock, B.M., Phelps, M.C., & Naessens, J. (1981). The structure and reliability of health belief indices. Health Services Research, 16(1), 81-98. Kaplan, R.M. (1984). The connection between clinical h ealth promotion and health status: A critical overview. American Psychologist, 39(7), 755-765.

PAGE 185

175 Kasl, s.v., & Cobb, S. (1966). Health behavior, illness behavior, and sick role behavior. 1. Health and illness behavior. Archives of Environmental Health, 12, 246-266. Keppel, G. (1973). Design and analysis: A researcher's handbook. Englewood Cliffs, N.J.: Prentice Hall. Kerlinger, F.N., & Pedhazur, E.J. (1973). Multiple regression in behavioral research. New York: Holt, Rinehart and Winston. Kiesler, C.A., & Morton, T.L. (1988). Psychology and public policy in the "health care revolution". American Psychologist, n(12), 993-1003. Kirscht, J.P., Haefner, D., Kegeles, S., & Rosenstock, I.M. (1966). A national study of health beliefs. Journal of Health and Human Behavior, 2, 248-254. Kirscht, J.P. (1974). Research related to the modification of health beliefs. Health Education Monographs, ~(4), 455-469. Kirscht, J.P. (1983). Preventive health behavior: A review of research and issues. Health Psychology, ~(3), 277-301. Knapp, D.N. (1988). Behavioral management techniques and exercise promotion. In R.K. Dishman (Ed.), Exercise adherence: Its impact on public health (pp. 203-235). Champaign, IL: Human Kinetics Books. Knowles, J.H. (1977). The responsibility of the individual. In J.H. Knowles (Ed.), Doing better and feeling worse (pp. 57-80). New York: Norton. LaPorte, R.E., Montoye, H.J., & Caspersen, C.J. (1985). Assessment of physical activity in epidemiologic research: Problems and prospects. Public Health Reports, 100(2), 131-146. Lau, R.R., Hartman, K.A., & Ware, J.E. (1986). Health as a value: Methodological and theoretical considerations. Health Psychology, 2(1), 25-43. Lewin, K. (1935). A dynamic theory of personality. London: McGraw-Hill. Maiman, L.A. & Becker, M.H. (1974). The health belief model: Origins and correlates in psychological theory. Health Education Monographs, ~(4), 336-353.

PAGE 186

176 Maiman, L.A., Becker, M.H., Kirscht, J.P., Haefner, D.P., & Drachman, R.H. (1977). Scales for measuring health belief model dimensions: A test of predictive value, internal consistency, and relationship among beliefs. Health Education Monographs, 2, 215-230. Martin, J.E. (1981). Exercise management: Shaping and maintaining physical fitness. Behavioral Medicine Advances,~, 3-5. Martin, J.E., & Dubbert, P.M. (1982). Exercise applications and promotion in behavioral medicine: Current status and future directions. Journal of Consulting and Clinical Psychology, 50, 1004-1017. Martin, J.E., Dubbert, P.M., Katell, A.O., Thompson, J.K., Raczynski, J.R., Lake, M., Smith, P.O., Webster, J.S., Sikora, T., & Cohen, R.E. (1984). Behavioral control of exercise in sedentary adults: Studies 1 through 6. Journal of Consulting and Clinical Psychology, 52(5), 795-811. Matarazzo, J.D. (1980). Behavioral health and behavioral medicine: Frontiers for a new health psychology. American Psychologist, 35(9), 807-817. McAlister, L., Farquhar, J.W., Thoresen, C.E., & Maccoby, N. (1976). Behavioral science applied to cardiovascular health: Progress and research needs in the modification of risk-taking habits in adult populations. Health Education Monographs, ~(l), 45-73. Meichenbaum, D., & Turk, D.C. (1987). Facilitating treatment adherence. New York: Plenum Press. Morgan, W.P. (1981). Psychological benefits of physical activity. In F. Nagle & L.H. Montoye (Eds.), Exercise health and disease (pp. 299-314). Springfield, IL: Charles C. Thomas. Morgan, W.P., & O'Conner, P.J. (1988). Exercise and mental health. In R.K. Dishman (Ed.), Exercise adherence: Its impact on public health (pp. 91-121). Champaign, IL: Human Kinetics Books. Naughton, J.P. (1974). Physical activity and coronary heart disease. In P.K. Wilson (Ed.), Adult fitness and cardiac rehabilitation (pp. 3-80). Baltimore: University Park Press.

PAGE 187

177 O'Connell, J.K., Price, J.H., Roberts, S.M., Jurs, S.G., & McKinley, R. (1985). Utilizing the Health Belief Model to predict dieting and exercising behavior of obese and nonobese adolescents. Health Education Quarterly, 12(4), 343-351. Oldridge, N.B. (1981). Dropout and potential compliance: Improving strategies in exercise rehabilitation. In F.J. Nagle & H.J. Montoye (Eds.), Exercise in Health and Disease (pp. 250-258). Springfield, IL: Charles C. Thomas. Paffenbarger, R.S., & Hyde, R.T. (1988). Exercise adherence, coronary heart disease, and longevity. In R.K. Dishman (Ed.), Exercise adherence: Its impact on public health (pp. 41-73). Champaign, IL: Human Kinetics Books. Perkins, K.A. & Epstein, L.H. (1988). Methodology in exercise adherence research. In R.K. Dishman (Ed.), Exercise adherence: Its impact on public health (pp. 399-416). Champaign, IL: Human Kinetics Books. Perri, M.G., & Richards, C.S. (1977). An investigation of naturally occurring episodes of self-controlled behaviors. Journal of Counseling Psychology, (3), 178-183. Powell, K.E. (1988). Habitual exercise and public health: An epidemiological view. In R.K. Dishman (Ed.), Exercise adherence: Its impact on public health (pp. 15-39). Champaign, IL: Human Kinetics Books. Prochaska, J.O., & DiClemente, c.c. (1983). Stages and processes of self change of smoking: Toward an integrative model of change. Journal of Consulting and Clinical Psychology, 51, 390-395. Ray, A.A. (Ed.). (1982). SAS user's guide: Statistics. Cary, North Carolina: SAS Institute. Riddle, P.K. (1980). Attitudes, beliefs, behavioral intentions, and behaviors of women and men toward regular jogging. Research Quarterly for Exercise and Sport, 51(4), 663-674. Rosenstock, I.M. (1974a). Historical origins of the health belief model. Health Education Monographs, ~(4), 328-335. Rosenstock, I.M. (1974b). The health belief model and preventive health behavior. Health Education Monographs, 2(4), 354-386.

PAGE 188

Silver, S.E. (1983). A test of the applicability of the Health Belief Model to the arena of contraceptive decisionmaking. Unpublished doctoral dissertation, University of Florida, Gainesville, Florida. Sinyor, D., Schwartz, S.G., Peronnet, F., Brisson, G., & Serganian, P. (1983). Aerobic fitness level and reactivity to psychosocial stress: Physiological, biochemical, and subjective measures. Psychosomatic Medicine, 45(3), 205-215. 178 Slenker, S E., Price, J.H., Roberts, S.M., & Jurs, S.G. (1984). Joggers versus nonexercisers: An analysis of knowledge, attitudes and beliefs about jogging. Research Quarterly for Exercise and Sport, 55(4), 371-378. Sohn, R.S., & Micheli, L.J. (1984). The effect of running on the pathogenesis of osteoarthritis of the hips and knees. Medicine and Science in Sports and Exercise, 16, 150. A solution for health costs? (1988, 18 December). New York Times, 3, p. 1. Sonstroem, R.J. (1982). Attitudes and beliefs in the prediction of exercise participation. In R.C. Cantu & W. J. Gillespie (Eds.), Sports medicine, sports science: Bridg i ng the gap (pp. 3-16). Lexington, MA: Collomore Press. Sonstroem, R.J. (1987). Stage model of exercise adoption. Paper presented at the national convention, American Psychological Association, New York, NY. Sonstroem, R.J (1988). Psychological models. In R.K. Dishman (Ed ), Exercise adherence: Its impact on publ i c health (pp. 125-153). Champaign, IL : Human Kinetics Books. Stern, M.J. & Cleary, P. (1981). National exercise and heart disease project : Psychosocial chang e s observed during a low-level exercise program. Archiv e s of Internal Med i cine, 1 4 1, 1463-1467. Such m an, E.A. (1970). H e alth attitud e s and b e h a v i or. Archive s of E nvi r onm e ntal H ea lth, 20 10 5 110 Taylor, C.B., S a llis J.F., & N e edle, R. (1985 ). T h e relation of physica l activity and exe rci se to me nt al h ea lth. P ubl i c H ealt h R e p or t s 100(2 ), 1 95 202

PAGE 189

Thoresen, C.E., & Eagleston, J.R. (1985). Counseling for health. The Counseling Psychologist, .l]_(l), 15-87. 179 Tirrell, B.E., & Hart, L.K. (1980). The relationship of health beliefs and knowledge to exercise compliance in patients after coronary bypass. Heart & Lung, 2(3), 487-493. Turk, D.C., Rudy, T.E., & Salovey, P. (1984). Health protection: Attitudes and behaviors of LPNs, teachers and college students. Health Psychology,~, 189-210. Wagner, P.J. (1983). Application of the Health Belief Model to psychosomatic problems: A factor analytic approach. Unpublished manuscript. Wallston, B.S., Wallston, K.A., Kaplan, G.D., & Maides, S.A. (1976). Development and validation of the health locus of control (HLC) scale. Journal of Consulting and Clinical Psychology, 44, 580-585. Wankel, L.M., & Thompson, C. (1977). Motivating people to be physically active: Self-persuasion vs. balanced decision making. Journal of Applied Social Psychology, 7, 332-340. Ward, A., & Morgan, W.P. (1984). Adherence patterns of healthy men and women enrolled in an adult exercise program. Journal of Cardiac Rehabilitation,~' 143-152. Washburn, R.A., & Montoye, H.J. (1986). The assessment of physical activity by questionnaire. American Journal of Epidemiology, 123(4), 563-576. Washburn, R.A., Cook, T.C., & LaPorte, R.E. (1989). The objective assessment of physical activity in an occupationally active group. The Journal of Sports Medicine and Physical Fitness, 29(3), 279-284. Weinstein, N.D. (1984). Why it won't happen to me: Perceptions of risk factors and susceptibility. Health Psychology, 3(5), 431-457. Weintraub, M.S., Rosen, Y., Otto, R., Eisenberg, s., & Breslow, J.L. (1989). Physical exercise conditioning in the absence of weight loss reduces fasting and postprandial triglyceride-rich lipoprotein levels. Circulation, 79(5), 1007-1014. Wilfey, D., & Kunce, J. (1986). Differential physical and psychological effects of exercise. Journal of Health Psychology, TI(3), 337-342.

PAGE 190

World Health Organization (1981). Global strategy for health for all by the year 2000. Geneva: World Health Organization. 180

PAGE 191

BIOGRAPHICAL SKETCH Larry Gage was born on 17 February, 1953, in Fort Lauderdale, Florida. Growing up in south Florida helped to mold Larry's interest in meteorology and the sciences, as a result of several vivid encounters with Atlantic hurricanes. He was selected as the outstanding science student in his class at Stranahan H.S., where he graduated ranked seventh in a class of 770. Larry's initial undergraduate major at Florida State University was meteorology, which soon lost out to the appeal of psychology and counseling. Following graduation from Florida State in 1975, he left Tallahassee to do master's study at Ohio University in a counselor education program. The affiliation with Ohio University and Athens, Ohio, continued following graduation in 1978, as Larry worked in counseling and administrative capacities at the University College. The experience served to cement Larry's interest in higher education, both as a pursuit and a place to work. With the encouragement of colleagues, Larry began another round of advanced study, returning to Florida to pursue the doctorate in counseling psychology at the University of Florida. Through study and work, he has pursued the goal of becoming a counseling psychologist. He interned at George Washington University in Washington, 181

PAGE 192

182 interned at George Washington University in Washington, D.C., finding there an affirmation of his career choice. He has settled near Philadlphia in Doylestown, Pennsylvania, where he lives with his wife Karen Forbes. He joined the staff at the Psychological Counseling Services at Trenton State College in 1988, and will remain there following graduation.

PAGE 193

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequcte, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. d7.r, Chadi-:::r~c__....Professor of Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. James Archer, Jr., Coch Professor of Counselor Education I certify that I have read this study and that in my opinion it conforms to acceptabl:~ standards of scholarly presentation and is fully adequa te, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Pegg Fong Associate Professor of Counselor Education

PAGE 194

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequa te, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Associate Professor of Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarl y presentation and is fully adequate, in scope and quality as a dissertation for the degree of Doctor of Philosophy. gcu-/4,,J / Barbara S. Probert Associate Professor of Psychology I certify that I have read this study a nd that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in sco i.e -a~ quality as a dissertation for the degree of D oc r of PhN sophy. T h is dissertation was submitted to the Graduate Faculty of the Depar tment of Psychology in the College o f Liberal Arts and Sciences and to th e Gradua te School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August, 1990 Dean, Graduate School

PAGE 195

UNIVERSITY OF FLORIDA II I II IIIII I Ill I l l ll l ll l l lll I I I II II I I I II I I llll 1 1 11 1 111 1 1111 1 1111 1 3 1262 08553 7487


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EPCN16G1P_ST38Q2 INGEST_TIME 2017-07-12T21:03:49Z PACKAGE AA00003323_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES