V
. ,
u .
V~
.l Usiiwxﬂ VIA:
.VVaLQtEEi.?. V
.Vft..r)§.hi1tr~ E V V
V :5). .VCLVO‘EerhVI. tutti.
I. Vigirsyﬁmé. (Ink-5.7.4. \L . ‘
s Aﬁxgéblhr $5.51!...qu E
illftnuuiiztsu? .V:
V ....w IEF?‘ (t rd in
’1» 3|1¢Luwi .1; V.
. t! ﬁﬁ‘e.§$¢lx\h EVEVJHWWVQUH
.. h . n V I it. I: ugh-“Ft $50 ’057. v;
V . . .V . .5: .3: +5.16..an
.3!qu . V. . . . 5......Lxlrrf-3VE». nil .
.L.Lﬂrﬂﬁrnbrﬂdkiih..vtrrx}€.§; . V V V .. V V P V tins; LL .«kh
.VVVsIVVVVuVVVVVVVVtVVVItVrtiVSVC VV .3. V V
v

I I. I V J“ .9111} III! I\b
: . . . :5 . V . V V.. “volt“! «5.1 9&1?
. V. 93,15?! .V ... .V . V V . . a... v

V . y l . V . I Or~

V . I. V | V . . .

x, E”?

k

v (1Y2.

1
if I... .515’ .
V V! ..v .3533”!!! 3.13.}..2 V
L.\.tVV.t..V. if. tdvktzi .5 i
§£i§ﬁ$VVVnVF§V2i§3 Va
I}. \o r r. v‘ﬂVIUHYVI-hFPYE13rsJKSsI .fhrv P439». .
2.9.55 u.’1tb.9« vair~§V . 305.58.!EVX: ’Ysthr .Qixﬁg. V
§31§¥§}1Vl.f£yu?i‘b§
leTvictV‘rih‘V.f.§x‘r-l:s V. . 1.1...LV
r ‘Ll .Il‘viniAerCL : 9m .
g... 2.90.9!!!

a“; :3?
V. V V:

“3.53%.
92% V‘ V
. LthIIVLVaNIEPE.
ht... : .21....5grlefk‘op... :..
~V.3J§{1‘ji)zs‘.nﬂ. V
Aﬁitﬁ‘iao6r3hrhﬁiirsr "V.
V». V .. V sV~ V V It!
.1qu urifzilfe . . ..V V V. p. E}. Hm" . CH. nkﬂmiwsnwﬁ L.
V . . V . . V . . V2 ”Stuntihnmfrxi
. V. Vat. 1.7.. V . Vat-visit.
V §l~riViluVr~ V V V . i1} «
V wrathrﬁﬂgéagt .V‘r‘Jﬂtint its." V.VV .V . . . a
. Lﬁﬁkéﬁﬁxsukﬁ rum" rimmﬁuﬁ 3% . F. V:
V . b‘r 2.3% . w} it... Vrzzmmnvtpyls .
E. VKVEVVEVVVVV§ m3» VithE, .VVVVVV.
tr.

ﬂ . I &. 522.3%»; I. i:
{LIL—VI? id; .f-obhtv. ii): afilrt. .197! .39.. ”WW"!
7...!!! I V. :1 5.5:? 9??

T!

‘r

Fiﬁ;

ﬁ
I

-:
“it:
‘ .a‘

‘

5

c

‘2

if: :. >~ "
i‘ . s

V
“MN-nu. ggltmmmgbﬂhat+ {If V uiuNL.
h V ktsilg‘; 2!: V . _ . 971.??? L. as: V... ..
. 9%:me

V d... 1.). $55.!
i . LL

‘
‘2
l
\

rWt’iiPVt LabECVEYESJ-ngﬁﬁui
V .3. .V .8. .mpuairui‘tiru , V
.V: V an“. Vii 4. . cite, V V1 e»? 4%.}.
V,. .7, .E.. r... .3. r...
V V . Vt! Ittrg‘réetlfd, V .
V ‘ 1 V I... :1
VuVVVVVV.d.a.....V...VVVV.VV..m ﬁtﬁmhﬁdnktizi; ataxia: V
A... V! . V S ‘ ﬁ}?§g}y\ .
. § F Ebro. ....N V g. V
<. :50“; “giv‘ix , , . . V1. V.V V V.
N 2 _ A. . ... It‘mﬁuauixit. :. .Ii
1. Hunk.) L. VI...- V V .VrrdwELEF Euvv‘ 55.1st V
_L-Nuﬂ$£€¢.~$.\u.n\cb\k .IV Vii ibiswvﬁfﬁu
erIVIV: t:.1.r..~§ﬂ.x§. {satiititﬁyg V .
.«th: .Llituﬁitiv ¥§.L..V9.&LLVI\V. titsizhh. .
.v .Vuiy. .igl:tottWI.‘YAV€.fP.LLEL1E5‘39)???’
VOttvtIAL tibi‘hr‘tl.nrys::i§ V .34;
V V .V. ‘giisluru .25 .
tit”. LE1
.u‘ﬁéﬁu . .V V~ .

t.
LVL .
2";
V 1?...

x

w 3"
0,).
1.
e
\V it.
~

V}
g
unﬁt.» .

..V.r nit» . k .
+LqubV.$wi.uﬁnﬂv5M$as;S. . . 5!. V.
V L: . $6922.57: tiff». t V r71 V? 2‘ 3.1!. §x its}!
. 3?! 9. [.1 . V 2 V V
Vivi F31“ kaiﬁyf‘sﬂﬁwuhuuafwtﬁvmmﬁiﬂV i9... ~xVa..u)~.:7o .yLNuh..nJV\.¥-ﬁkﬂum03 .
5%}.Ex..KK~ILV-le .ttsﬁsi lxﬁ-TEEVEtSI‘VEItSVINeRHHwﬂ. t...vl.h£...l~t . 3L3?
Lin. ‘ﬂu i‘h‘ihn:‘rz INIVAENLVFIV} . ELKVVIiwtl. ..b3..0Ltrq. .82.»...-
. . hVQoV‘mEiRE‘. - (tn-L‘ix‘va. t I: . 8 V . é
p tr: KP??? V V gktiut‘kbﬁé cV I‘V.’1tA.Ik-l£.y7.§ .ztyrlrta! r V . I V . V . , pV.
h§§t$ft 9.3.1...th £§Ehlu§d¢€f V. .ttlstV V .
.V V V cgik .6... V V 71.0.11? :1}?- ! tut; 1.
V V V V. s . . . _. Vb“ . u , V .. it.) V, [tut-UL
gr}?)v}l7.(lcz.5.¢ch{t It...lsx~:v.£l(.e inunwit V V. . . . V $.21..ch
hxﬂiEEKV£ttiE§ ii): . 5 V .5?! V
1 igﬁir1f§ V V V. V. . .
53% V . . V V, . VS.» 1
.ltiixf ulbirwb. :V. . . V . . V , !.u.v—-3V.n.. .
tsiititck . .....§Rt¢.£ StV :13
.1. . V... - . . V . ﬁiiiftuﬁi
.1 . ,er I a. V . . KEV; i§7€1t§+1~
. LL. kﬂi.‘ . .V . V. V. §‘\§th. .ﬁhg.i*b.ﬂ$}rhtké
. I...» . . V V 1.9!! V, Mr Li. V)-
V . V in! viuwhsgcbk z
L‘LF.»
Iii}:
é.

. .VVSigle-snﬂmthrﬂﬁums.
t gxﬁtlrr‘rtﬁxti.’

attis

 

. V
l: I; . V
V. 5»! uniﬁerﬁﬁkti . V . . . ,VsprttiV .
» fun eiyoaVrILVSVIxLVEEV VV. . V. g: £9.7ka V , V
3V %..§.V.§§V§.V .2...) . V 93.1.11... t .‘xutrir... ~33.th 6U». .V . ,
l L. \leiklkpv‘t \Kxikéaﬂin XI V «{txiﬁz‘. Oi‘liﬂnwg-H. . .u 0%.?! ...... 8.91
9.: iii? Xituwhwnumufﬁnrtgi 2111.. i» t) t;tt\:¥a59££ist35:§r
Efrik§lrn§\l\kt it. .kE‘ELL LVI.
g§:¥i.it\vﬁl\ . . .. Lulu.
{1?}tLVJilVWNHnD.}Ic§V V. 2. . V
tit. V *qul‘x)‘ a: V

E
V 3, B. V . fun: 3.1.; V
an, 3.1.04.1 haunt»... V. .i . .
g5? . V VVVtVVrEVXVVVVVVV.Vuu $1.» .
yt‘ \ilsta‘r'? t
A...‘ ‘51-‘10: EN
1 V ll.IV V» I V . v . V E ‘II ’1
.V V . VV 1 .V V :1 V u. ...f . V V 1. . .V V V
. vQEAI. . V V. . V V ‘ L. .V .v .
i V. Van“ . V)! 5s .. k

.l .
9:9; hiVstIVvLV.
V V . . V \E.71..§. . . .. .111
.V V. :IriV ; . .L . .. .VV . . . . V . . V gt!!! (VLVQH. V .9: b.5‘oé’ I“?
V .VV.. VVV V . V It: Vt iytﬂm‘ku
V .

 

 

 

igﬁo}
, .

 

KixiAl‘ V.
t . .VVIV. (1|
V: . . tmnxn. angmutﬁustf
ZutttttkEEFV. L.ct.\{ak;h\tt
xx. ¥~§fta:~\u{li tutti‘kitki r:
ﬁat-0?)... .L. flu-{i}. 2. n-
.n .v nati+$¢.u.anﬂk.vﬁl .1 E534»? r .K
.1? k. *3. FE;
IV I - fitih 5.5.. Wig. .L .a f
‘lti V E.
th‘. gt.” V
.‘AiEubo‘leVw: :1 A LV VV
.yVo {Dip}.V|Y.l bv - V V
(tuft. {xv/Vent] unlit 29.1 3.... .15.. . . V V . .
i‘L’ll’ ‘IEJ’VI.-{ \ V V

1.255 .V V
I“ it...
1909}; \ aanxi

i .11

 

” Ill/IIIIIIHIII'll/Ill!!!” 11771

028 J

LIBRARY
Michigan State
University

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

This is to certify that the
dissertation entitled
A PATH ANALYSIS OF THE BIOPSYCHOSOCIAL VARIABLES
RELATED TO EXERCISE PERFORMANCE AND ADHERENCE
presented by

Mary Clearing-Sky

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Psychology

Major professor

Date July 23, 1986

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution 0-12771

u.

\

 

 

 

MSU

 

RETURNING MATERIALS:
Place in book drop to

 

 

 

 

LIBRARIES remove this checkout from
w your record. FINES will
be charged if book is
returned after the date
stamped below.
, I mm MAGl02
“J v‘ "3 ___—_.——,
ff», 5/ (7‘10 é‘T/r- {K ,q72__ PE§§}1998
- QDX A011 - z 53
“CLO” 1: I": g 5-9.!“ fJANk12000
~ “‘8’ \27 “26%,” Iggff’. 04
" ' .. APRaMOEU ‘0
9 0208 t;§$§

 

APR? 5 1992‘
W33

3 v

 

 

A PATH ANALYSIS OF THE BIOPSYCHOSOCIAL VARIABLES

RELATED TO EXERCISE PERFORMANCE AND ADHERENCE

BY

Mary Clearing-Sky

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Psychology

1986

ABSTRACT

A PATH ANALYSIS OF THE BIOPSYCHOSOCIAL VARIABLES
RELATED TO EXERCISE PERFORMANCE AND ADHERENCE
BY

Mary Clearing-Sky

Path analysis was utilized to examine the assumptions
of a hypothesized model of the causal relationships of
several physical, psychological, and psychosocial variables
influencing exercise adherence and performance. The model
extracted variables from human performance, exercise
adherence, social learning, and cognitive mediation
theories. Social support, empathy, self-concept, physical
fitness, perceived fitness, goals, anticipated satisfaction
with performance, anticipated satisfaction with adherence,
self-efficacy for performance, self-efficacy for adherence,
and self-motivation were predicted to affect each other and
exercise performance and adherence.

Eighty-five male and female college students completed
questionnaires and exercise tests in the first week of three
ten-week exercise and conditioning classes. Missing data
and class attrition created sampling problems. Fifty-four
cases in Week 1 and 44 in Week 10 were usable in the path
analyses. The Harvard Step Test, the Twelve-Minute Run, the

Tennessee Self-Concept Scale, self-efficacy scales,

Mary Clearing-Sky

Dishman's (Falls, Baylor, & Dishman, 1980) Self-motivation
Inventory, and several scales developed for this study were
utilized. A planned follow-up was abandoned due to low
returns.

Sonstroem's (1984) exercise adherence theory was
consistently supported. Fitness affected perceived fitness
which in turn predicted attitudes toward exercise
performance and adherence. Fitness also predicted adherence
and performance.

Bandura's (1977) self-efficacy theory was partially
supported. Self-efficacy mediated the effect of goals,
perceived fitness, and social support on adherence. Self-
efficacy did not positively influence performance or mediate
the effects of perceived fitness, goals, and self-concept on
performance as expected.

Garland's (1985) cognitive mediation theory was
partially supported. Anticipated satisfaction with
adherence mediated between goals and performance.
Anticipated satisfaction with performance, self-efficacy for
performance, and self-efficacy for adherence did not
influence performance as expected.

Dishman's (1982a) exercise adherence theory was
partially supported. Fitness influenced performance and
adherence and several variables affecting adherence. Self—
motivation did not significantly influence performance or

adherence as expected.

Mary Clearing-Sky

Social support had a significant effect on perceived
fitness and self-efficacy for adherence. The separation of
adherence and performance aspects of measures suggested by
the present study was supported by the divergent results
wherever these separations were made.

To enhance adherence, it is suggested that exercise
program leaders and exercisers should emphasize social
support, adherence and performance goals, periodic fitness
and performance tests, as well as reinforcement of the
exercisers' perceptions of their progress, fitness,
performance, goal attainment, ability to adhere, and
adherence.

The study emphasizes the value of path analysis in the
multivariate study of adherence. Further research needs
include long-term studies of adherence, direct comparison of
the explanatory power of theories of exercise adherence,
improved measurement devices, and standardized definitions

of variables.

DEDICATION

To Madgequomoqua, the original Rolling Thunder
Clearing Sky Woman, who gave me life and taught me
that industry, education, and speaking English would
help me "get ahead" in the White Man's World and that
gentleness, caring, and appreciation of the Earth

would help me live in the Natural World.

 

ii

ACKNOWLEDGMENTS

I wish to acknowledge Bob Caldwell and Deb Feltz for
their assistance in focusing and identifying a viable area
of study among my many interests, and for their help in
conceptualizing a project that utilized several of those
interests. I wish to thank them both for their teaching,
support, and patience through the lengthy process of
proposal, data collection, analysis and interpretation, and
reporting. I want to also thank Deb Feltz for her aid in
assuring that I had subjects when registration problems
nearly eliminated my project.

John Hurley once again demanded the precision and
polish expected of those to be called scholar upon
completing their degree requirements. I appreciate John's
ability to demand quality as well as to smile and laugh at
"stupid" errors and be a colleague in my progress.

I thank Dozier Thornton for his regular expressions of
interest and concern for my project and me. His suggestion
that I also examine social support and empathy was important
to the study. His support, empathy, and caring were
important to me.

Finally, I acknowledge my beautiful family: Jim,
Elizabeth, Thomas, and Nanette, whose love, encouragement,

iii

and sacrifice assured the time and freedom I needed to

pursue a fifth love.

iv

TABLE OF CONTENTS

List of Tables. . . . . . . . . . . . . . vii

List of Figures . . . . . . . . . . . . . viii

CHAPTER I - INTRODUCTION . . . . . . . . . . 1

CHAPTER II

Literature Review. . . . . . . . . . . . 4
Performance Variables . . . . . . . . . . 4
Perceived Physical State . . . . . . . . 4
Self— concept . . . . . . . . . . . . 5
Goals. . . . 6
Anticipated Satisfaction With Performance. . . 7
Self— —efficacy . . . . . . . . . . . . 8
Adherence Variables . . . . . . . . . . . 10
Perceived Fitness. . . . . . . . . . . 10

Self-concept . . . . . . . . . . . . 11
Physical States . . . . . . . . . . . 12
Goals. . . . . . . . . . . . . 12
Social Support. . . . . . . . . . . . l3
Empathy . . . . . . . . . . . . . . l4
Self-efficacy . . . . . . . . . . . . 15
Self—motivation . . . . . . . . . . . 15
Exercise Benefits. . . . . . . . . . . . 16
Physiological Benefits . . . . . . . . . l6
Psychological Benefits . . . . . . . . . 18
Adherence Problems . . . . . . . . . . . 20

CHAPTER III

The Present Study. . 23
Multivariate Relationships in Exercise Adherence
and Performance . . . . . . 23
Variables and Their Definitions . . . . . . 26
Path Model Hypotheses . . . . . . . 31

Other Hypotheses and Analyses. . . . . . . 32

CHAPTER IV

Method

S.

Subjects .

Method

Measures . . . .

CHAPTER V

Result

S.

Path Analyses. .
Path Model Hypotheses
Further Comparisons.

Summary of Path Analysis Results . .
Limitations of Path Analysis.
A Proposed Respecified Model.

Other Hypotheses and Analyses . .

CHAPTER V

Discussion.

I

Major Findings .
Other Analyses .

Uncontrolled Factors and Measurement Problems .
Conclusions . . . . . .
Implications for Exercise Programs. . . .
Implications for Future Research . . .
Appendix A: Questionnaires . . . . . .
Appendix B: Informed Consent and Information Release
Appendix C: Course Outline . . . . . .
Appendix D: Testing Schedule . . .
Appendix E: Pearson Correlations . . .
Appendix F: Statistical Formulae . . . .
Appendix G: E—tests . . . . . . .
REFERENCES. . . . . . . . . .

vi

Page

118
121
123
125
128
131

133

LIST OF TABLES

Table Page
E—I Correlations Among Variables, Week 1 . . . 125
E—2 Correlations Among Variables, Week 10 . . . 125

3 Means and Standard Deviations for Variables

in Path Analyses . . . . . . . . . 46

4 Standardized Regression Coefficients for
Multiple Regression Analyses . . . . . 48

G—S E—tests Testing Full Models Against
Hypothesized Models . . . . . . . . 131
6 Comparison of the R25 for Performance and
Adherence Between the Hypothesized Model
and the Full Model . . . . . . . 78
7 Univariate g-test of MANOVA Test for

Significance of Change . . . . . . . 91

Figure

l

2

10

ll

12

13

14

15

l6

17

18

LIST OF FIGURES

Hypothesized Model, Week 1.

Hypothesized Model, Week 10

Predictor and Criterion Variables for

Regression Equations. .

Hypothesis 1,
Hypothesis 2,
Hypothesis 3,
Hypothesis 4,
Hypothesis 5,
Hypothesis 6,
Hypothesis 7,
Hypothesis 8,
Hypothesis 9,

Hypothesis 10,

Illustrated
Illustrated
Illustrated
Illustrated
Illustrated
Illustrated
Illustrated
Illustrated

Illustrated

Illustrated.

Status of the Path Hypotheses.

Significant Paths, Week 1

Significant Paths, Week 10.

Proposed Respecified Model, Week 1

Proposed Respecified Model, Week 10.

viii

 

Multiple

Page
28

29
51

56
57
59
61
63
65
67
70
72
76
79
80
81
85

87

Chapter 1

INTRODUCTION

In the last decade researchers have reported several
physiological and psychological benefits of vigorous

exercise. An exercise revolution has resulted which may
improve life-style, cardiovascular health, and longevity"
(Paffenbarger, Hyde, Wing, & Steinmetz, 1984, p. 491). In a
recent two-year period an estimated 50% increase in
exercisers occurred in American society. However, the
dropout rate in exercise programs has continued to be 30—70%

1 has

throughout the last decade, and exercise adherence
become an intriguing problem (Danish, 1983; Dishman, 1982c;
Dishman,l985; Dishman, Ickes, & Morgan, 1980; Falls, Baylor,
& Dishman, 1980; Sachs, 1982; Snyder & Spreitzer; 1984).
Research_attempts to understand who will persist in the hard
work and discipline required to maintain an exercise program
have resulted.

Several variables have been empirically related to

the performance2

of various exercise, sport, or other
activities or tasks. Several studies have also examined
Variables related to adherence or persistence in the
activity as a specific issue. Critically, the same
Variables relate to both performance and adherence which in

turn affect each other. Adherence or persistence of effort

1

often leads to improved performance which often leads to
continued adherence. The effects of these variables on
adherence versus performance have not been clearly or
consistently separated either theoretically or
methodologically. The present study separated the adherence
and performance measures and the adherence and performance
aspects of two other measures in order to clarify the
separate phenomena. The purpose of the present study was to
examine the relationship of several of these variables to
each other and to adherence to and performance of an
exercise activity.

Dishman (1982c) suggested that an important step in
exercise research is the testing of these relationships in a
multivariate analysis. The present study developed and
tested a hypothesized model of the causal relationships of
several physical, psychological, and psychosocial variables
influencing exercise adherence and performance. To test
this biopsychosocial model, repeated measures of these
variables were collected from 42 women and 43 men students
enrolled for one term in exercise and conditioning classes
at Michigan State University. Measures were collected at
Week 1 and Week 10 of the 10 week course. In Week 1, 85
students provided data. In Week 10, 75 students provided
data. Fifty-four cases in Week 1 and 44 cases in Week 10
were usable in the path analyses.

The present study also attempted to determine which

factors were involved in the decision to adhere to an

exercise program or to quit exercising despite its widely
known benefits. Such information can help in our efforts to
understand and promote adherence to exercise and other
health-seeking behaviors. It may also contribute to the
understanding and promotion of compliance3 with the
treatment of illness, such as compliance to an obesity

management plan, a diabetes management plan, or a cardiac

rehabilitation plan.

Chapter II

Literature Review

 

Researchers and wellness promoters in medicine and
psychology have cited several benefits of exercise for
physical and psychological health. As a result, many people
have begun exercising. However, the number of people who
drop out of exercise programs has remained high.

Observation of this drop out problem has led to several
studies of adherence. This chapter reviews (a) several
variables that have been shown to affect performance in
exercise and other activities, (b) variables that have been
linked specifically to exercise adherence (c) the benefits

of exercise, and (d) the exercise adherence problem.

Performance Variables

 

Several biological, psychological, and psychosocial
variables have been shown to function as predictors of
exercise behavior or participation in sports. These
variables include perceived physical state, self-concept,
goals, anticipated satisfaction with performance, and self-

efficacy.

Perceived Physical State

 

Thoughts one has about one's physical condition have an
important influence on one's performance of physical tasks.

In Bandura's (1977) theory of self-efficacy, (the belief

4

5
that one can successfully perform a specific behavior

required to produce a specific outcome), cognition about
physical condition is a strong predictor of self-efficacy.
Such cognitions provide a source of information upon which
self-efficacy expectations are built. Bandura's theory
predicts perceived physical states would influence
performance indirectly through self-efficacy. Feltz and
Mugno's (1983) study supported this aspect of Bandura's
theory. The perceived autonomic arousal of their student

divers influenced performance through self-efficacy.

Self-concept

 

Physical fitness also influences self—concept (Brown,
Livingston, & Morrow, 1982; Sonstroem, 1974, 1978, 1984;
Trujillo, 1983). The attitudes, beliefs, or cognitions one
holds toward the self make up self—concept. Several studies
have found that improvement in self-concept resulted from
increases in physical fitness (see Sonstroem, 1984 for a
recent review). Trujillo (1983) found the self-esteem of the
college women in weight training and running exercise
programs increased. Sonstroem's (1974, 1978, 1984) findings
suggest exercise increased self-esteem or improved self—
concept. Brown et al. (1982) found that physical self-
concept, measured by a subscale of the Tennessee Self-
Concept Scale, improved more than global self-concept in a
group of college women who participated in a conditioning
Program. Brown et al.'s (1982) study illustrated the

importance of studying selected aspects of self-concept.

6
Examining the opposite effect, Olszewska (1982) found

positive self-concept predictive of effective performance in

sport behavior.

Goals

Goals, self-set standards for performance, have been
shown to affect behavior in a number of tasks including
exercise. The influence appears to be mediated by self-
efficacy and persistence, with persistence being influenced
by self-efficacy (Bandura & Shunk, 1981; Garland, 1985;
Shunk, 1984). Goals, goal setting, and cognitions regarding
goals have been demonstrated to affect both self-efficacy
and the execution of behaviors. Bandura and Schunk
suggested goals were used to evaluate behavior. Children in
their study made their self-satisfaction conditional upon
those standards. Anticipated satisfaction with one's
achievement serves as an inducement to persist toward the
goal until those self-standards are met. Setting goals
provides a source of ongoing information about one's
behavior that affects self-efficacy, that in turn influences
persistence and strength of effort. Goals can also serve in
the development and maintenance of intrinsic interest in an
activity which also affects persistence. "Most of the
activities that people enjoy doing for their own sake
originally had little or no interest for them...However,
through favorable continued involvement, almost any activity
can become imbued with consuming significance" (Bandura &

Schunk, p. 587). Bandura and Schunk found children who set

7
goals completed math tasks rapidlly, had high self-efficacy,

and high interest in math. Schunk achieved similar results
with children formerly deficient in math skills.

Garland (1985) also found goals important in predicting
task activity and achievement. Subjects performing a
laboratory task and having high goals had both higher self-
efficacy and greater achievement. Tu and Rothstein (1979)
found that goals set by junior high school students or
teachers interacted with students' dependence or
independence to determine exercise persistence and
performance. Independence-oriented students improved at a
significantly faster rate when they set their own goals,
while dependence-oriented students improved significantly

faster when the teacher set goals.

Anticipated Satisfaction With Performance

Studies have reported both negative and positive
associations of anticipated satisfaction with performance.
Garland (1985) postulated two cognitive constructs that
mediate between goals and performance: performance
expectancy which is similar to Bandura's (1977) self-
efficacy construct, and performance valence, anticipated
satisfaction with the performance itself (not with expected
outcomes). Garland found performance valence negatively
associated with performance. People who aspired to
achievement of higher goals expected some dissatisfaction
and performed at higher levels. Both performance valence

and self-efficacy were predictive of magnitude and duration

 

of effort. Individuals sought to minimize anticipated self-
dissatisfaction with substandard performance or to increase
self-satisfaction through increasing effort. Others have
found anticipated satisfaction with performance to be
positively associated with performance (Bandura, 1977).
Bandura and Schunk reported that anticipated satisfaction
with performance served as in inducement to persist (see

Goals above).

Self-efficacy

Self—efficacy has also been recognized as an important
component in human behavior. Bandura (1977) proposed a
theory of self—efficacy to explain changes in behavior.
Bandura's self-efficacy is the belief that one can
successfully perform the specific behavior required to
produce a specific outcome. Bandura views self-efficacy as
a common cognitive mechanism mediating all behavioral
change. Self-efficacy affects one's choice of activities
and behavioral settings, and the amount and persistence of
effort expended in performance (Bandura, 1977; Bandura &
Adams, 1977; Bandura, Adams, & Beyer, 1977; Bandura &
Schunk, 1981). The level and strength of efficacy
expectations are built upon and altered by four principal
sources of information: performance accomplishments,
vicarious experience, verbal persuasion, and physiological
states. Performance accomplishment or personal mastery is
the most powerful source of efficacy information.

Successful performance will modify future efficacy

expectations. Given necessary skills and proper incentives,
self-efficacy is a major determinant of behavior (Bandura,
1977).

Self-efficacy theory has been tested in a wide array of
psychological experiments. Self—efficacy has been found to
predict behaviors in experiments that studied fear and
phobias (Bandura, Adams, Hardy, & Howells, 1980),
intellectural and social skills (Bandura & Schunk, 1981;
Kazdin, 1980), and self-regulatory behaviors such as dealing
with addictions (Marlatt & Gordon, 1980) or coronary
rehabilitation (Bandura, 1982). Recent self—efficacy tests
have seen diverse applications in such topics as weight
rehabilitation (Bernier & Poser, 1984; Weinberg, Hughes,
Critelli, England, & Jackson, 1984), school performance
(Schunk, 1984), and coping with chronic obstructive
pulmonary disease (COPD) (Kaplan, Atkins, & Reinsch, 1984).
(See reviews by Bandura, 1982; Bandura, 1984; Bandura,
Adams, Hardy, & Howells, 1980). Self-efficacy was reported
to affect these behaviors by influencing the subjects'
amount and persistence of effort.

Sport psychologists have tested whether self-efficacy
predicts sport performance. Self—efficacy has been found to
predict performance in gymnastics (Lee, 1982), children's
sport (Roberts, Kleiber, & Duda, 1981), and diving
performance (Feltz, 1980, 1982; Feltz, Landers, & Raeder,
1979; Feltz & Mugno, 1983). It has been demonstrated that

self-efficacy predicts performance in leg—strength and

10

endurance tasks and that manipulating efficacy by deception
alters that performance (Weinberg, Gould, & Jackson, 1979;
Weinberg, Gould, Yukelson, & Jackson, 1981; Weinberg,

Yukelson, & Jackson, 1980).

Adherence Variables

Many biological, psychological, and psychosocial
variables have been posited to function as antecedents or
predictors of adherence to exercise behavior (Albinson &
Moore, 1983; Brownell, 1984; Carmody, Senner, Malinow &
Matarazzo, 1980; Danish, 1983; Dishman, 1981, 1982a, 1982b,
1982c, 1985; Dishman, Sallis, & Orenstein, 1985; Falls et
al., 1982; Gayle, Eckhoff, Mogel, & Rodnick, 1984; Goodrick,
Warren, Hartung, & Hoepful, 1984; Haskell & Blair, 1980;
Keefe & Blumenthal, 1980; Laffrey & Isenberg, 1983; Paolone,
1983; Sachs, 1979, 1982; Serfass & Gerberich, 1984; Slenker,
Price, Roberts, & Jurs, 1984; Snyder & Spreitzer, 1984).
Dishman et al.'s (1985) review of the exercise and adherence
literature listed 37 exercise predictors that have been
studied. The following studies suggested the variables that

were subsequently examined in the present study.

Perceived Fitness

 

Dishman et al.'s (1985) review reported that two
studies found no association between perceived physical
competence and adherence to an exercise program. Dishman's
(1982c) review suggested beliefs in one's physical ability

had been found to predict initial involvement in physical

 

11

activity, but not long term adherence. Snyder and Spreitzer
(1984) found perceived athletic ability one of three
significant predictors of exercise adherence. The
inconsistent results may be due to the use of different
populations, study methods, and measures. Dishman et al.
(1985) referred to Dishman's (1981, 1982b) retrospective
studies of the medical records of cardiac testing and
exercise prescription program subjects, many of whom were
cardiac rehabilitation patients. Snyder and Spreitzer'
subjects were persons actually contacted for follow-up
adherence information up to 5 years after an exercise
program. These subjects had initially expressed an interest
in physical fitness, had a stress test, and an exercise
prescription. Because standardized definitions and
measurements are not in use (Dishman et a1.), and methods
and populations differ, it is difficult to interpret these
inconsistent results. Sonstroem (1974, 1978, 1984)
theorized that perceived athletic ability influences

adherence through its affect on self—concept.

Self-concept

Review of the literature failed to locate a study
utilizing physical self—concept per se as an exercise
adherence predictor. Perceived physical competence
mentioned above may have been measuring the same construct.
Little standardization of definitions existed in the
literature (Dishman et al., 1985). Brown et al.'s (1982)

finding that physical training directly affects self—concept

12

suggested self-concept may mediate the effect of physical
fitness on adherence. In Sonstroem's (1974, 1978, 1984)
adherence theory, general self-esteem or self-concept is
enhanced by increases in perceived athletic ability and
leads to furthur attraction to and involvement in physical
activity. Thus, it appeared important to study both

perceived fitness and a more general physical self-concept.

Physical States

Exercise and health programs have often measured
physical states but few researchers have included such
measures in studies of adherence (Dishman, 1981). The
following studies suggested physical states have a direct
effect on exercise adherence. Dishman and Gettman (1980)
found body weight and percent body fat significantly related
to adherence. Dishman et al.'s (1980) study reported that
of several variables tested only body fat, weight, and self—
motivation were significant predictors of exercise
adherence. Persons high in self-motivation and low in body
weight and body fat were high in adherence (Dishman, 1981,
1982a, 1982b, 1982c, 1985; Dishman et al.,1980; Falls, et
al., 1980).

Goals

Review of the literature revealed two studies that
provided some evidence that adherence depends on goal
attainment or expected goal attainment. Tu and Rothstein's

(1979) study suggested that goal types interacted with

l3

personality types to predict adherence. Student or teacher-
set goals interacted with the children's dependent or
independent personalities to predict adherence.
Independence-motivated children performed better when they
set their own goals. Dependence-motivated children
performed better when the teacher set goals. The intention
to jog, a general goal, correlated highly (r = .82) with
jogging behavior in Riddle's (1980) study of nonexercisers

and joggers.

Social Support

 

Social support (the caring, encouragement, assistance,
and positive attitudes toward exercise of significant
others) has also been linked to exercise adherence. Support
appeared to be a component of setting or situation that
Dishman (1982b) suggested was important to adherence.
Participants in a physical activity program whose spouses
had favorable attitudes toward exercise had better
complianCe patterns than those whose spouses had negative or
neutral attitudes (Heinzelmann & Bagley, 1970). Bandura
(1982) reported the significant relationship of the spouse's
estimation of patient self-efficacy to exercise compliance
in cardiac rehabilitation.

Godin and Shephard (1985) reported that in their study
of 20 couples social support significantly facilitated male
spouses' beginning and adhering to an exercise program.
Dishman et al.'s (1985) review of the literature cited

several studies that concluded social support of spouse,

14

exercise partners, or program staff strongly predicts
exercise adherence. Two studies found no association of
social reinforcement of program staff or exercise partners
with adherence. It may be that social support of intimates
is more effective in maintaining adherence than support of
staff or exercise partners. Hirsch (1981) suggested social
supportmaintains critical identities and role changes
during life transitions. One's attempt to change one's
life-style by becoming an exerciser certainly qualifies as
such a transition. It appeared that support of the role or
self-concept of exerciser would enhance one's adherence to

an exercise program.

Empathy

In the medically-oriented compliance literature,
rapport (mutuality or understanding, trust, and caring) in
the social relationship between the medical practitioner and
the patient has been reported to be basic to appointment
keeping and compliance (DiMatteo & DiNicola, 1982; DiMatteo,
Friedman, & Taranta, 1979; Friedman & DiMatteo, 1979; Truax
& Carkhuff, 1965). Rapport has been found to depend in part
upon empathy, the practioner's ability to understand the
patient's communications and relate that understanding to
the patient. Empathy, it has been suggested, is crucial to
compliance (Hurtado, Greenlick, & Columbo, 1973; Nelson,
Gold, Hutchinson, & Benezra, 1975). Empathy has not been
empirically linked to adherence as narrowly defined in this

study. The construct is quite similar to social support and

15

it's operation in compliance appeared similar to the

influence of social support on compliance and adherence.

Self-efficacy

 

Several of the studies reviewed under Performance
Variables suggested self-efficacay had its effect on
performance through its effect on persistence of effort.
Accomplishment in weight management (Bernier & Poser, 1984;
Weinberg et al., 1984), management of fear and phobias
(Bandura et al., 1980), addiction control (Marlatt & Gordon,
1980), school performance (Schunk, 1984), and children's
sports (Roberts et al., 1981), required continued effort
that appeared to be sustained by self-efficacy beliefs.

Exercise compliance per se was predicted by high self-
efficacy in cardiac rehabilitation patients (Bandura, 1982)
and chronic obstructive pulmonary disease patients (Kaplan
et al., 1984). Albinson and Moore (1983) reported that
exercisers high in self-efficacy were significantly more
likely to continue in their exercise programs. Dishman et
al.'s (1985) review of the literature cited unpublished
observations that perceived self-efficacy and self-estimates
of the likelihood of adherence have predicted future

exercise activity.

Self-motivation

 

Bandura (1977) suggested self-efficacy information
affects self-motivation and thus persistence of effort.

Falls et al. (1980) found motivation, a general disposition

16

to persevere, to be one of the most frequently examined
factors in medical compliance literature. More than 85
percent of the studies they reviewed that utilized the
motivation construct found it to be a significant factor in
influencing compliance. Falls et a1. (1980) tested the
power of several psychological and biological factors to
predict adherence to exercise in a group of rowers and a
group of joggers. Self-motivation was the best
discriminator of adherence compared to aesthetic experience,
locus of control, and physical estimation, attraction, and
attitudes toward physical activity.

In another Dishman study (1981), self-motivation, body
weight, and body fat were the only significant predictors of
exercise adherence. Gayle and Eckhoff (1984) however,
reported very low correlations of self-motivation as well as
several other posited exercise predictors. Snyder and
Spreitzer (1984) also reported a very low correlation for

self-motivation and exercise adherence.

Exercise Benefits

 

Physiological Benefits

 

Several studies have suggested regular, vigorous
exercise reduces the risk of cardiovascular and other
diseases (Hagberg, 1986; Mustafa, Garrett, & Freeman, 1984;
Paolone, 1983; Sharkey, 1975; Shephard, 1984; Snyder &
Spreitzer, 1984; Spirduso, 1985; Stamford, 1984; Thompson &

Martin, 1984; Van Andel & Austin, 1984). Cardiovascular

17

fitness has been correlated with physical fitness and
regular, vigorous exercise (Blair, Goodyear, Gibbons, &
COOper, 1984; Falls et al., 1980; Haskell & Blair, 1980;
Kaplan, 1984; Merkin & Hoffman, 1978; Paffenbarger et al.,
1984; Serfass & Gerberich, 1984). Increases in maximum
oxygen consumption (deVries, 1970, 1979), cardiac output,
stroke volume, heart volume, and decreases in heart rate
(Falls et al., 1982) and blood pressure (Blair, Collingwood,
Reynolds, Smith, Hagan, & Sterling, 1984; Blair, Goodyear,
Gibbons, & Cooper, 1984) as well as changes in lipids and
lipidprotein level (Goldberg, Elliot, Schutz, & Kloster,
1984), are reported in those persons engaged in vigorous
exercise.

Dishman's (1985) review reported that exercise may
decrease Type A-coronary prone behavior and may decrease the
endocrine response to stress. Shephard, Corey, and Cox
(1982) found reduced health risk behavior in their fitness
and fitness education program participants. After six
months participants had increased physical activity,
decreased or quit smoking and alcohol consumption, and had
reduced blood pressure. Sinyor, Schwartz, Peronnet,
Brisson, and Seraganian (1983) reported quicker
Physiological recovery from experimentally induced stress in
exercisers. Body composition and weight of vigorous
exercisers also differed from that of nonexercisers and was
an important correlate of cardiovascular fitness (Falls et

al., 1983; Hagberg, 1986) as well as a source of psychic

18

pleasure to the person living a culture that idealized
'thinness. Exercisers had less fat relative to fat-free body
mass than nonexercisers (deVries, 1970; Falls et al., 1980).
Exercise has been widely recommended as an adjunctive
treatment for obesity and excessive cholesterol rather than
diet or drugs alone (Paffenbarger et al., 1984). Muscular
strength, the ability to contract muscles against a
-resistance, and muscular endurance, the ability to repeat
the contraction or sustain it, improve with exercise (Falls
et al., 1980; Merkin & Hoffman, 1978; Ryan, 1980). The
Olympic Issue of the Journal of the American Meical
Association (Lundberg, 1984) was devoted entirely to
exercise and sport research and featured an editorial
acclaiming the effects of exercise on cardiovascular health

(Kaplan, 1984).

Psychological Benefits

Several psychological benefits of regular, vigorous
exercise have been reported. Vigorous exercise has been
linked with improved mood, self-concept, stress management,
and work behavior (Dishman, 1982a; Folkins & Sime, 1982;
Parent & Whall, 1984; Sheehan, 1984). Anti-depressive and
anti-anxiety effects have been reported (Bahrke & Morgan,
1978; Greist, Klein, Eischens, Faris, Gurman, & Morgan,
1978; Greist, Eischens, Klein, & Paris, 1979; Morgan, 1979,
1981, 1985). Sinyor et al. (1983) reported aerobically fit
subjects recovered faster from experimentally induced

psychosocial stress. Others reported subjects managed

19

stress better as a result of exercise (Blair, Collingwood,
Reynolds, Smith, Hagan, Sterling, 1984; Walsh, 1983).
Laffrey and Isenberg (1983) found exercise led to feelings
of well—being, increased capacity to work, increased ability
to relax, improved self-image, and decreased anxiety and
depression. Blumenthal, Sanders, Needels, & Wallace (1982)
found exercisers had more vigor and less anxiety, tension,
depression, and fatigue than nonexercisers. Exercisers in
several studies reported a ”feeling better" phenomenon
(Carmack & Martens, 1979; Kau & Fischer, 1974; Morgan, 1979,
1985).

Improvement in self-concept, self-esteem, and
subjective well-being of exercisers was reported by several
investigators and reviewers (Arends, 1983; Brown, Morrow, &
Livingston, 1982; Dishman, 1982a; Harris, 1981a, 1981b;
Hilyer & Mitchell, 1979; Jasnowski, Holmes, Solomon, &
Aguir, 1981; Jorgenson & Jorgenson, 1979; Laffrey &
Isenberg, 1983; Parent & Whall, 1984; Riddle, 1980; Sachs &
Pargman, 1979; Sonstroem, 1974, 1978, 1984; Trujillo, 1983;
Young & Ismail, 1976, 1977). Several other studies have
provided additional evidence that vigorous exercise is
connected to positive mental health characteristics
(Brownell, 1984; Blair, Collingwood, Reynolds, Smith, Hagan,
& Sterling, 1984; Dishman, 1982c, 1985; Dishman, Ickes, &
Morgan, 1980; Gunn & Zwingmann, 1984; Eagan, 1984; Haskell &
Blair, 1980; Heaps, 1978; Muir-Gray, Young, & Ennis, 1983;

Mustafa, Garrett, & Freeman, 1984; Snyder & Spreitzer, 1984;

20

Thompson & Martin, 1984; Tucker, 1983; Van Andel & Austin,
1984).

Morgan (1985) concluded that there is "considerable
evidence attesting to the affective beneficence of vigorous
exercise" (p. 94). Morgan reported that was also the
conclusion of the participants of an exercise and mental
health workshop sponsored by the National Institute of
Mental Health in 1984 where the following consensus
statements were formulated:

1. Physical fitness is positively associated with
mental health and well being.

2. Exercise is associated with the reduction of
stress amd emotions such as state anxiety.

3. Anxiety and depression are common symptons 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,
electro-convulsive 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 and
both sexes.

8. Physically healthy people who require
psychotropic medication may safely exercise when
and medications are titrated under close medical
supervision. (p. 95)

Adherence Problems

 

Clearly, several studies have provided suggestive
eVidence of the benefit of vigorous exercise. Health care

PrOfessionals and the public have appeared enthusiastic to

21

prescribe and utilize exercise to derive such benefits.
Despite the enthusiasm and evidence of benefit, adherence
problems have continued (Albinson & Moore, 1983; Brownell,
1984; Carmody, Senner, Malinow, & Matarazzo, 1980; Danish,
1983; Dishman 1981, 1982a, 1982b, 1982c, 1985; Dishman et
al., 1985; Falls et al., 1982; Gayle, Eckhoff, Mogel, &
Rodnick, 1984; Goodrick, Warren, Hartung, & Hoepful, 1984;
Haskell & Blair, 1980; Keefe & Blumenthal, 1980; Laffrey &
Isenberg, 1983; Paolone, 1983; Sachs, 1980; Serfass &
Gerberich, 1984; Slenker, Price, Roberts, & Jur 1984; Snyder
& Spreitzer, 1984). Exercise program dropouts have remained
at 30-70% during the last decade (Falls et al.) even though
the evidence for physiological and psychological benefits of
exercise has recruited many new exercisers. Lack of
adherence to health-promoting behaviors and lack of
compliance with treatment programs are well known problems
in the health care field (DiMatteo & DiNicola, 1982). These
problems share many similarities with the exercise dropout
problem (Dishman, 1982c). Non-compliance with diabetes,
obesity, and smoking management programs is a serious
problem (Surwit, 1983). Inability or unwillingness to
merely finish a full course of antibiotics to treat an
infection is also common.

The emphasis on the consequences of exercise in
Previous research has obscured the problem of sustained
involvement. Dishman (1982c) suggested exercise benefit is

El moot question if people continue to drop out. An

22

important concern is the discovery of factors that

contribute to exercise adherence. Studying factors that are

related to adherence to an exercise program will promote our

understanding of persistence in
understanding could enhance our
adhere to an exercise program.

factors may also be utilized by
the related issue of compliance

treatment.

such activity. That
ability to help people
Information about adherence
those researchers examining

to medically prescribed

Chapter III

The Present Study

 

Multivariate Relationships in Exercise Adherence and
Performance

 

Review of the literature thus yielded a number of
social, psychological, and physical variables that have been
shown to affect persistence in and/or accomplishment of a
behavior. A variety of theoretical and empirical links have
been made among these variables as their effects on
adherence and/or accomplishment of a behavior have been
studied. Examination of the posited links suggested that
the variables are related to one another and that their
interrelations in turn influence adherence and/or
accomplishment of various behaviors. The clearest
expositions of the paths of these inter-relationships were
found in the writings of Bandura (1977, 1982), Bandura and
Adams (1977), Dishman (1982a, 1982b,1982c), Feltz (1980),
Feltz and Mugno (1982), and Garland (1985). A brief summary
of those theoretical/empirical connections follows. Several

researchers provided evidence that social support and

 

empathy influence persistence in an activity (Bandura, 1982;
Heinzelmann & Bagley, 1970; Hurtado et al., 1973; Nelson et
al., 1975). Bandura (1982) reported a significant
relationship of significant others' beliefs about patients'
self-efficacy to exercise adherence in cardiac

23

24

rehabilitation. Support of significant others may provide
information that one uses to bolster one's self—concept
(Hirsch, 1981) or strengthen one's self-efficacy. Another
source of information regarding self-concept and self-

efficacy is one's physical state or one's perceived physical

 

spapg. Olszewska's (1982) data regarding self-concept of
exercisers supported the theory that self-concept influences
performance. Sonstroem's theory predicted that improved
self-concept would lead to further physical activity.
Bandura's (1977) theory predicted perceived physical states
would influence performance indirectly through self—
efficacy. Feltz' (1980) and Feltz and Mugno's (1982) data
regarding springboard diving behavior supported this aspect
of Bandura's theory. Physical states have a direct impact
on adherence and performance of a physical activity (Dishman
& Gettman, 1981). One's fitness level sets limits on the
amount of activity one can physically tolerate.

Egalp and goal setting also have an impact on
self-efficacy (Bandura & Schunk, 1981; Schunk, 1984;
Garland, 1985). Goals reflect one's cognitions about one's
ability (Garland, 1985) and one's physical state. Goal
setting and achievement have been posited to influence self-
efficacy directly (Garland, 1985) or through anticipated
satisfaction with performance achievements (Bandura &
Schunk, 1981). Goals may also influence adherence as they
interact with other personality variables (Tu & Rothstein,

(1979). The level of anticipated satisfaction with one's

25

accomplishments will affect motivation (Bandura, 1977)

 

persistence in and performance of an activity (Bandura,
1977, 1982; Bandura & Schunk, 1984; Garland, 1985). Self-
efficacy affects self-motivation, adherence, and.performance
in that activity (Bandura, 1977, 1982; Bandura & Schunk,
1984; Roberts et al., 1981; Weinberg et al., 1981).

Adherence or persistence in an activity has often been
treated as an integral part of performance. While it is
true that a close relationship exists, it is also important
to conceive of and measure adherence and performance
separately wherever possible. Separate measures will both
(a) clarify the boundaries of the two closely related
phenomena and (b) allow a methodology with which to study
both the influence of the many antecedent variables on
adherence and performance and the relationships of adherence
and performance to each other.

The preceding discussion served as a rationale for the
model that was the focus of this study. Dishman (1982c)
suggested that an important statistical procedure in
exercise research is a multivariate analysis of the several
variables posited to influence exercise adherence. It
appeared that a multivariate model that posited the
interrelationships of several of the variables reviewed
above and their relationships to adherence 229 performance
was necessary. Such a model was developed and tested. The
variables chosen to be examined and their definitions are

listed below.

26

Variables and Their Definitions

 

Physical fitness, the physical state, capacity, or

 

ability of the individual.

Perceived Fitness, how one views one's physical

 

fitness.

Social Support, the caring and encouragement the

 

individual receives from the social environment.
Empathy, the understanding the individual receives from
the social environment.

Self-concept, attitudes and beliefs about one's self.

 

Goals, self-set standards for frequency of training.
Self-efficacy for Performance, the belief that one can
successfully perform a specific behavior required to produce

a specific outcome.

Self-efficacy for Adherence, the belief that one can

 

successfully continue to perform a specific behavior over
time.
Anticipated Satisfaction with Performance, the belief
regarding how pleased one will be with one's performance.
Anticipated Satisfaction with Adherence, the belief
regarding how pleased one will be with one's adherence.

Self-motivation, a general disposition or attitude

 

toward perseverance.

Performance, the quality of the exercise behavior.

 

Adherence, the persistence of effort in continuing to

 

perform a task.

27

The hypothesized models of the predicted relationships
of these variables following the rationale above are
presented in diagramatic form in Figures 1 and 2. Each
arrow represents a predicted path of influence. -Fer
example, the arrow extending to self-concept from social
support represents the prediction that self-concept is
directly influenced by social support (see Hypothesis 2
below). When such a network of causal relationships is
posited, a traditional experimental paradigm is difficult to
employ.

Path analysis was selected as the primary technique in
the analysis of this model. Path analysis allows the
investigation of the various influences where experimental
paradigms are difficult to design. Path analysis tested
whether the hypothesized model fits a set of data by
comparing the observed relationships among the variables
with the predicted relationships. Path analysis also
allowed an estimation of the relative direct and indirect
effects (Feltz, 1980). Duncan (1975) stated that the
advantages of using path analysis lie in

...(1) making our arguments consistent
(so that we are not altering our premises
surreptitiously in the course of a discussion),
(2) making our conclusions precise (so that it is
easier to see what evidence is, and what is not,

compatible with them), and thereby (3) rendering

228

UUSQEHOHHOQ

 

 

oucouosv< ecu
soooauuouuaum

 

 

acaun>quoeluaom

 

 

 

oucouuxv<
and:

cauuuaunaunm

uuuuaeoguc<

¢

 

 

oucmEMOuuwm haw
soauauuuuugom

UUENEHONHU“
as“:
COHUUQuQﬁUQw

souoaauauc<

 

 

canoe

.H xv»: .ﬁovo: vouﬁuosuoamm

noucuwm
vo>auuu0m

.H ouswam

 

caucuum
Acouazsm

 

 

 

 

unaccounuuom

 

 

snowman

 

 

 

 

uuoaaam
kuoow

 

 

 

 

29

0.0
I.
ovo-
0.9:
o.

   

 

 

 

 

 

coco-ooooOouvo on coo-o
0.
O

o.

 

 

 

 

o....oo‘Oooooooo-ooo9.9.6.09... 609.
.0... 00...... O “"‘.
op no... coo-cocoooooaooo sot... .. so .0
o. .00... 0'0." ontchQo-QQAQQOoaoOo V .0... I. 00
no a... 0.5... o .0. o to 7" so a. . 00 00
o .oo 0-. o... o. o
O. O .0. .000 O O O
O b 00 9 O O
o O o o. c o O
. . o. o v . o
o O o o. O o
.. . .. .. ..
u n no .... . W .4
o . . . cocouuzv< n
sud: . unocuum
.... o..o... ..o......a oocouosv< ecu coauumumuuum . vo>ﬁouuom
accouuzv< v :.. ..... monogamouuaom vauunuuuuc< u -
O... 0.000. “C
c0000 5'
83932738 ... .. 3.8
I V oucmELOuuwm
saw:
vacuELOHuom ouanMONLum new acquomwmauam
suuuauuw-u~om coauaHUHu=<

 

 

 

 

.o~ xooz .Huvo: ponﬁmocuoazm .N gunman

 

mmucuam
Hauuuasm

 

 

 

acuansm

 

 

 

 

‘

unaccounudom

 

 

 

uuonaaw
amwoom

 

 

30

our conclusions susceptible of empirical

refutation (p. 13).
Path analysis is not a method for discovering causal
directions. It is a technique for testing the directions of
causation specified by a model (Kenny, 1979). The present
model was recursive, testing only one-way causal direction
hypotheses. This does not deny the two-way flow of causal
direction that could be tested with a more complex,
nonrecursive model. It is simply a reflection of the choice
to study a limited portion of a growing set of assumptions
and theories in the theory building occurring in the
exercise adherence/health compliance area. The variables
are also not exhaustive, but reflect the most viable
research and theory in this area. The number of variables
is limited to retain the power of the path analysis. Path
analysis is reduced in power by a large number of variables
and a small sample.

Diagrammatic representation of the path analysis for
the hypothesized model is presented here for greater
comprehension of the cause and effect network. However, due
to the large number of variables and possible paths,
presenting the results in such a diagram would be cumbersome
and perhaps impede comprehension. Thus the data are
presented in table form and diagrams are utilized only when

smaller amounts of information are being presented.

_

I .. FFlll~1l...,
(l I 1

 

 

31

The hypotheses involved in the path model are listed
below. Additional hypotheses, which were tested utilizing

E-tests and a MANOVA, are listed following that.

Path Model Hypotheses

The following hypotheses were tested by path analysis
at Week 1 and Week 10 in order to support or reject the
model (see Figures 1 and 2).

1. Self-concept is directly influenced by social
support, empathy, and physical fitness.

2. Physical fitness is the only predictor of perceived
fitness.

3. Goals is directly influenced by self-concept,
perceived fitness, and physical fitness.

4. Goals is the only predictor of anticipated
satisfaction with performance.

5. Goals is the only predictor of anticipated
satisfaction with adherence.

6. Self-efficacy for performance is directly
influenced by self-concept, perceived fitness, goals,
anticipated satisfaction with performance, and anticipated
satisfaction with adherence.

7. Self-efficacy for adherence is directly influenced
by self-concept, perceived fitness, goals, anticipated
satisfaction with performance, and anticipated satisfaction
with adherence.

8. Self-motivation is directly predicted by self-

efficacy for performance, self-efficacy for adherence,

 

32

anticipated satisfaction with performance, and anticipated
satisfaction with adherence.

9. Performance is directly influenced by self-
motivation, self-efficacy for performance, self-efficacy for
adherence, anticipated satisfaction with performance,
anticipated satisfaction with adherence, perceived fitness,
and physical fitness.

10. Adherence is directly predicted by self-
motivation, self-efficacy for performance, self-efficacy for
adherence, anticipated satisfaction with performance,
anticipated satisfaction with adherence, perceived fitness,
and physical fitness. Adherence was not tested until Week 10

when adherence data were available.

Other Hypotheses and Analyses

The hypotheses and analyses beyond the path analyses
were:

1. Subjects who initially set goals and were high in
self-efficacy and perceived fitness and low in anticipated
self-satisfaction with performance in the pretest would
experience greater increases in their distance on the 12-
minute run than subjects who initially did not set goals and
were low in self-efficacy and perceived fitness and high in
anticipated satisfaction with performance.

2. Subjects with greater distance increases would
experience increases in self-efficacy for performance and
perceived fitness over the 10 weeks and still experience low

anticipated satisfaction with performance. Subjects with

33

smaller distance increases would experience little or no
change in self-efficacy, perceived fitness, or anticipated
satisfaction with performance.

3. Physical fitness measures would not change
significantly from Week 1 to Week 10 (due to the short
period of time since exercise was begun).

4. A MANOVA was planned to test the data for
significant change in the 9 variables suggested by the 3

hypotheses above.

Chapter IV

Methods

Subjects

Subjects were 43 male and 42 female college students
enrolled in three sections of an exercise and conditioning
course in the Health and Physical Education (HPE) Department
of Michigan State University. The course sections were
selected by the students for schedule convenience. The
students were from 18 to 28 years old with a mean age of
20.49. Thirty-four of the students were former high school
athletes. They had been members of track and field,
football, basketball, and tennis teams. Of this group, 7
were Black Americans, 2 were foreign-born Orientals, 1 was
Mexican-American, and 1 was Native American. The remainder
were White Americans. Several were already in training for
amateur 10k races. (The classes had been changed by the HPE
Department during registration from track and field to
exercise and conditioning.) Only one student appeared
somewhat obese. Because of the large number of former and
amateur athletes, the group could not be considered a
representative sample of college students.

Due to missing data, the actual number of cases

analyzed in the first week's path analyses was 54. In
addition, 10 students dropped the class. Drop-out excuses

included an appendicitis, a broken foot, and schedule

34

35
problems. Thus the number of cases was 44 in the tenth
week. The nonpath analyses utilized a specific combination
of variables. Only cases which had no missing data for
these variables in Week 1 and Week 10 could be used. This
combination of specified cases resulted in different
numbers. The reader is advised to observe the degrees of
freedom to avoid confusion. Of the 54 cases used in Week 1,
30 subjects were women and 24 were men. They ranged in age
from 18 to 28 years old. Twenty were former high school
athletes. Of the 44 cases used in Week 10, 25 subjects
were women and 19 were men. They were from 18 to 28 years
old. Seventeen were former high school athletes. Four of
these students had been injured during the term and suffered
from problems such as strained muscles. Eighteen were ill
during the term with such illnesses as flu, mononucleosis,
and allergies. Mean class attendance for the group was

16.89 days of 18.

Method

During the first class of the term the general nature
of the research and data collection was explained and the
participation of class members was invited. Volunteers were
asked to sign a consent form allowing the collection of data
and a release of information form allowing a confidant to
verify their exercise in the follow-up measures (see Appen-
dix A). The planned follow—up subsequently failed (see end
of Methods). Measures were collected in Week 1 and Week 10

(see Appendix B). The classes met twice a week for 1 hour

36
and 15 minutes, and students were encouraged by the
instructor to exercise 2 to 3 additional times per week.
The course usually consisted of (a) pretest measures of
fitness such as resting heart rate and the Harvard Step
Test, (b) deveIOpment and implementation of individualized
conditioning programs, and (c) a posttest. The course was
altered for the present study only in that a broader range
of measures was taken, and follow-up information was to be
acquired by mail and telephone. Conditioning program
choices included running, swimming, weight lifting, and
stationary bike. After programs were developed, each class
consisted of a 15 minute stretching and warmup routine
followed by a workout in the individualized programs. The
instructor and I participated in the stretch and warm up and
were available to answer questions. I often did my own
running while the students were exercising. Course grades
were based only on attendance and a conditioning log kept by
each student (see Appendix C). The tests were performed

according to the testing schedule (see Appendix D).

Measures
Three measures of physical fitness were performed: the
Harvard Step Test, Resting Heart Rate, and Body Weight.
Harvard Step Test. The heart rate, taken 15 seconds
after stepping up and down on the Harvard Step-Test
apparatus at the rate of 22.5 complete ascents and descents

per minute for 5 minutes, provided a measure of aerobic

37

fitness. The apparatus is simply a box 40 cm in height for
men and 33 cm for women.

Resting Heart Rate. An average of 5 heart rate
measures taken on consecutive days before rising out of bed
in the morning was utilized for the pretest and post-test.
The student was instructed to feel the pulse on the wrist,
count the number of beats in 10 seconds and multiply by 6.
The 5 rates were averaged during the analyses.

Body Weight. Each student was to be weighed on a

 

standard scale and their weight was recorded. A Pearson
Correlation indicated that the three physical measures,
Harvard Step Test, Resting Heart Rate, and Weight were not
well correlated (r = -.39 for Harvard Step Test and Resting
Heart Rate, r = .22 for Weight and Harvard Step Test,
r = -.044 for Resting Heart Rate and Weight). In addition
to the lack of correlation, the Resting Heart Rate was
regarded as unreliable because students admitted forgetting
to take it before rising or averaging days to give readings
for forgotten days. The standard scale was unavailable and
the weights were provided by students weighing themselves on
a variety of scales. The Harvard Step Test is a well known,
respected measure of perceived fitness. Therefore only the
Harvard Step Test was used for the physical fitness
variable.

Perceived Social Support Inventory. The social support
variable was derived by averaging the 6 items of the social

support scale (see Appendix A). A six item Likert-like

38

measure of perceived social support assessed the degree to
which subjects believed close associates supported their
exercise behavior. This measure was examined for
readability and for test-retest reliability by administering
it to ten University secretaries participating in exercise
programs outside the University. The measure was
administered on two occasions, ten days apart. Test—retest
reliability on the pilot sample equalled 73.3% exact
agreement. That is, 73.3% of the items were answered on the
second occasion exactly as they had been on the first. The
measure of reliability (Cronbach's alpha) of the social
support scale was .86 in Week 1 and .88 in Week 10.

Perceived Empathy Inventory. The empathy variable was
derived by averaging the 3 empathy questions. This 3 item
Likert-like measure examined the degree to which subjects
believed the instructor understood, cared about, or
encouraged their efforts. Test-retest reliability on the
pilot sample equalled 63.4% (N = 9). The empathy inventory
achieved an Alpha of .97 in Week 1 and .73 in Week 10.

A Pearson Correlation was calculated between the social
support and empathy variables (r = .33 in Week 1 and r = .62
in Week 10). Since the two variables were not highly
correlated they remained separate variables in the analyses.
The criterion level for this and all other analyses was p 5
.05.

Tennessee Self-Concept Scale: Physical Self Subscale.

The physical self-concept variable was the total score of

39
the 18 items of the physical self subscale of the Tennessee
Self-Concept Scale (TSCS). The TSCS is a 100 item scale
that has been reported to reliably assess the
multidimensional aspects of self (Brown et al., 1982; Fitts,
1965). Trujillo (1983) suggested the TSCS was among the
most frequently used means for assessing self-concept.
Brown et al. reported its use in several studies relating
physical fitness and self-concept. Suinn (1972) concluded
the TSCS was among the better measures that combined group
discrimination with self-concept information. Test-retest
reliability for the subscale reported by Brown et al. was
.92. In the present study the physical self-subscale
measured .50 on Cronbach's Alpha, the lowest reliability of
all the measures used. Items 3, 11, and 12, which had
negative correlations to the total score and contributed
most to the low alpha, were reviewed for coding error and
the possibility of reverse scoring due to misunderstanding
of the item by the subjects. This appeared not to be the
case. To remain consistent with the previous research use
of the instrument, the decision was made to use the scale as
written.

Perceived Fitness Scale. The perceived fitness
variable was derived by averaging the 2 perceived fitness
scores. Pearson correlations revealed they were highly
correlated measures (r = .86 in Week 1, r = .91 in Week 10).
This scale consists of a two-item Likert-like scale

assessing the subjects' estimations of their level of

40

physical fitness. Test—retest reliability on the pilot
sample equalled 70% exact agreement (N = 10). Because there
were only 2 items, Pearson correlations were calculated.

The items correlated E = .86 in Week 1 and p = .91 in Week
10.

Egalg. The measure for the variable, goals, was the
number of times per week the student planned to exercise.
This goal appeared in the plan sheet required by the
instructor.

Anticipated Satisfaction with Performance. The
anticipated satisfaction with performance and the
anticipated satisfaction with adherence variables were
single estimations of percent satisfaction and needed no
further treatment. The Anticipated Satisfaction with
Performance measure asked subjects how satisfied or pleased
with performance they believed they would be. This rating
was made on a 100-point scale ranging from very satisfied to
completely dissatisfied.

Anticipated Satisfaction with Adherence. The
Anticipated Satisfaction with Adherence measure asked
subjects how satisfied or pleased with adherence they
believed they would be. This rating was made on a 100-point
scale ranging form very satisfied to completely
dissatisfied.

Self-efficacy for Performance. The self-efficacy for
performance variable was derived by averaging the 16 self-

efficacy for performance scores. This measure assessed the

41
percent of certainty with which subjects believed they would
complete a given number of quarter-mile laps around a track.
The measure was a grid listing sixteen 1/8 mile segments
along one axis and percent certainty from 0-100 in
increments of 10 along the other axis (see Appendix A).
Test-retest reliability in the pilot was 90 % exact
agreement.

Self-efficacy for Adherence. Self-efficacy for

 

adherence was derived by averaging the 7 self-efficacy for
adherence scores. This measure assessed the percent of
certainty with which the subject believed they would
continue their exercise program over the 6 months following
the class. The measure was a grid listing 1 to 7 days per
week along one axis and 0-100 % along the other axis (see
Appendix A). Test-retest reliability in the pilot was 80 %
exact agreement.

Eastman and Marziller (1984) and Marziller and Eastman
(1984) suggested both theoretical and methodological
difficulties plague self-efficacy theory. They maintained
efficacy expectations were not unambiguously differentiated
from outcome expectations. Further, Bandura and his
colleagues have not made clear what is being assessed. The
construction and exploration of self-efficacy scales is a
particularly thorny methodological problem requiring
attention in future research (Eastman & Marziller, 1984).

The present study addressed Eastman and Marziller's

criticisms. It drew on Garland's (1985) model which was

42

conceived independently of Bandura's self-efficacy concepts.
Garland measured separately performance expectancy, similar
to Bandura's self-efficacy, and performance valence or
anticipated satisfaction with performance. Satisfaction
with performance was thus differentiated from self-efficacy.
The present study measured self-efficacy and anticipated
satisfaction with performance to make that same distinction.

To improve the clarity of the self-efficacy measure, a
definition of self-efficacy in Ehig situation was used in
the scale intructions. The self-efficacy scales read from
0-100 instead of 10-100, and the verbal labels that Eastman
and Marziller (1984) suggested were misleading did not
appear. The scale was used only as a measure of the
subjects' beliefs about their ability to perform or adhere.
Specific use was made of the words, performance and
situation. Outcome expectancy words, such as fitness,
better figure, better condition, and words such as
probability, which Eastman and Marziller suggested confused
the meaning of the scale, were avoided.

Dishman's Self-Motivation Inventory. The self-
motivation variable was derived from the total score on
Dishman's Self-Motivation Inventory. This 7-item Likert-
like inventory assessed the subject's tendency to persevere
or be motivated. The Self-Motivation Inventory achieved

Cronbach's Alpha of .70 in Week 1 and .74 in Week 10.

43

Twelve Minute Run . The performance variable was the

 

number of laps and quarter laps on the 1/8 mile track the
subject ran in 12 minutes.

Adherence. The adherence variable was calculated at

 

Week 10 by averaging the frequency of exercise reported by
the participant in Weeks 4 through 10.

Other preliminary data treatment. Change in distance,

 

self-efficacy, and perceived fitness scores required for the
MANOVA were calculated by simple subtraction. Pearson
Correlations were performed with the variables in each path
analysis and are presented in Tables E-l and E-2. These
correlations were examined for intercorrelations and
independence of measures. The same correlations are
referred to in the path analyses.

Failure of the Follow-up. During the term the parti-

 

cipants were told their efforts would be rewarded by plac-
ing the names of all persons who completed all forms in a
lottery for two sport watches. The drawing was held after
both follow-ups were completed. The first follow-up was
due 10 weeks after the term ended. Letters were mailed a
week in advance to the 85 subjects who began the class and
to 85 confidants they had named to remind them to return
the forms. Extra forms for the previous ten-week period
and new forms for the next ten-week period were included
along with stamped return envelopes. A week after the due
date an effort was made to locate by telephone all every-

cnae who had not returned a form. Phone calls were made to

 

44

the numbers given by the subjects. Public telephone Opera-
tors in several states and university Operators were con-
tacted for updated numbers. Calls were placed several
times a day to try to locate the subjects at a convenient
time. Information forms were completed by telephone if

the subjects said they didn't have the time to complete

and mail the form. Seventeen forms were completed. Path
analysis with this low number and several variables could
not be justified.

At the time of the final follow-up, many students had
returned to the University. Again 85 letters asking for
the second follow-up form and 85 stamped return envelopes
were sent to the subjects. Eighty-five letters and stamped
envelopes were also sent to the confidants. A slightly
higher return that was probably due to the proximity of
the students occurred. Once again, several attempts were
made to call everyone who had not returned a form. This
time 26 of the 85 forms were completed, still far to few
to justify the use of path analysis, especially with a
large number of variables. After conferring with my path
analysis consultant and my committee chair and with much
sadness, I agreed that it was not appropriate to perform
these final analyses. The longer term adherence informa-
tion I had hoped to gain would have to wait for a replica-

tion with a more stable population.

Chapter V

Results

Path Analyses

 

Before proceeding with the path analyses it is
important to remind the reader of the small sample size in
the present study and offer a warning regarding the
interpretation of the results. The use of path analytic
techniques with a small sample and a large number of
predictors is problematic (see Kerlinger & Pedhazur, 1973).
The reader is advised to use some caution in interpreting
the results of these analyses. The means and standard
deviations for the variables in the path models are
presented in Table 3. The correlations of the variables are
shown in Tables E-l and E-2.

Both the Week 1 and Week 10 models were considered to
be recursive. This means that causal flow is
unidirectional. A variable cannot be the cause and the
effect of another variable at the same time. A recursive
model permitted the use of the ordinary least squares
solution to estimate path coefficients in each multiple
regression analysis. Table 4 displays the total path
analysis for the hypothesized and full models. The path
analysis contained 9 regression analyses in Week 1, and 10

in Week 10, one for each criterion variable.

45

46

.oocouo;v< now Succuwuocuuow I cum

.oucnauOUuoa new Auuuauuouuuom u mum

.0ucouo£v< sud: acuuuuuuuuum vouna«u«u:< I <m<

.oucuILOuuom xuﬁa cauuuquoauam vouua«u«u:< I mw<

 

 

 

 

 

 

 

 

 

 

Nu.“ m~.~ 05.: n~.o~ ~n.- mo.w~ No.m~ am.~ ea. ~o.o ne.n mm.“ n~.~ am
OH

no.n no.n~ no.n~ ~n.mo cc.- ~0.~n mn.mn m~.m ﬁe.n no.ma o~.om w~.m mN.N :

In ne.~ o~.¢ ma.- nm.a~ om.q~ 05.0“ cm.~ be. ~n.» no.m so.~ -.~ on
u

nu nmuw— No.a~ m~.o~ we.we ~¢.Oo o—.an m~.m nn.n no.ma e~.cn ~o.~ om.~ :
oucouvsv< uucnsuOHLom ecuuu>auos <mw mum <m< mm< cameo unocuum unusuam mayocounuaom Asuamlu uuommam xvos

.28 .c n N a 833:... 132:: 18%.: a: 8m

 

.mmnmﬂw:< :uom :« cowbouuo> new accuuua>~a vuuvcmuw

was occur .n «ﬁnch

47

The full model contains all possible paths between
the variables. The techniques used to compare the ability
of the hypothesized model and the full model (see Table 4)
to explain the correlations shown in Tables E-l and E-2 were
(a) a Chi-square goodness of fit statistic testing the total
path models, (b) a "Q" coefficient, testing the total path
models and (c) a comparison of individual multiple
regression equations within each path model by means of an F
-test (Ott, 1977). The computational formulas for the "Q"
coefficient and the F—test are found in Appendix G.

A Chi-square goodness-of-fit statistic was computed to
examine how well the hypothesized model fit the data in
comparison to the full model. As stated previously, the
full model contains all possible pathways among the
variables. Thus, it can explain all the correlations among
the variables. If the Chi-square was nonsignificant the
data fit the hypothesized model. Therefore, the Chi-square
test also represents the overall ability of the hypothesized
model to account for the correlations in Tables E-1 and
E-2. The degrees of freedom for the Chi-square test are
equal to the number of path coefficients hypothesized to be
zero in the hypothesized model. For example, in Week 1 the
degrees of freedom equal 31 since 31 possible paths are
omitted in the hypothesized model (see Table 4). Thirty-one
additional path arrows could be drawn on Figure 1 to

represent those paths omitted on the hypothesized model.

4E3

Table 4. Standardized Regression Coefficients for Multiple Regression Analysis.

 

 

 

 

Hypoth- yﬁypothesizsdinodel Full Model
e313 Predictor Variable Criterion Variables Week 1 Week 10 Week 1 Week 10
1 Social Support .058 .014 .058 .014
Empathy Self-concept -.221 -.128 -.221 -.128
Physical Fitness .267 .112 .267 .112
2 Physical Fitness .625* *ﬁ .493* .618* .467*
Social Support Perceived Fitness .229* .151
Empathy -.031 .096
Self-concept -.241* .012
3 Self-concept -.277 -.104 -.274 -.072
Perceived Fitness -.043 -.042 -.058 -.048
Physical Fitness Goals .174 .192 .170 .239
Social Support .039 -.106
Empathy .029 .214
4 Goals -.032 -.103 -.001 -.112
Self-concept l ** ** .103 -.034
Perceived Fitness ASP .422* .560*
Social Support .166 .063
Empathy -.333 -.053
Physical Fitness -.273 -.022
5 Goals .028 -.077 ** —.019 -.107
Self-concept 2 -.173 .059
Perceived Fitness ASA .125 .617*
Social Support .366 -.007
Empathy -.235 .038
Physical Fitness .001 -.011
6 Self-concept -.071 -.033 -.112 -.039
Perceived Fitness -.015 ~.387* -.075 -.325
Goals Self Efficacy -.038 .191 -.046 .244
aspl for Performance .190 .363 .210 .325
1542 .077 .037 .130 .121
Social Support —.245 .068
Empathy .009 -.144
Physical Fitness .222 -.152
7 Self-concept .019 -.174 -.101 -.158
Perceived Fitness .195 .117 -.027 .066
Goals Self Efficacy .375* ** .554* .345* .554*
ASPl for Adherence -.200 .246 -.139 -.263
ASAZ .438* .584* .294* .590*
Social Support .392* .269*
Empathy -.051 .143

Physical Fitness .137 -.039

459

Table 4 (Continued)

 

 

 

 

 

Hypoth- ﬁypothesized Model .2 Full updel
sis Predictor Variable- Criterion Variables Week 1 Week 10 Week 1 Week 10
8 EP2 .280* -.108 .224 —.058
EA 1 ' -.047 .393* .058 .484*
ASP2 .233 .444* .183 .442*
ASA .242 .055 .256 -.028
Goals Self-motivation -.080 -.156
Self-concept ' .039 .091
Perceived Fitness .137 -.058
Social Support —.172 .061
Empathy -.104 .043
Physical Fitness .003 .162
9 Self-motivation .073 .108 .039 .183
£22 .004 -.235 —.017 -.302*
EA 1 -.210 -.043 -.122 -.250
ASP2 -.066 .087 -.109 .098
ASA .242 .113 ** .379* .206
Perceived Fitness Performance .435*' ** .180 .623* .199
Physical Fitness .294* .330 .130 .240
Self-concept .384* .102
Social Support -.328* -.123
Empathy .127 -.093
Goals .161 .395*
10 Se f-motivation .120 .272
EP -.100 -.200
EA“1 .500* .020
ASP2 -.028 -.061
ASA -.152 ** .089
Perceived Fitness Adherence -.106 -.085
Physical Fitness .135 .032
Self-concept -.019
Social Support -.155
Empathy -.000
Goals .667*
1Anticipated Satisfaction with Performance.
2Anticipated Satisfaction with Adherence.
3Efficacy for Performance.
4

Efficacy for Adherence

* - Significant (p 2 .05) for individual path coefficients.

** Significant
regressions.

F
§_(p 2 .05) for comparison of hypothesized and full model individual

50

Since the Chi-square test is sensitive to sample size,
a nonsignificant Chi-square could still result with a
relatively poor fit of data, indicating in error that the
model fit the data. Pedhazur (1982), therefore, suggests
also conducting a "Q" coefficient test for the model that is
not affected by sample size. The “Q" coefficient represents
the ratio variance "explained" by the hypothesized model
relative to that explained by the full model. ”Q" varies
from 0 to 1. Values closer to 1 indicate that the
hypothesized model can explain nearly all explainable
variance in the dependent variables.

After these overall comparisons were made, the
individual multiple regression analyses (listed in Figure 3)
between the hypothesized and full models were examined for
significant differences between the models. For instance,
in the Week 1 hypothesized model, physical fitness was
considered to be the only predictor of perceived fitness in
‘the second group of variables listed in Figure 3. Social
support, empathy, and self-concept variables were
hypothesized to be zero pathways in the model. That is,
they were not considered to be significant predictors of
perceived fitness. The hypothesized regression equation is
then tested against the full regression equation that
includes social support, empathy, and self-concept as
predictor variables by means of an ﬁttest (Ott, 1977). If
the ﬁfteSt is not significant, than the hypothesized

Predictor variable, physical fitness, adequately accounts

Figure 3. Predictor and Criterion Variables for Multiple
Regression Equations in Week 1 and Week 10 Models.

 

Hypothesized Model

Predictor Variables

Full Model

Predictor Variables

Both Models'
Criterion Variables

 

Social Support
Empathy
Physical Fitness

Physical Fitness

Self-Concept
Perceived Fitness
Physical Fitness

Goals

Goals

Self-Concept
Perceived Fitness
Goals

ASP

ASA

Self-Concept
Perceived Fitness
5

ASP
ASA

Social Support
Empathy
Physical Fitness

Physical Fitness
Social Support
Empath
Self-Concept

Self-Concept
Perceived Fitness
Physical Fitness
Social Support
Empathy

Goals
Self-Concept
Perceived Fitness
Social Support
Empathy

Physical Fitness

Goals
Self-Concept
Perceived Fitness
Social Support
Empathy

Physical Fitness

Self-Concept
Perceived Fitness
Goals

ASP

ASA

Social Support
Empathy

Physical Fitness

Self-Concept
Perceived Fitness
5

ASP

ASA

Social Support
Empathy

Physical Fitness

SEP
SEA
ASP
ASA

Goals
Self-Concept
Perceived Fitness
Social Support
Empathy

Physical Fitness

Self-Concept

Perceived Fitness

Goals

SEA

Self-Motivation

 

Figure 3 (Continued)

52

 

Hypothesized Model
Predictor Variables

Full Model
Predictor Variables

Both Models'
Criterion Variables

 

Self-Motivation
SEP

SEA
ASP

ASA
Perceived Fitness
Physical Fitness

Self-Motivation
SEP

SEA
ASP

ASA
Perceived Fitness
Physical Fitness

Self—Motivation
SEP

SEA

ASP

ASA

Perceived Fitness

Physical Fitness

Self-Concept

Social Support
pathy

Goals

Self-Motivation
S P

Perceived Fitness
Physical Fitness
Self-Concept
Social Support
Empathy

Goals

Performance

Adherence

 

1
2
3
4

Anticipated Satisfaction with Peformance
Anticipated Satisfaction with Adherence
Self-Efficacy for Performance
Self—Efficacy for Adherence

53
for the variablity in perceived fitness without the need for
the additional predictor variables of social support,
empathy, and self-concept.

Examination of the overall fit of the hypothesized
model to the data by the chi-square goodness-of-fit test
resulted in nonsignificant values in both Week 1, X2(31) =
30.74, and Week 10, x2(35) = 19.18. This would ordinarily
indicate a good fit to the hypothesized model. However,
since the present study has a small sample size, the "Q“
test is a better test of fit. The "Q" ratios were: Week 1,
Q = .263, and Week 10, Q = .119, indicating that the fit was
poor in both Weeks.

The next set of analyses was a comparison of individual
multiple regression equations by means of an F—test. The
hypotheses upon which the equations for the hypothesized

model were constructed are stated below.

Path Model Hypotheses

1. Self-concept is directly influenced by social
support, empathy, and physical fitness.

2. Physical fitness is the only direct predictor of
perceived fitness.

3. Goals are directly influenced by self-concept,
perceived fitness, and physical fitness.

4. Goals is the only predictor of anticipated

satisfaction with performance.

 

54

5. Goals is the only predictor of anticipated
satisfaction with adherence.

6. Self-efficacy for performance is directly
influenced by self-concept, perceived fitness, goals,
anticipated satisfaction with performance, and anticipated
satisfaction with adherence.

7. Self-efficacy for adherence is directly influenced
by self-concept, perceived fitness, goals, anticipated
satisfaction with performance, and anticipated satisfaction
with adherence.

8. Self-motivation is directly predicted by self-
efficacy for performance, self-efficacy for adherence,
anticipated satisfaction with performance, and anticipated
satisfaction with adherence.

9. Performance is directly influenced by self-
motivation, self-efficacy for performance, self—efficacy for
adherence, anticipated satisfaction with performance,
anticipated satisfaction with adherence, and physical
fitness.

10. Adherence is directly predicted by self-
motivation, self-efficacy for performance, self-efficacy for
adherence, anticipated satisfaction with performance,
anticipated satisfaction with adherence, and physical
fitness.

Each path hypothesis was tested by performing a
multiple regression analysis for the hypothesized and full

models in Week 1 and Week 10. The full and hypothesized

55
versions of the multiple regression equations for all
hypotheses except Hypothesis 1 were then tested using an F-
test. The results of the E—tests follow:

Hypothesis 1 stated that social support, empathy, and
physical fitness directly predict physical self-concept and
is illustrated in Figure 4. The regression equations in the
hypothesized and full models were the same for this
hypothesis so no Fetest was needed for comparison. None of
the path coefficients leading to self-concept were
significant for either Week 1 or Week 10, thus refuting the
first hypothesis. As can be observed from Table 4, the path
coefficient decreased in strength from Week 1 to Week 10.
These results failed to support the assumption of this study
that information about the self that one gains from such
sources as one's own physical fitness would predict initial
physical self-concept measures. Other studies have found
increases in self-concept resulting from improved physical
fitness (Brown, 1982; Sonstroem, 1984; Trujullo, 1983). The
results also failed to support Hirsch's (1981) contention
that social support would influence self-concept. Nor did
the present study's contention that empathy would predict
self-concept receive support.

Hypothesis 2 stated that physical fitness is the only
direct predictor of perceived fitness and is illustrated in
Figure 5. The full equations for Week 1 and Week 10
included social support, empathy, and self-concept as

additional predictors of perceived fitness. The resulting

56

Social Support.

Empathy Self-concept

 

Physical Fitness

FIGURE 4. HYPOTHESIS 1 ILLUSTRATED.

(Connecting line indicates hypothesized predictor.)

 

1. 1.41. .. (1.14 111131 ..1 1.1111(3“

 

 

57

Physical Fitness

Social Support \\\\\\‘Perceived Fitness
Empathy

Self-concept

FIGURE 5. HYPOTHESIS 2 ILLUSTRATED.

58

F—tests (see Table G-5) indicated a significant difference
between the full and hypothesized equations for the Week 1
models but not for the Week 10 models. This significant
difference indicated that there were significant path
coefficients other than physical fitness that should be
examined in the full equation. In examining the full model
for Week 1 (see Table 4), it can be seen that social support
and self—concept were also significant predictors of
perceived fitness, but self-concept was in the opposite
direction than that hypothesized. A specific aspect of
physical self-concept, perceived fitness was influenced by
social support, thus providing support for Hirsch's (1981)
social support theory and Sonstroem's (1974, 1978) exercise
adherence theory. Physical self-concept had a negative
influence on perceived fitness. The low correlation between
them in Table E-l suggests the two variables did measure
separate constructs. At Week 10 (see Table 4), physical
fitness was the only significant predictor of perceived
fitness, fully supporting the hypothesis. Physical fitness
significantly predicted perceived fitness at both Week 1 and
Week 10, supporting Sonstroem's finding that fitness
influences self-perception of fitness or athletic ability.

Hypothesis 3 stated that self-concept, perceived
fitness, and physical fitness directly predict the extent to
which one sets goals and is illustrated in Figure 6. The
full equations for Week 1 and Week 10 included social

support and empathy as additional predictors of goals. The

 

59

Self-concept

Perceived Fitness

 

Physical Fitness Goals
Social Support
Empathy

FIGURE 6. HYPOTHESIS 3 ILLUSTRATED.

 

60

resulting F—tests (see Table G-5), indicated no significant
differences between the full and hypothesized equations for
either Week 1 or Week 10. None of the path coefficients
leading to goals were significant for either Week 1 or Week
10, thus supporting neither the third hypothesis nor the
assumption of this study that exercise goals would be
predicted by actual fitness and perception of one's self and
fitness.

Hypothesis 4 stated that goals would be the only
predictor of anticipated satisfaction with performance and
is illustrated in Figure 7. The full equations for Week 1
and week 10 included self-concept, physical included self-
concept, perceived fitness, social support, empathy, and
physical fitness as additional predictors of anticipated
satisfaction with performance. The resulting F—tests (see
Table G-5), indicated a significant difference between the
full and hypothesized equations for Week 1 and Week 10
models. This significant difference indicated that there
were significant path coefficients other than goals that
should be examined in the full equation. In examining the
full model for Week 1 and Week 10 (see Table 4), it is
apparent that goals was not a significant predictor whereas
perceived fitness was a significant predictor of anticipated
satisfaction for performance in both weeks. A substantial
increase in the strength of the path coefficient for
perceived fitness can also be observed. Since goals did not

significantly predict anticipated satisfaction with

 

61

 
   
 

Self-concept

Perceived Fitness

Social Support

Empathy

Physical Fitness
FIGURE 7. HYPOTHESIS 4 ILLUSTRATED.

1

ASP Anticipated Satisfaction with Performance

 

 

62
performance, Garland's (1985) assumption that anticipated
satisfaction with performance mediates between goals and
performance was not supported. Hypothesis 4 was not
supported. That perceived fitness predicted anticipated
satisfaction with performance lends more support to
Sonstroem's (1974, 1978) contention that self-concept or
perceived athletic ability affects performance and attitudes
toward performance.

Hypothesis 5 stated that goals is the only direct
predictor of anticipated satisfaction with adherence and is
illustrated in Figure 8. The full equations for Week 1 and
Week 10 included self-concept, perceived fitness, social
support, empathy, and physical fitness as additional
predictors of anticipated satisfaction with adherence. The
resulting F—tests (see Table G-5), indicated no significant
differences betweeen the full and hypothesized equations for
Week 1. There were no significant path coefficients in
either the hypothesized model or the full model for Week 1.
Therefore the hypothesis was not supported nor were any of
the variables significant precictors of anticipated
satisfaction with adherence.

There was, however, a significant difference between
equations in Week 10, indicating that there were
significant path coefficients other than goals that should
be examined in the full equation. In examining the full
model for Week 10, (see Table 4), it can be seen that

Perwzeived fitness was the lone significant predictor of

 

63

  

Perceived Fitness
Social Support ASA
Empathy

Physical Fitness

FIGURE 8. HYPOTHESIS 5 ILLUSTRATED.

 

64
anticipated satisfaction with adherence. Thus Hypothesis 5
was also not supported in Week 10 and perceived fitness was
the lone significant predictor of anticipated satisfaction
with adherence in Week 10. Goals did not predict
anticipated satisfaction with adherence, failing to support
this study's assumption that anticipated satisfaction with
adherence mediates goals and adherence. That perceived
fitness became a very strong predictor of anticipated
satisfaction with adherence provides strong support for
Sonstroem's (1974, 1978) theory that perceived fitness
increases with training and increases positive attitudes
toward continued physical activity.

Hypothesis 6 stated that self-concept, perceived
fitness, goals, anticipated satisfaction with performance,
and anticipated satisfaction with adherence directly
influence self—efficacy for performance. Hypothesis 6 is
illustrated in Figure 9. The full equations for Week 1 and
Week 10 included social and Week 10 included social support,
empathy, and physical fitness as additional predictors of
self-efficacy for performance. The resulting F-tests (see
Table G-5), indicated no significant differences between the
full and hypothesized equations for either Week 1 or Week 10
models. In Week 1 (see Table 4), none of the path
coefficients leading to self-efficacy for performance were
significant, thus not supporting the hypothesis. In the
hypothesized model for Week 10, (see Table 4), it can be

seen that perceived fitness was the only significant

65

Self-concept
Perceived Fitnes

Goals

1

ASP EA

 

ASA2

Social Support
Empathy
Physical Fitness

FIGURE 9. HYPOTHESIS 6 ILLUSTRATED.

Anticipated Satisfaction with Performance
Anticipated Satisfaction with Adherence
Self—efficacy for Adherence

 

 

66

predictor of self-efficacy for performance, but in the
direction opposite that predicted. The results did not
provide support for that part of Bandura's (1977) self—
efficacy theory that states that cognitions about physical
condition is a strong predictor of self-efficacy. Bandura
and Schunk's (1981) and Garland's (1985) findings that goals
predict self-efficacy were not supported. The assumption of
the present study that anticipated satisfaction with
adherence and anticipated satisfaction with performance
would predict self-efficacy for performance was also not
supported.

Hypothesis 7 stated that self-concept, perceived
fitness, goals, anticipated satisfaction with performance,
and anticipated satisfaction with adherence directly predict
self-efficacy for adherence. Hypothesis 7 is illustrated in
Figure 10. The full equations in Week 1 and Week 10
included social support, empathy, and physical fitness as
additional predictors of self-efficacy for adherence. The
resulting F—tests (see Table G-5), indicated a significant
difference between the hypothesized and full equations for
the Week 1 but not for the Week 10 models. This significant
difference indicated that there were significant path
coefficients other that goals and anticipated satisfaction
with adherence that should be examined in the full model.
Examination of the full model for Week 1 and Week 10 (see
Table 4) indicates that social support was also a

significant predictor of self-efficacy for adherence.

 

67

Self-concept
Perceived Fitnes
Goals

ASP SEAl

 

ASA

Social Support
Empathy

Physical Fitness

FIGURE 10. HYPOTHESIS 7 ILLUSTRATED.

1SEA Self—efficacy for Adherence

 

. .ﬁnloa'ln‘l...‘ ‘.J.H.iltilu‘

 

 

68
Reviewing the hypothesized model for Week 1, it can be seen
that only two of the hypothesized paths, goals and
anticipated satisfaction with adherence were significant.
Thus the hypothesis was only partially supported.

The results indicate support for Bandura and Schunk's
(1981) and Garland's (1985) findings that goals predict
self-efficacy. It is important to note that goals predicted
self-efficacy for adherence but not self-efficacy for
performance. The contention of this study that adherence
and performance elements of measures should be separated was
also supported. The assumption of this study that
anticipated satisfaction with adherence would predict self-
efficacy for adherence was supported. Bandura and Schunk's
finding that social support provides information that one
uses to strengthen self-efficacy was supported. The path
was directly from social support to self-efficacy and is not
mediated by self-concept as the hypothesized model
suggested. The results failed to support the assumption of
this study that perceived fitness and physical self-concept
would predict self-efficacy for adherence attempted in this
Study. The assumption that these variables would serve as
Sources of self-efficacy information, based on Bandura's
(1977) contention that cognition about physical condition is
aStrong predictor of self-efficacy, was not supported. The
results did not support the attempted connection of

anticipated satisfaction with performance and self-efficacy

 

69
for adherence although anticipated satisfaction with
adherence did predict self-efficacy for adherence.

Hypothesis 8 stated that self-efficacy for performance,
self-efficacy for adherence, anticipated satisfaction with
performance, and anticipated satisfaction with adherence
directly influence self—motivation. Hypothesis 8 is
illustrated in Figure 11. The full equations for Week 1 and
Week 10 included goals, self-concept, perceived fitness,
social support, empathy, and physical fitness as additional
predictors of self-motivation. The resulting F—tests (see
Table G-5) indicated no significant difference between the
hypothesized and full equations for the Week or the Week 10
models. Examination of the path coefficients for Week 1
(see Table 4), indicates only one significant path
coefficient. Thus the hypothesis was only partially
supported. Self-efficacy for performance was the only
significant predictor of self—motivation. There were no
significant path coefficients in the full model for
Week 1.

Examination of the path coefficients for Week 10
reveals that self—efficacy for adherence and anticipated
satisfaction with performance are significant predictors of
self-motivation in both the hypothesized and full models.
Thus partial support of the hypothesis was indicated.
Bandura's (1977) theory that self-efficacy affects self-
motivation was supported by the results. Once again,

differing results for self-efficacy for performance and

 

 

70

ASP

 

ASA Self-motivation
Goals

Self-concept

Perceived Fitness

Social Support

Empathy

Physical Fitness

FIGURE 11. HYPOTHESIS 8 ILLUSTRATED.

Self-efficacy for Performance
Self-efficacy for Adherence

 

71
self-efficacy for adherence occurred. Self-efficacy for
adherence but not self-efficacy for performance was a
significant predictor of self-motivation, affirming the
importance of separating adherence and performance elements
of measures.

That anticipated satisfaction with performance was a
significant positive predictor of self—motivation is
contrary to Garland's (1985) finding that anticipated
satisfaction with performance (performance valence in his
words) was predictive of perseverence in a negative
direction. Lower anticipated satisfaction with performance
predicted higher performance and intensity of effort in
Garland's study. The lack of significant influence of
anticipated satisfaction with adherence on self-motivation
failed to support this study's assumption that anticipated
satisfaction with adherence would predict self-motivation.

Hypothesis 9 stated that self—motivation, self-efficacy
for performance, self-efficacy for adherence, anticipated
satisfaction with performance, anticipated satisfaction with
adherence, perceived fitness, and physical fitness directly
influence performanace. Hypothesis 9 is illustrated in
Figure 12. The full equations for Week 1 and Week 10
included self-concept, social support, empathy, and goals as
additional predictors of performance. The resulting F—tests
(see Table G-5) indicated significant differences between
the hypothesized and full equations for both Week 1 and Week

10. These significant differences indicated that there were

72

Self-motivation

SEA
SEP
ASP

ASA Performance

 

   

Perceived Fitness
Physical Fitness
Self-concept
Social Support
Empathy

Goals

FIGURE 12. HYPOTHESIS 9 ILLUSTRATED.

73
significant path coefficients in addition to those
hypothesized that should be examined in the full equations.
Four significant path coefficients appeared in the full
model for Week 1 (see Table 4) in contrast to only 2 in the
hypothesized model. In the full model, anticipated
satisfaction with adherence, perceived fitness, self-
concept, and social support were significant predictors of
performance. However, social support had a negative path
coefficient, opposite the predicted direction. Perceived
fitness and physical fitness were significant predictors of
performance in the hypothesized model for Week 1, providing
partial support for the hypothesis.

In Week 10, examination of the path coefficients shows
that physical fitness dropped out as a significant predictor
while self-efficacy for performance and goals were
significant predictors of performance. However, self-
efficacy has a negative coefficient, opposite the predicted
direction. In the hypothesized model for Week 10 only
physical fitness was a signifiant predictor of performance.

Again, the hypothesized model received only partial
support. The results failed to support the assumption of
this study that physical fitness would directly, as well as
through other paths, predict exercise performance. The
results failed to support several theoretical links that the
present study sought to test. Self—motivation did not
predict performance thus failing to support Bandura's (1977)

assumption that motivation mediates between self-efficacy

74
and performance. Bandura's theory that self-efficacy
directly predicts performance was not supported. (Both
self—efficacy for performanace and self-efficacy for
adherence failed to reach positive significance.) Garland's
(1985) contention that negative anticipated satisfaction
with performance predicts performance was supported.

That this study's construct, anticipated satisfaction
with adherence, did predict performance while anticipated
satisfaction with performance had no strong effect, provides
support for the separation of these constructs. The
positive direction of the influence is counter to what
Garland found. Other studies, however, have found
anticipated satisfaction with performance to predict
performance in a positive direction (Bandura, 1977).
Perceived fitness and physical self-concept directly
affected performance, counter to Bandura's theory that
states such information affects performance through its
effect on self-efficacy. That goals directly predicted
performance also counters Bandura's theory that self-
efficacy mediates goals and performance, but lends support
to Garland's and Tu and Rothstein's (1979) findings that
goals also directly predict performance.

Hypothesis 10 stated that self-motivation, self—
efficacy for performance, self-efficacy for adherence,
anticipated satisfaction with performance, anticipated
satisfaction with adherence, perceived fitness, and physical

fitness directly predict adherence. Hypothesis 10 is

75
illustrated in Figure 13. Adherence was not measured in
Week 1. The full equations for Week 10 included self-
concept, social support, empathy, and goals as additional
predictors of adherence. The resulting {-test (see Table
G-5) indicated a significant difference between the full and
hypothesized equations for the Week 10 model. In the
hypothesized model, self-efficacy for adherence was the only
predictor of adherence. The hypothesis was only partially
supported. The results failed to support Dishman and
Gettman's (1980) and Dishman's (1981) direct connection of
self-motivation and adherence. This study's test of
anticipated satisfaction with performance, anticipated
satisfaction with adherence, perceived fitness, and physical
fitness as direct predictors of adherence also failed to
achieve support.

Support for Bandura and Schunk's (1981) findings that
self-efficacy predicts adherence was found in the results.
That self-efficacy for adherence and not self-efficacy for
performance was a significant predictor again supports the
separation of adherence and performance elements of
measures. Goals was a direct predictor of adherence (as
well as of performance) supporting Tu and Rothstein's (1979)
and Riddles' (1980) findings that goals predict exercise

adherence.

Further Comparisons
As a furthur comparison of the hypothesized and full

models, squared multiple correlation coefficients (R25)

76

Self-motivation
SEP

SEA

ASP

ASA

Perceived Fitness

Physical Fitness 7 ‘ Adherence

 

Self-concept
Social Support
Empathy

Goals

FIGURE 13. HYPOTHESIS 10 ILLUSTRATED.

-—_--————-—-—s——---——.-——-—————--—-—---———-—————-———--——
-——-—-————-—.——-—--——.--——————-—-——--—-—-——----—--—--—

77
("explained variance") were calculated for equations that
differentially predicted performance and adherence. As can
be seen from Table 6, neither the Week 1 nor the Week 10
hypothesized models accounted for exercise performance as
well as the corresponding full models. As indicated in
Table 10, the hypothesized model did not account for

exercise adherence as well as the corresponding full model.

Summary of Path Analysis Results

The hypothesized accounted for the performance and
adherence data very pooryl compared to the full model (see
Figure 14). Figures 15 and 16 illustrate the significant
positive paths from both the hypothesized and full models.
In Week 1 (see Figure 15), performance was influenced by
physical fitness, self-concept, perceived fitness, and
anticipated satisfaction with adherence, but not by self—
motivation, self-efficacy for performance, self-efficacy for
adherence, anticipated satisfaction with performance,
empathy, or goals. Adherence was not measured in Week 1.
Self-motivation, self-efficacy for performance, anticipated
satisfaction for adherence, goals, and self—concept had no
significant predictors. Self—efficacy for adherence was
significantly predicted by goals, anticipated satisfaction
with adherence, and social support. Anticipated
satisfaction with performance was significantly influenced
by perceived fitness. Perceived fitness was significantly

predicted by physical fitness.

78

 

 

 

 

 

N

mom. mma. mwm. 05m. Nam. nmq. OH
qu. mmo. Hoe. H
monouwro< oodeuowuom successes Homo: oocmaorc< auto: x603
mucouowmwm monopoWMHQ Homo: Hash Hash Hope: caseworuoazm
couﬁmoLDOQSE
m .Hocoz Hash ecu
com Hovoz oouﬁmoﬁuomzm orb cmoBuum oodouonv< can oocmEMOMaom now m orb mo domaumuﬁoo .o manmw

 

79

 

Hypothesis

PS

 

Hypothesis 1. Self—concept is directly
influenced by social support, empathy,
and physical fitness

Hypothesis 2. Physical fitness is the
only direct predictor of perceived fit-
ness

Hypothesis 3. Goals are directly influ-
enced by self-concept, perceived fitness,
and physical fitness

Hypothesis 4. Goals is the only pre-
dictor of anticipated satisfaction with
performance.

Hypothesis 5. Goals is the only pre-
dictor of anticipated satisfaction with
adherence.

Hypothesis 6. Self-efficacy for perfor-
mance is directly influenced by self—
concept, perceived fitness, goals, anti-
cipated satisfaction with performance,
and anticipated satisfaction with adher-
ence.

Hypothesis 7. Self-efficacy for adher-
ence is directly influenced by self-
concept, perceived fitness goals, anti-
cipated satisfaction with performance,
and anticipated satisfaction with
adherence.

Hypothesis 8. Self-motivation is directly
predicted by self-efficacy for performance,

selfeefficacy for adherence, anticipated
satisfaction with performance, and uti-
cipated satisfaction with adherence

Hypot hes is 9. Performance is directly
influenced by self-motivation, self-
efficacy for adherence, anticipated sat—
isfaction with performance, anticipated
satisfaction with adherence, perceived
fitness, and physical fitness.

Hypothesis 10. Adherence is directly
predicted by self-motivation,

efficacy for performance, self-efficacy
for adherence, anticipated satisfaction
with performance, anticipated satisfac-
tion with adherence, perceived fitness,
and physical fitness.

 

S - Supported
l = Refuted
PS - Partially Supported

Figure 14. Status of the Path Hypotheses.

 

8 O

.—ov0l anuumozu0n>£ cu accouuacm«mgsuwotu
.aovos ”Hue cu ueuuacucwam.llllluusog

coast-

a
con

 

be...

    

 

   

cocoaozv< ecu .
>uno«uuolw~ow

aucouuzo<
sea:

cauuuounuuem

wouuaauaa=<

 

 

 

 

    

   
 

oucuEaOunoa sou
succuuuuuuﬁow

oocesuousum

 

 

 

soda.
ecuuuuumuucm
wouueauau=<

    

 

 

a saw: .ozuna “cacaca=wam .ma ot=maa

secessionl‘caOtae-u
tt§§a§t
at...
no.

anocu«m Hoounhzm

       

    

so...

   

sons-o»...

 

ou>uoouum. .....

cameo

unaccouuuﬁww

    

  

uaoaasm
_s«uom

  

dupes vun~mogu0azr cu ucwu«_«=Mum .4.
.~uvol Aﬁcu cu ucuouuucmum-Illlllnaox

 

as
sent. .0...-
o -

 

 
 

 

     

 

   

I
h
.
Oucuhoimi . q u c s u a n a a o oucwuvzmz
........ .2«:

IL on ouco~uso< new ..0 coauuuunuuwm
8 . o.o mucouuuuuuﬁum 8:393:
O

couum>uu02|u~ow
ouanuOunum
rad:
.. oocwEHOuuwm ecu comuoeuodunm
wocmsaoucwn .. honouuuunwmuw covenaouuc<
-
O

 

 

 

.
.o.

 

.
. I...
. -.
I'd... IDODI‘
..c-o..a.a-.a...

 

.o— so»: .naaua saunaaacuam .ea arsed.

      

  

omocuun
aau.n»;a

meocudm
. oo>«oouom

 

oﬁwoo
..
.51
r||
.... arugula
‘ ..... uawucooluawm
..
Inovblb

uuonaam
~uuuom

 

82

In Week 10 (see Figure 16), performance was influenced
by physical fitness and goals, but not by self-motivation,
self—efficacy for performance, self-efficacy for adherence,
anticipated satisfaction with performance, anticipated
satisfaction with adherence, perceived fitness, physical
fitness, self-concept, social support, or empathy.
Adherence was significantly predicted by goals and self-
efficacy for adherence. Self-motivation was only an effect
and not a causal influence of anything. Self-motivation was
significantly predicted by self-efficacy for adherence and
anticipated satisfaction with performance. Both anticipated
satisfaction with performance and anticipated satisfaction
with adherence were significantly predicted by perceived
fitness. Self-efficacy for adherence was significantly
predicted by goals, anticipated satisfaction with adherence
and social support. Self-efficacy for performance, goals,
and self—concept had no significant predictors among those
tested in this study. Perceived fitness was significantly

predicted by physical fitness.

Limitations of Path Analysis

Although causal modeling and path analysis are becoming
increasingly popular and important to the advancement of
social science research (Kenny, 1979), several limitations
of this nonexperimental research method must be considered.
Discretion must be exercised in making inferences.
Inferences made in the present study are justified for two

reasons. First, the theoretical model and its proposed

83
linkages were formulated a priori from established theory
and the careful observation of previous research. The
strength of interpreting results from path analysis is
derived from explicit a priori model construction (Cook &
Campbell, 1979; Duncan, 1975; Hunter & Gerbing, 1982; Hunter
& Levine, 1983; Kenny, 1979). In order to make strong
conclusions in causal modeling, the researcher must begin
with strong assumptions (Kenny).

Secondly, an implicit weakness in path analysis is
causes due to variables left out of the model. The model of
the present study considered several variables. Cook and
Campbell (1979) stated that if other probable causal
variables can be tested and excluded, the basis for causal
inference is strengthened.

Cook and Campbell (1979) suggested that the researcher
is in a stronger position to establish complex and realistic
path models when there are 3 or more points of measurement.
Unfortunately, the planned remeasurement points were not
carried out due to the follow-up failure.

As mentioned above, path analysis is not a method for
discovering causal directions, but for testing directions of
causation already specified by a model. Therefore, the
researcher must make the theoretical framework with which
the model was built explicit (Hunter & Levine, 1982;
Pedhazur, 1982). The present study reviewed the theoretical

and research background of each of the variables utilized in

84
the model and presented a rationale for the model based on
that theoretical framework.

The data may fit competing models equally well.
Therefore path analysis is more effective as a method for
rejecting untenable models than for supporting one of
several competing models (Kerlinger & Pedhzur, 1973).

Finally, the present study includes all the limitations
which are characteristic of passive observational studies
such as sampling problems (Hunter & Levine, 1982) and

extraneous variable control problems.

A Proposed Respecified Model

In the early stages of model or theory development it
is sometimes useful to delete the nonsignificant path
coefficients and examine the resulting respecified model
against the full model with another sample population. This
approach is referred to as theory trimming (Kerlinger &
Pedhazur, 1973). Kerlinger and Pedhazur also suggested the
meaningfulness of paths should be considered. Therefore the
respecified models contain only the significant paths that
eventually influence performance or adherence (see Figures
16 and 17).

Week 1 In Week 1, performance is directly predicted by
self-concept, perceived fitness, and anticipated
satisfaction with adherence (see Figure 17). Social support
and physical fitness indirectly influence performance

through their significant influence on perceived fitness.

85

.H wiS .Huvoz vo«uuuwnmo¢ oncogene

.2 33:...

 

 

oucouwzv<
sou:

cauuomumauwm

vouunﬁouuc<

 

 

O8 ﬁEhOwhwh

 

wwwcuah
amounzsm

 

onocuam
vo>uouu0m

 

 

uuonasw
«maqu

 

unoccooluﬁwm

 

 

 

86

Week 10 In Week 10, performance is directly predicted
by goals and physical fitness (see Figure 18). Adherence is
directly predicted by goals and self-efficacy for adherence.
Social support and goals also influence adherence through
their influence on self-efficacy for adherence. Anticipated
satisfaction with adherence directly affects self—efficacy
for adherence. Perceived fitness affects adherence through
anticipated satisfaction with adherence and physical fitness
has its influence through perceived fitness. A future step
in the research would be the examination of these proposed

respecified models with a new sample population.

Other Hypotheses and Analyses

 

The other hypotheses and analyses beyond the path
analyses were:

1. Subjects who initially set goals and were high in
self-efficacy and perceived fitness and low in anticipated
self-satisfaction with performance in the pretest would
experience greater increases in their distance on the 12
minute run than subjects who initially did not set goals and
were low in self—efficacy and perceived fitness and high in
anticipated satisfaction.

2. Subjects with greater distance increases would
experience increases in self—efficacy for performance and
perceived fitness over the 10 weeks and still experience low
anticipated satisfaction with performance. Subjects with

smaller distance increases would experience little or no

 

8'7

.oH sow: .Huvoz vauuuuwamwm wouonoah .nH shaman

          

uucmuwsv<

 

      

oucwnwzo< :uua
coauumwmuuuw mmocuum moccuum
vouonwuquc< vo>wouuom Huuuozzm

     

 

oucmuuco< new
suuuacauuuaom

 

 

 

eucmEHOWuwm

uuoansw
ﬁuHuom

88
change in self-efficacy, perceived fitness, or anticipated
satisfaction with performance.

3. Physical fitness measures would not change
significantly from Week 1 to Week 10 (due to the short
period of time since exercise was begun).

4. A MANOVA was planned to test the data for
significant change in the 9 variables suggested by the
hypotheses above.

To examine the first hypothesis an attempt was made to
form two groups at each data collection point based on
whether or not subjects had set goals and by median split of
the data for perceived fitness, self-efficacy for
performance, and anticipated satisfaction with performance.
A t—test to examine the difference between the two groups in
improvement in performance (increased distance on the 12
minute run) was planned. The groups fitting the required
parameters were extremely small (N = 6, N = 2). There was a
nonsignificant negative association of the groups with
distance change, 3(6) = —.65. (The criterion level for this
and all other analyses was 2 5 .05.) In a further attempt to
gain information on the relationship of these variables to
performance, four unplanned t—tests were calculated. First,
two groups were formed based on whether the subject had set
goals in Week 1. A t—test to examine the difference between
the two groups in improvement in performance (increased
distance on the 12 minute run) was calculated. There was a

nonsignificant negative association 3(45) = -.14, of the two

89
groups with distance change. Next, two groups were formed
from the top 25% and the low 25% of the scores of the Week 1
data for perceived fitness. A t—test to examine the
difference betweeen the high and lowest perceived fitness
groups in improvement in performance was calculated. There
was a nonsignificant negative association, 3(30) = —.17.
Two groups were then formed from the top 25% and the lowest
25% of the Week 1 data for self—efficacy for performance. A
t—test to examine the difference between the high and low
self-efficacy for performance groups in improvement in

performance was calculated. There was a nonsignificant

negative association, 3(33) = -.55. Finally, two groups
were formed from the top 25% and the lowest 25% of the Week
1 data for anticipated satisfaction with performance. A 3-
test to examine the difference between the high and low
anticipated satisfaction with performance groups in
improvement in performance was calculated. There was a
nonsignificant association, 3(39) = .46.

These t—tests indicated that there was not a
significant difference in improvement between groups high or
low in perceived fitness, groups high or low in self-
efficacy, groups high or low in anticipated satisfaction
With performance, or groups which did or did not set goals.
The hypothesis remained unsupported in this second set of
tests.

Four t—tests were conducted to test the second

hypothesis. A median split of the distance scores formed

90
groups high and low in performance change. A t—test
examined the difference in the two groups in changes in
self-efficacy for performance, perceived fitness, and
anticipated satisfaction with performance. Results
indicated that no significant differences existed between
the high and performance groups on changes in self-efficacy
for performance, 3(52) = 1.25, perceived fitness, 3(56) = -
.44, or anticipated satisfaction with performance, 3(63) = -
.96.

A paired t—test was calculated to examine the
difference in physical fitness in Week 1 and Week 10.
Results indicated physical fitness did not change
significantly from Week 1 to Week 10, 3(56) = -1.68.

A MANOVA was conducted to test the variables involved
in these 3 hypotheses for significance of change for Week 1
to Week 10. The results indicated significant changes
occurred. The multivariate {(9,37) = 7.14. Significant
changes occurred in the following variables in the
univariate F—tests: self—efficacy for performance,

performance, perceived fitness, and empathy (see Table 7).

91

Table 7. Univariate F—tests of MANOVA Test for Signi—
ficance of Change.

 

 

Variable Week 1 Mean Week 10 Mean F
ASPl 76.52 75.87 .06
ASA2 79.13 73.91 3.25
SEP3 43.44 26.96 34.73*
SEA4 66.34 62.83 1.8
Performance 12.42 12.96 28.37*
Perceived Fitness 3.21 3.49 4.55*
Social Support 2.58 2.61 .18
Empathy 3.43 3.94 6.18*
Self-Motivation 24.35 24.02 .18

 

Anticipated Satisfaction

DMNH

With Performance.

Self—efficacy for Performance.

Self—efficacy for Adherence.

Anticipated Satisfaction With Adherence.

Chapter VI

Discussion

This study examined the predictive power of a model of
the posited relationships of several biopsychosocial
variables to exercise performance and adherence. Utilizing
causal modeling and path analysis, its purpose was to
contribute to theory building in the exercise adherence and
health compliance area. The model was recursive, testing
only one—way causal direction hypotheses. This does not
deny the two-way flow of causal direction that could be
tested with a more complex, nonrecursive model. It is
simply a reflection of the choice to study a limited portion
of a growing set of assumptions and theories. The variables
are also not exhaustive, but reflect the most viable
research and theory in the area. The number of variables is
limited to retain the power of the path analyses. Path
analysis is reduced in power by the use of a large number of
variables and a small sample.

A major caution regarding the interpretation of the
results of this study is offered. The final sample of data
that could be accepted for path analysis is relatively small
for this type of analysis and the number of variables
utilized (N = 54 in Week 1 and N = 44 in Week 10). The

reader is advised to interpret the results with caution.

92

93

Major Findings

The overall results provide only weak support for the
hypothesized model as formulated in the path diagram. The
model was constructed from several theories and studies
regarding performance and adherence. Some of the theories
are consistently supported by the results, and others are
partially supported.

Sonstroem's (1974, 1978, 1984) exercise adherence
theory suggested that perceived athletic ability gained by
athletic or exercise participation enhances self-esteem or
self-concept which makes further physical activity
attractive and insures continued involvement. People who
feel good about themselves will be attracted to and continue
exercise. The significant paths which lead to performance
and adherence (see Figures 17 and 18) support Sonstroem's
proposal. Perceived fitness, a specific aspect of self-
concept is predicted by physical fitness and directly
affects performance in Week 1. (That the subjects were
primarily former high school athletes probably had an
important effect on perceived fitness in Week 1.) By Week
10, additional information on physical fitness gained
through performance and physical fitness tests influenced
perceived fitness, adherence, and performance. The effect
on adherence was mediated by both attitude toward continued
activity (anticipated satisfaction with adherence and self-
efficacy for adherence). This path of influence closely

follows Sonstroem's theory.

94

Bandura's (1977) theory of self-efficacy was partially
supported by the results. Bandura's theory suggests that
self-efficacy mediates the effect of perceived fitness,
goals, and social support on performance and adherence
(persistence of effort in his words). Self-efficacy for
adherence (but not self-efficacy for performance) was
predicted by goals and social support and influenced
indirectly by perceived fitness. Also consistent with
Bandura's findings, self-efficacy for adherence (but not for
performance) directly predicted adherence.

Several aspects of Bandura's theory were not supported.
Cognitions about physical states are an important source of
self-efficacy information in Bandura's theory. This element
of Bandura's theory was only partially supported by the
present study. Perceived fitness did indirectly influence
self-efficacy for adherence. Perceived fitness negatively
predicted self-efficacy for performance. This may reflect
several factors. The students were not all practicing
running. Some of the effect may have been from students
whose conditioning did not involve running. Perhaps they
viewed their fitness as enabling them to perform another
task. It may reflect the real difficulty im improving
performance a third time in a ten week period. It is highly
likely that it reflects the belief they would not do well
because they were forced to run on a differenct track when
University officials set up a commercial event on the track

the students had planned to run on. The new track was

95

bigger, and was outdoors, as opposed to the shaded indoor
track. It was a hot windy day during exam week. (The laps
were converted mathematically for the data analyses.) Even
runners may perceive their conditioning and fitness as
preparation unrelated to an all-out run lasting 12 minutes.
To test these hypotheses, future researchers may wish to
replicate the study and solidly assure same—track testing of
performance. They should add an outcome expectancy measure
to the replication which asks the subjects how certain they
are that thier physical fitness will be important to their
performance on the 12-minute run.

Secondly, self-efficacy for performance did not predict
performance as one would expect following Bandura's theory.
Goals and self—concept were not mediated by self-efficacy as
one would also expect, but had direct effects on
performance.

Finally, Bandura also suggests self-motivation mediates
the effect of self-efficacy on performance and adherence.
The results of the present study do not support this aspect
of Bandura's theory. While self—efficacy for performance
did predict self-motivation, self-motivation had no effect
on performance or adherence.

Dishman's (1981, 1982a, 1982c) adherence studies
reported that self—motivation and physical states were the
only significant predictors of exercise adherence. The
results of this study only partially support Dishman's

findings. Self—motivation never achieved significance as a

 

96

predictor of adherence or performance. Physical fitness did
directly predict performance in Week 1 and Week 10 but its
effect on adherence was mediated by several other path
variables. The complexity of the model tested may have
contributed to the difference in results of the present
study. The biopsychosocial model tested additional relevant
variables that were not included in Dishman's study.

Another contributing factor may be that the Self-Motivation
Inventory measures a general disposition to persevere, which
is not specific to exercise adherence as were the other
measures in the present study.

Garland (1985) postulated that self-efficacy (his
performance valence) and anticipated satisfaction (his
performance expectancy) mediated between goals and
performance. The results of the present study partially
support Garland's theory. Garland's construct of
performance expectancy or anticipated satisfaction with
performance was the source of this study's construct,
anticipated satisfaction with adherence. Anticipated
satisfaction with adherence did have significant paths in
both Week 1 and Week 10. It directly predicted performance
in Week 1 and mediated between perceived fitness and self-
efficacy for adherence which significantly predicted
adherence. However, the remainder of Garland's proposition
was not borne out in the results. Goals did not influence
anticipated satisfaction with performance, anticipated

satisfaction with adherence, or self—efficacy for

97

performance. The way the goals variable was operationalized
certainly contributed to the difference in results. Goals,
in the present study, was the number of times per week the
student planned to train. Garland's (1985) goal was defined
as the level of performance that an individual wishes to
achieve. The present study measure quantity of adherence,
while Garland was speaking of standards of performance
goals.

These were the major findings with regard to the
theories utilized in the present research. Another premise
of this study, that adherence and performance aspects of
measures should be separated wherever possible to clarify
constructs and effects, was well supported by the results.
The variables which predicted anticipated satisfaction with
performance versus anticipated satisfaction with adherence,
self-efficacy for performance versus self—efficacy for
adherence, and performance versus adherence differed. The
predictive power of anticipated satisfaction with
performance versus anticipated satisfaction with adherence
and self-efficacy for performance versus self-efficacy for
adherence also differed. It appears that the subjects had
distinctly different efficacy and expectancy beliefs with
regard to performance and adherence. Other measures which
may have given different results or more information had
they been separated into their performance and adherence
aspects are goals and self-motivation. Goals measured only

number of planned training sessions per week. A measure of

 

98

the standards for performance would give information needed
to measure the separate effect of adherence goals and
performance goals as predictors, as well as the effect of
other predictors on them. Future researchers may wish to

attempt that separation.

Other Analyses

The results of the nonpath analyses (E-tests) indicated
that the groups formulated according to Bandura's self-
efficacy theory and Garland's cognitive mediation theory
(goal setters high in self-efficacy for performance and
perceived fitness and low in anticipated satisfaction with
performance) did not have greater increases in performance
as expected. These results are consistent with the path
analysis.

Another t—test indicated no significant change in
physical fitness over the term. This test confirmed the
assumption of the study that physical states would not
change significantly in that time.

Finally, the MANOVA indicated significant increases in
performance, perceived fitness, and empathy and a
significant decrease in self-efficacy. According to
Bandura's theory, performance is one of the best sources of
efficacy information. By Week 10 the subjects had 2-3
performance scores that could contribute to self-efficacy.
Bandura's theory is not supported by this result. However,
as mentioned above the decrease in self-efficacy may have

been due in large part to the unplanned use of a different

 

99

track. It may be due to some students' beliefs that (a)
their conditioning was not related to an all—out run, and
(b) therefore they couldn't improve further. It may be due
to the belief that additional improvement in ten weeks was
not reasonable. The increase in perceived fitness which
probably utilizes the same information lends support to
Sonstroem's adherence theory. Performance increases were
most likely due to the effect of training. Empathy
increases are most certainly due the instructor's
friendliness which was perceived by the students as empathy

as well as social support.

Uncontrolled Factors and Measurement Problems

 

Several uncontrolled factors that may have contributed
to the present results are now mentioned. The students
became friendly and supportive to each other in their
exercise effort inside and outside the class. The
instructor was very friendly and supportive of the students'
exercise during class. This additional support may help
account for the switch in the role of social support from
predicting perceived fitness in Week 1 to predicting self—
efficacy for adherence in Week 10. As mentioned above, it
may also have influenced the increase in perceived empathy.

The students' status as former athletes means the group
was not a representative sample of college students and
tends to confound the results. For example, the fact that
many of the students were former athletes may account for

the early effect of physical fitness and perceived fitness

100

on performance. These students did not need several weeks
of training to know they were in good condition.

This study did not examine male and female differences.
Whether the variables have greater saliency for men's or
women's adherence or performance is an interesting and
perhaps vital question. Godin and Shepherd (1985) found
some male-female differences in adherence related to
response to social support. Future researchers may wish to
include further study of male-female differences in
adherence and variables related to adherence.

As mentioned above, the present study had a very small
sample for path analysis. Path analysis should have a
greater sample to achieve reliable results.

Yet another problem was the failure of the collection
procedure on proposed follow-up studies which was discussed
in the Methods section. Because the study began in spring
term, the follow-up reports were due in August and October.
Most students had changed addresses at least twice during
this time period making adequate mail and telephone follow
up impossible. The much needed information on the longer
term effect of the variables on adherence this part of the
study was designed to gain was not collected.

Several measurement problems that created
interpretation problems and results that were inconsistent
with the literature require discussion and revision before
any replication takes place. The decision to measure self-

concept only at the beginning of the study became a

101

problem. It would have provided a better test of
Sonstroem's theory of self-concept and adherence had a
posttest also been done. In addition, a broader measure of
self-concept or self-esteem appears to have been needed.

The physical self variable, while broader than perceived
fitness, probably did not capture the self—valuing (beyond
the physical self—valuing) that Sonstroem posits is enhanced
with increases in perceived athletic ability. Yet it was
enough broader than the perceived fitness variable that it
could not be used as that variable. Its intermediate status
may have been the reason self-concept paths did not achieve
significance where it was expected. The low reliability of
the measure in this study would caution against its use
without revision in any replication study.

As discussed above, several questions surround the
significant drop in the self-efficacy for performance scores
and the unexpected negative influence of physical fitness on
self-efficacy for performance. The self—efficacy for
performance scale should ask the subject to estimate their
actual performance rather than improvement in performance.
An outcome expectancy question that asks the subject to
state how certain they are that their physical fitness will
affect their performance is also recommended. Together,
these changes may resolve the questions regarding the
unexpected results.

Definition, operationalization, and separation of the

goals measure into its adherence and performance aspects

102

could similarly resolve the question of unexpected results
and interpretation problems encountered with the results
related to goals. This change could also give more
information about the differential effects of performance
goals and adherence goals.

Similarly, the Self-motivation Inventory appears too
general a measure for the present study. Motivation
measures specific to adherence and performance could provide
more specific information. Perhaps self-motivation would
also have greater predictive power with the use of these
separate measures.

The change of the performance measurement device, the
running track, also affected the reliability of the data.
Running on the bigger, warmer, windier track certainly
affected the subjects' performance. Anticipating the
different run problably also affected their self-efficacy
for performance scores. Future researchers are cautioned to
secure agreements from buildings officials that would
prevent such problems.

Finally, the large number of theories and variables
included in the design presented a problem in analyses.
Besides the reduction of explanatory power with the small
sample and large number of variables, the results were often
difficult to conceptualize at once. Comparison of models
from two concise theories rather than from a collection of
theories may provide a more manageable research design in

the future.

103

Conclusions

Although the hypothesized model did not fare well, the
results were informative:

1. Sontroem's exercise adherence theory was
consistently supported. Fitness did affect perceived
fitness which did in turn predict attitudes toward exercise
performance and adherence, and did predict adherence and
performance.

2. Bandura's self-efficacy theory was partially
supported. Self-efficacy (for adherence) mediated the
effect of goals, perceived fitness, and social support on
adherence. Self-efficacy did not mediate the effects of
perceived fitness, goals, and self-concept on performance.
Self-efficacy never had a positive influence on performance.

3. Garland's cognitive mediation of goals and tasks
theory was partially supported. Anticipated satisfaction
with adherence, a construct built on Garland's thinking,
mediated between goals and performance. However,
anticipated satisfaction with performance, self-efficacy for
performance, and self-efficacy for adherence did not
influence performance as one would expect from Garland's
theory.

4. Dishman's exercise adherence theory was partially
supported. Physical states did have an important influence
on performance and adherence as well as other variables
affecting adherence. Self-motivation, however, never

Significantly influenced either performance or adherence.

104

5. The present study was not designed to compare
theories but to examine the relationships of variables to
each other and to adherence. However, the results provide
the beginnings of such comparisons for a few theories and
emphasize the need for such comparisons.

6. The present study's suggestion that adherence and
performance aspects of all measures require separation was
consistently supported. Wherever such separation took place
it allowed the observation of separate predictors for those
variables as well as the observation of their differing
predictive powers.

7. Studies that require follow up and utilize college
students should be started in the beginning of the academic

year to assure data collection.

Implications for Exercise Programs

 

While the challenge to learn more about exercise
adherence continues, findings of the present study along
with the related literature, suggest the following
recommendations for exercise programs or individual exercise
prescriptions. Exercise leaders (or prescribers of
individual exercise prescriptions) should:

1. Encourage the exerciser to enlist the support of
the spouse or close friend. Suggest they discuss goals,
progress, and feelings about the exercise activity.

2. Assist the exercise participant in setting
adherence and performance goals. Emphasize reasonable goals

so that improvement and favorable assessment of fitness and

 

105

performance are perceived. Recognize and reinforce progress
the participant makes in reaching goals.

3. Schedule periodic fitness and performance tests to
provide feedback and reinforcement of actual and perceived
fitness levels.

4. Emphasize improvements the exerciser makes in
fitness, performance, self-esteem, and general mental
health.

5. Provide and encourage Close associates to support
and reinforce the exerciser's ability to adhere as well as

their actual adherence.

Implications for Future Research

 

1. Path analysis apppears a valuable model for the
multivariate study of exercise adherence.

2. Replication of this study and examination of the
proposed respecified models with a new sample population
appears to be a next step in future research.

3. The direct comparison of the explanatory power of
theories such as those of Sonstroem and Bandura appears to
be another next step. Such theory comparison would
contribute to identifying of viable theory to guide exercise
adherence efforts as recommended by Dishman et al. (1985).

4. Self-concept ought to be measured pre and post
study to provide a more complete test of Sonstroem's theory.
A broader measure of self—concept/self-esteem appears

required.

106

5. The self—efficacy for performance scale should ask
the subjects to estimate their actual performance rather
than their performance improvement.

6. An outcome expectancy scale should ask the subjects
to state how certain they are that their physical fitness
will be important to their performance.

7. The use of standardized definitions of variables
among exercise researchers is needed.

8. Follow-up and longitudinal studies are still needed
to provide information about adherence over time. I echo
Dishman (19820) and Sonstroem (1984) in saying that a ten

week study is just a beginning in any study of adherence.

107

Footnotes

In the present study exercise adherence refers to
continued involvement or persistence in exercise
behavior. Exercise behavior refers to involvement in
exercise activity such as jogging, aerobic dancing,
weight training, calisthenics, or swimming.

In this study Eerformance refers to the quality of
an exercise be av1or. How well one performed on a

timed run was the measure of one's performance.

In the present study compliance refers to persistence
in following a medical treatment plan.

108

APPENDICES

109

APPENDIX A

Questionnaires

110

Name:

 

Date:

 

Exercise Data Collection

Instructor's Measures On Exerciser

Part I — Physiological States

 

1. Resting Heart Rate pre post

2. Harvard Step Test pre post

 

Part II - Performance Measure

 

12 Minute Run pre Week 5

post

111

Name:

 

Date:

 

Exerciser's Se1f—report Questionnaire

 

Part III - Perceived Fitness

 

1. I be1ieve I am in good physica1 condition.

1 2 3 4 5

somewhat moderate1y strong1y be1ieve
2. I be1ieve I am a physica11y fit person.

1 2 3 4 5

 

somewhat moderate1y strong1y be1ieve

Part IV - Socia1 Support

 

P1ease write the name of a few of your c1ose friends or fami1y who
give you support or encouragement for your fee1ings, thoughts, or
activities.

 

1. Do peop1e c1ose to you 1et you know they be1ieve your thoughts
about exercising are okay.

0 1 2 3 4 5
never very se1dom moderate1y often very often
se1dom

2. Do peop1e c1ose to you he1p you think about your exercise p1ans
or strategies?

0 1 2 3 4 5
never very se1dom moderate1y often very often
se1dom

3. Do peop1e c1ose to you do things 1ike drive you to exercise
c1ass or buy exercise shorts or shoes to encourage your exercise.

0 1 2 3 4 5
never very se1dom moderate1y often very often
se1dom

 

 

112

4. Do peop1e c1ose to you do things 1ike exercise with you or ca11
you so you can remember to exercise?

0 1 2 3 4 5
never very se1dom moderate1y often very often
se1dom

5. Do peop1e c1ose to you share your joy when you've met a new
exercise goa1?

0 1 2 3 4 5
never very se1dom moderate1y often very often
se1dom

6. Do peop1e c1ose to you he1p you with your anger, gui1t, or sad-
ness when you aren 't meeting your exercise goa1s?

0 1 2 3 4 5
never very se1dom moderate1y often very often
se1dom

Part V - Empathy

1. Does the instructor understand your attitudes and fee1ings about

exercise?
0 1 2 3 4 5
never very se1dom moderate1y often very often
se1dom

2. Does the instructor appear to care if you exercise?

0 1 2 3 4 5
never very se1d0m moderate1y often very often
se1dom

3. Does the instructor encourage you to exercise?
0 1 2 3 4 5

never very se1dom moderate1y often very often
se1dom

113

Part VI — Tennessee Se1f-Concept Sca1e

 

Items 1—18 make up the Physica1 Se1f subsca1e. The TSCS is copy—
righted. Copies are avai1ab1e from Western Psycho1ogica1 Services,
12031 Ni1shire B1vd., Los Ange1es, CA 90025.

Part VII - Goa1s Inventory

 

1.

Do there seem to be specific exercise goa1s for this c1ass?
(Some target or 1eve1 of exercise to work for)

Yes No

 

 

If there are c1ass goa1s are they set for the term?

Yes No

 

 

If there are c1ass goa1s are they set for each week?

Yes No

 

If there are c1ass goa1s, are they about "sticking to it” or
persistence?

Yes No

 

 

If there are c1ass goa1s, are they about gua1ity of your exercise
performance?

Yes No

 

 

Do you set specific exercise goa1s yourse1f?

Yes No

 

 

If you set your own goa1s, are they set for the term?

Yes No

 

 

If you set your own goa1s, are they set each week?

Yes No

 

 

If you set your own goa1s, are they about "sticking to it" or
persistence?

Yes No

 

 

If you set your own goa1s, are they about the gua1ity of your
exercise performance?

Yes No

114

Part VIII - Se1f-Efficacy for Performance

 

Instructions:

1/2 1ap

1 1ap
1—1/2 1aps
2 1aps
2-1/2 1aps
3 1aps
3-1/2 1aps
4 1aps
4-1/2 1aps
5 1aps
5-1/2 1aps
6 1aps
6-1/2 1aps
7 1aps
7—1/2 1aps
8 1aps

Se1f-efficacy is the be1ief that you can perform a
particu1ar task. It is se1f-confidence in a parti-
cu1ar or specific situation. Rate how strong1y you
be1ieve (have se1f-efficacy or se1f—confidence) that
you can improve your performance on the 12 Minute
Run by each of the amounts 1isted:

% 0F CERTAINTY

0 10 20 30 40 50 60 70 80 90 100

 

115

Part IX — Se1f-Efficacy for Adherence

 

Instructions: Rate how strong1y you be1ieve that you wi11 continue
to do your exercise program during the next 6 months.
(The specific task here is persisting at an exercise
program.) Rate how strong1y you be1ieve (have se1f-
efficacy or se1f-confidence) that you can continue
to exercise the number of times per week 1isted:

% 0F CERTAINTY

0 1O 20 30 40 50 60 70

every day

0‘

times a week

0'!

times a week

4:.

times a week

0.)

times a week

N

times a week

 

—l

time a week

Part X - Anticipated Satisfaction With Performance

 

Instructions: How satisfied or p1eased or accepting of your exercise
performance in this c1ass do you expect to be? Rate
how satisfied you think you wi11 be:

extreme1y extreme1y
dissatisfied satisfied

_0% __10% _20% __30% _40% _50% __60% _70% _80% _90% _100%

 

116

Part XI - Anticipated Satisfaction With Adherence

Instructions: How satisfied or p1eased or accepting of your exercise
adherence in this c1ass do you expect to be? Rate
how satisfied you think you wi11 be:

extreme1y extreme1y
dissatisfied satisfied

_0% _1o% _20% _30% _40% _50% _60% _70% _80% __90% _100%

Part XII - Dishman's Se1f-motivation Sca1e

Instructions: Read each of the statements be1ow, and circ1e the
number beneath the 1etter corresponding to the a1ter—
native that best describes how characteristic the
statement is when app1ied to you. The a1ternatives

are:
A. extreme1y.uﬂcharacteristic of me.
B. somewhat yﬂcharacteristic of me.
C. neither characteristic nor uncharacteristic of me.
D. somewhat characteristic of me.
E. extreme1y characteristic of me.

A B C D E

5 4 3 2 1 1. I get discouraged easi1y.

5 4 3 2 1 2. I don't work any harder than I have to.

1 2 3 '4 5 3. I se1dom if ever 1et myse1f down.

5 4 3 2 1 4 I'm just not the goa1-setting type.

1 2 3 4 5 5. I'm good at keeping promises, especia11y
the ones I make to myse1f.

5 4 3 2 1 6. I don't impose much structure on my activities.
1 2 3 4 5 7. I have a very hard-driving, aggressive person—
a1ity.

 

*Copyright 1978 by R. K. Dishman and W. J. Ickes. Permission of the
copyright hoIders granted 4/16/85.

 

APPENDIX B

Informed Consent and Information Release

 

118

MICHIGAN STATE UNIVERSITY
Department of Health and Physical Education
Department of Psychology

RESEARCH CONSENT FORM

1. I have freely consented to take part in a scientific study being
conducted by Mary Clearing Sky a graduate student in Michigan
State University's Clinical Psychology Department, as part of
her doctoral dissertation research and as an authorized part of
the research and evaluation program in the Department of Health
and Physical Education under the supervision of Robert Caldwell,
Ph.D., Assistant Professor of Psychology and Deborah Feltz, Ph D.,
Associate Professor of Physical Education.

2. The study has been explained to me and I understand the explana-
tion that has been given and what my participation will involve.
I understand that information about my attitudes and physical
preparation for exercise and my exercise performance will be
collected. I understand that the questionnaire contains some
personal questions. The information will be used to learn more
about people who exercise and will be helpful to educators and
health-care providers.

 

3. I understand that I will complete a 30 minute questionnaire in
the first and tenth weeks of class, brief questionnaires in the
fifth week of class and at 10 and 20 weeks after class. I under-
stand that as part of the follow-up I will also complete a summary
log of my weekly exercise.

4. I understand that my name will be removed from the questionnaire
as soon as the forms are stapled together.

5. I understand that I am free to discontinue my participation in
the study at any time without penalty.

6. I understand that the results of the study will be treated in
strict confidence and that I will remain anonymous. Within these
restrictions, results of the study will be made available to me
at my request.

7. I understand that my participation in the study does not
guarantee any beneficial results to me.

8. I understand that, at my request, I can receive additional
explanation of the study after my participation is completed.

Signed:

 

Title of Exper: An Analysis of the Predictors of Exercise Perfor—
mance and Adherence Date:

 

 

 

119

MICHIGAN STATE UNIVERSITY
Department of Psychology
Department of Health and Physical Education

Exercise Adherence Study

Dear
verifying person

 

The student who signed below is a volunteer in a study of exercise
behaviors. He/she has given you permission to report his/her
exercise behavior. Please complete the attached form and return

it to us in the enclosed self-addressed, stamped envelope as quickly
as possible. Thank you for your assistance in our study.

Students name:

 

I, , grant permission to
student's signature

 

, to report my exercise

 

person who will verify your exercise
behavior to Mary Clearing Sky for use in an exercise adherence study.
This information is to be strictly confidential and for use only

in this study.

Date:

 

Student's summer address:

 

City, State, Zip:

 

Phone Number: ( )

 

Name of person verifying your exercise:

 

Address:

 

City, State, Zip:

 

Phone Number: I, )

 

 

120

APPENDIX C

Course Outline

 

12].

Eggrsg: General Conditioning

Instructor: Jeanne Foley

ijige: IM Circle, Ex. Phys. Lab Office E
ghgge: 5-4734

Suggested References:

1. Physiology of Fitness — Brian J. Sharkey
(Human Kinetics Publishers, Inc.) $12.95

2. Fit or FAT - Covert Bailey
(Houghton-Mifflin Co.) 3 4.95
. _Course Outline:
- 10 Heeks, 2 classes per week
- Heek One: Monday - Intro, paperwork
Hednesday - Fitness Test. intro to stretching/
warmup routine
- Week Two: Monday - Hark out individual conditioning programs,
then warmup and first workout
Hednesday - First 15:00 warmup (in gym) then follow
workout program
- Heek Three-Ten: Harmup/workout as on Bed. of Heek 2

Note: On Monday or last week of class. we will do the Fitness
Test again - mandatory

Gradiny
Your course grade will be detenmined in two parts:
1) Attendance (701 of grade)
0 or 1 obsence: 4.0, grade drops by 0.5 for every absence
over 1. (Note: more than 10 minutes late to class . absence)

2

v

Conditioning Log (30% of grade)

Daily log of workouts. record of fitness tests. etc.

(Log sheets and further details on this will be given in next
class period.)

Final Grade = .7 x (attendance grade) + .3 x (109 grade) rounded
to next lower half grade point

Example: 3 absences, A on notebook (3 absences . 4.0 - 1.0)
Grade: .7 (3.0) + .3 (4.0) - 2.1 + 1.2 - 3.3 Final Grade: 3.0

 

122

APPENDIX D

Testing Schedule

 

123

Testing Schedule

Week 1 and Week 10

Week

Harvard Step Test

Resting Heart Rate

Body Weight

Perceived Social Support

Perceived Empathy

Tennessee Se1f-Concept Sca1e

Perceived Fitness

Goals Inventory

Anticipated Satisfaction With Performance
Anticipated Satisfaction With Adherence
Se1f-efficacy for Performance
Se1f-efficacy for Adherence

Dishman's Se1f-Motivation Inventory
Twelve Minute Run

10 only

Adherence Computation

 

124

APPENDIX E

Pearson Correlations

 

125

 

 

 

 

cc; oceanouuum
5°. ooé coda-">30...
:u—om
:. 3. ooé <mm
No. can. 2. 004 sun
can. cum. «on. cm. 004 <w<
«cm. «an. oo. «2. «an. ac.” mw<
co. co. .3. we. 50. co. 8% 3:00
«3.. «Cu. «2. 8.: «ma. «nu. Nu. co.“ nave»:
vo>uoouom
2. 2. co. .5. 2.: no. “of no. 004 unoucoo::ww
~uu«n>;m
«mm. 2. can. no. «3. S. no. «em. tau. ooé nous:
“coughs;
no. I 3.: no. 3.: 2.: «3.: mo. no. cc. 8.: 094 3:95
co. : 2. «mm. 3.: «3. 2. 2. (on. «MN. com. «mm. co; 2233
Hnuuow

3:550:3— coUuZuOs <mw mum <m< mw< 3.30 :05: umuu=ou::om one—5: Nﬁummu uuommzw

::wm vuzouuom :utmcm magma-E :«oom
.H see: Edi—33> wcos< 25:33.25 .Tm o3:

 

126

 

 

 

co; coconmtv<
no. oo.~ ouc:Eu0uuom
cﬁ. can. oo.~ cauuu>«uos
:u—d»
can. «ma. cm~. oo.— <Mm
No. «~n.: mo. NH. cc.~ mum
No. can. «ma. can. no. oo.— <m<
oo.: «en. cow. «nu. ea. can. oo.d mm<
«no. we. oﬁ. «ca. oo. 90. No.: Do.“ mason
o—.: «mm. «Nm. «on. 00.: «co. «on. —O.: oo.~ anocuum
v~>uuouom
so.: c_~. n—. «o. No.: «aw. c—N. sc.: «KN. oo.# unwoccunu~om
Huuaasza
cn~. cum. ON. -. so.: Hm. 0‘. no. «on. ma. oo.~ unocuum
~uu«m»:&
dd. 00.: «mm. ann.: ~—.: 00. No. o~. mu. wo.: o—.: co." >£unaau
mo.: «0.: «ca. -.: «0. ON. «MN. co. can. ~H. «u. «we. oo.~ uuonnsw
~u—uow
uocouozv< oucuEuOHHum coauu>uuoE <um mum <w< mm< a—uou aau:u«m amoucooiuaum onucudm Nsunumm uuommaw
:uauw vu>«oonom gou«u»;m , unouo>sm auuuow
ed xvo: .oounounu> meoE< mucuuwmounou .~nm oupqh

 

 

127

APPENDIX F

Statistical Formulae

 

128

Formula for fetest

The computational formula for testing a full model against a

reduced modellis:

SS SS
If (k-g,df ) = -—- reduced - —— ) k-g
——ﬂ§e full e e full
MS
e full
Where SSe = sum of squares error and K = the number of independent
variables in the full regression equation and g = the number of

independent variables in the reduced regression equation.

1L. Ott (1977). An introduction to statistical methods and
data analysis. North Scituate, MA: Duxburg Press.

 

 

 

129

Formula for "Q" Coefficient-I

Full Model

2 _ _ _ 2 _ 2 _ 2
Rm — 1 (1 Rl)(l R2) ...(1 Rp)

Reduced Model

_ _ 2 _ 2 _ 2
M — l (1 R1) (1 R2) ...(1 RP)
1-R2
W = - (n — d) log Q = — (N - d) log ( m) i
e e >
1 - M
where as:
W = measure of goodness of fit
N = sample size
d = number of path coefficients hypothesized

to be equal to zero (number; overidenti-
fying restriction)
log = natural logarithm

W = approximate X2 distribution w/df = d

 

1E. S. Pedhazur (1982). Multiple regression in

behavioral research: Egplanation and prediction. New
York: Holt, Rinehart, and Winston.

 

130

APPENDIX G

E-tests

 

 

 

131

Table G—5. fetests Testing Full Models Against Hypothesized Models.

 

 

 

 

Variables Week 1 Week 10
Se1f-concept f_(0, 51) = O f (O, 41) = 0
Perceived Fitness f_(3, 50) = 3.05* f_(3, 40) = .56
Goals f_(2, 49) = .07 £_(2, 39) = 1.19
Anticipated Satisfaction
,with Performance f_(5, 48) = 2.92* f_(5, 38) = 3.66**
Anticipated Satisfaction

with Adherence f_(5, 48) = 2.34 f (5, 38) = 4.73**
Self-efficacy for

Performance f_(3, 46) = 1.14 f (3, 36) = .47
Se1f-efficacy for

Adherence f_(3, 46) = 3.49* f (3, 36) = 2.83
Se1f-motivation f_(6, 44) = .47 f_(6, 34) = .53
Performance f_(7, 43) = 2.51* f_(4, 33) = 2.54
Adherence £_(4, 33) = 5.96***

 

* p < .05
** p < .01
*** p < .001

 

132

REFERENCES

 

References

Albinson, J. G., & Moore, J. R. (1983). Se1f-efficacy and

 

an participation in fitness activities. Project No.

 

457- 0001-2. Fitness and Amateur Sport, Canada.
Arends, J. (1983). Fifty percent of men, 97 % of women never

physically fit in their lives. Michigan Medicine, 5,

 

296-298.
Bahrke, M. S. & Morgan, W. P. (1978). Anxiety reduction

following exercise and meditation. Cognitive Therapy

 

and Research, 3, 232-333.
Bandura, A. (1977). Self efficacy: Toward a unifying

theory of behavior change. Psychological Review, 191-

 

215.

Bandura, A. (1980). Se1f-referent thought: A developmental
analysis of self-efficacy. In J. H. Flavell & L. D.
Ross (Eds.), Cognition and social development:

Frontiers and possible futures (pp.200-239). New York:

 

Cambridge Press.
Bandura, A. (1982). Se1f-efficacy in human agency.

American Psychologist, 31, 122-147.

 

Bandura, A. (1984). Recycling misconceptions of perceived

self-efficacy. Cognitive Therapy and Research, 3,

 

231-255.

 

 

 

134

Bandura, A., & Adams, N. E. (1977). Analysis of self-

efficacy theory of behavioral change. Cognitive

 

Therapy and Research, A, 287-310.
Bandura, A., Adams, N. B., & Beyer, J. (1977). Cognitive

processes mediating behavioral change. Journal of

 

Personality and Social Psychology, 35, 125-139.
Bandura, A. Adams, N. B., Hardy, A. B., & Howells, G. N.
(1980). Tests of the generality of self-efficacy
theory. Cognitive Therapy and Research, 4, 39-66.
Bandura, A. & Schunk, D. (1981). Cultivating competence,
self-efficacy, and intrinsic interest through proximal
self-motivation. Journal of Personality and Social
Psychology, 41, 586-598.
Bernier, M. & Poser, E. G. (1984). The relationship
between self-efficacy, attributions, and weight loss

in a weight rehabilitation program. Rehabilitation

 

Psychology, 22, 95-105.

Blair, 8. N., Collingwood, T. R., Reynolds, R., Smith, M.
Hagen, D., & Sterling, C. L. (1984). Health promotion
for educators: Impact of health behaviors,
satisfaction, and general well-being. American
Journal of Public Health, 2, 147-149.

Blair, S. N., Goodyear, N. N., Gibbons, L. W., & Cooper, K.
H. (1984). Physical fitness and incidence of
hypertension in healthy normotensive men and women.
The Journal of the American Medical Association, 252,

487-490.

135

Blumenthal, J.A., Sanders, W., Needels, T.L., & Wallace, A.
G. (1982). Psychological changes accompany aerobic
exercise in healthy middle-aged adults. Psychosomatic
Medicine, 53, 529-535.

Brown, E. Y., Morrow, J. R., & Livingston, S. M. (1982).
Se1f—concept changes in women as a result of training.
Journal of Sport Psychology, 3, 354-363.

Brownell, K.D. (1984). Behavioral and psychological aspects

of motivation to exercise. International Journal of

 

Sports Medicine, 5, 69-70.
Carmack, M. A., & Martens, R. (1979). Measuring commitment
to running: A survey of runners attitudes and mental

states. Journal of Sport Psychology, 1, 25-42.

Carmody, T. P., Senner, J. W., Malinow, A. R., & Matarazzo,
J. D. (1980). Physical exercise rehabilitation:
Long term dropout rate in cardiac patients. Journal
of Behavioral Medicine, 3, 163—169.

Cook, T. D., & Campbell, D. T. (1979). Quasi-
experimentation: Design and analysis issues for field
settings. Chicago: Rand McNally.

Cooper, K. A. (1968). A means of assessing maximal oxygen

intake. Journal of the Amerian Medical Association,
_93, 201-204.
Danish, S. J. (1983). Musings about personal competence:

The contributions of sport, health, and fitness.

American Journal of Community Psychology, 11, 221-240.

 

136

DeVries, H. S. (1970). Physiological effects of an exercise
training regimen upon men aged 52-88. Journal of
Gerontology, 25, 325-336.

DeVries, H. S. (1981). Tranquilizer effect of exercise: A
critical review. Physician and Sports Medicine, 2,
47-55.

DiMatteo, M. R. & DiNicola, D. D. (1982). Achieving
patient compliance: The psychology of the medical
practitioner's role. New York: Pergamon.

DiMatteo, M. R., Friedman, H. S., & Taranta, A. (1979).
Sensitivity to bodily nonverbal communication as a
factor in practitioner-patient rapport. Journal of
Nonverbal Behavior, 5, 18—26.

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, II: Charles C. Thomas

Dishman, R. K. (1982a). Compliance/adherence in health
related behavior. Health Psychology, 1, 237-267.

Dishman, R. K. (1982b). Contemporary sport psychology. In
R. L. Terjung, (Ed.). Exercise and Sport Sciences
an Reviews (Volume 10, pp. 120-139). Lexington, MA:
D. C. Heath.

Dishman, R. K. (1982c). Hea1th psychology and exercise

adherence. Quest, 3;, 166-180.

 

 

137

Dishman, R. K. (1985). Medical psychology in exercise and

sport. Medical Clinics of North America, 3311), 123

 

-143.
Dishman, R. K., & Gettman, L. R. (1980). Psychologic

influences on exercise adherence. Journal of Sport

 

Psychology, 3, 295-310.

 

Dishman, R. K., Ickes, W. J., & Morgan, W. P. (1980).
Self-motivation and adherence to habitual physical
activity. Journal of Applied Social Psychology, 33,
115-131.

Dishman, R. K., Sallis, J. F., & Orenstein, D. R. (1985).
The determinants of physical activity and exercise.
Public Health Reports, 399, 158-171.

Duncan, 0. D. (1975). Introduction to structural equations

 

models. New York: Academic Press.
Eastman, C. & Marziller, J. S. (1984). Theoretical and
methodological difficulties in Bandura's seff-efficacy

theory. Cognitive Therapy and Research, 3, 213-229.

 

Eagan, M. (1984). Why everyone isn't exercising. Employee

 

Services Management, 31, 34-35.

 

Falls, H. B., Baylor, A. M., & Dishman, R. K. (1980).

Essentials of fitness. Philadelphia: Saunders

 

College.

138

Feltz, D. L. (1980). A path analysis of the causal

 

elements in Bandura's theory of self-efficacy and

an anxiety-based model of avoidance behavior.
Doctoral dissertation, Pennsylvnia State University,
University Park.

Feltz, D. L. (1982). Path analysis of the causal elements
in Bandura's theory of self-efficacy and an anxiety-
based model of avoidance behavior. Journal of
Personality and Social Psychology, 33, 764-781.

Feltz, D. L., Landers, D. M., & Raeder, U. (1979).
Enhancing self—efficacy in high avoidance motor tasks:
A comparison of modeling techniques. Journal of Sport
Psychology, 3, 112-122.

Feltz, D. L. & Mugno, D. A. (1983). A replication of the
causal elements in Bandura's theory of self-efficacy
and the influence of autonomic perception. Journal of

Sport Psychology, 3, 263-277.
Fitts, D. H. (1965). The Tennessee Se1f-Concept Sca1e

 

manual. Nashville: Counselor Recordings and Tests.
Friedman, H. S. & DiMatteo, M. R. (1979). Health care as

an interpersonal process. Journal of Social Issues,

 

_5’ 1-11.

Folkins, C. H. & Sime, W. E. (1981). Physical fitness

training and mental health. American Psychologist,

 

6! 373-389.

139

Garland, H. (1985). A cognitive mediation theory of task

goals and human performance. Motivation and Emotion,

 

9, 345-367.
Gayle, J. B., 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, 39, 544-549.

 

Godin, G. & Shephard, R. J. (1985). Psycho-social
predictors of exercise intentions among spouses.

Canadian Journal of Applied Sport Sciences, 39, 36-43.

 

Goldberg, L., Elliot, D., Schutz, R. W., & Kloster, R. E.
(1984). Changes in lipid and lipoprotein levels after

weight training. The Journal of the American Medical

 

Association, 333, 504-506.

Goodrick, G. K., Warren, D. R., Hartung, G. H., & Hoepful,
J.A. (1984). Helping adults stay physically fit:
Preventing relapse following aerobic exercise

training. Journal of Physical Education, Recreation,

 

and Dance, 3, 48-49.

Greist, J. H., Eischens, R. R., Klein, M. H. , & Faris, J.
W. (1979). Antidepressant running. Psychiatric
Annals, 3, 134-140.

Greist, J. H., Klein, M. H., Eischens, R. R., Faris, J.,
Gurman, A. S., & Morgan, W. P. (1978). Running

through your mind. Journal of Psychosomatic

 

Resesarch, 33, 259-294.

 

140

Gunn, D. G., & Zwingmann, C. H. (1984). The dangers of
ill-health through recreation. Journal of
Adolescence, 3, 73-81.

Hagberg, J. M. (1986, February). Health benefits of

 

exercise for older people. Paper presented at Center
for the Study of Human Performanace. Michigan State
University. East Lansing, Michigan.
Harris, M. B. (1981a). Runner's perception of the benefits
of running. Perceptual and Motor Skills, 33, 153-154.
Harris, M. B. (1981b). Women runners' view of running.

Perceptual and Motor Skills, 33, 395-402.

 

Haskell, W. L., & Blair, S. N. (1980). The physical
activity component of health promotion in occupational

settings. Public Health Reports, 93, 109-118.

 

Heaps, R. A. (1978) Relating physical and psychological
fitness: A psychological point of view. Journal of
Sports Medicine and Physical Fitness, 39, 399-408.

Heinzelmann, F., & Bagley, R. W. (1970). Response to
physical activity programs and their effects on health

behavior. Public Health Reports, 33, 905—911.

 

Hilyer, J. C., & Mitchell, W. (1979). Effect of systematic
physical fitness training combined with counseling on
self-concept of college students. Journal of

Counseling Psychology 33, 427-436.

141

Hirsch, B. (1981). Social networks and the coping process:
Creating personal communities. In B. H. Gottlieb

(Ed.). Social networks and social support (pp. 149-

170). Beverly Hills: Sage.
Hunter, J. E., & Gerbing, D. W. (1982). Unidimensional

measurement, second order factor analysis, and causal

models. In B. M. Staw and DL. L. Cummings (Eds.)

Research in Organizational Behavior. Volume IV

267-320). Greenwich, CT: JAI Press.

Hunter, J. E. & Levine, R. L. (1983). Regression

Correlation, meta-analysis, confidence

(pp-

methodolgy:
intervals, and reliability. Journal of Leisure
Research, 33, 323-343.

Hurtado, A., Greenlick, M., & Columbo, T. (1973).

Determinants of medical care utilization: Failure to

keep appointments. Medical Care, 33, 189-198.

Jasnoski, M. L., Holmes, D. 8., Solomon, 8., & Aguiar, C.

(1981). Exercise, changes in aerobic capacity, and

Changes in self-perceptions:
Journal of Research in Personality,

An experimental

investigation.

5, 460-466.

Jorgenson, C. B., & Jorgenson, D. E. (1979). Effect of

running on perception of self and others. Perceptual

and Motor Skills, 53, 242.

Kaplan, N. M. (1984). Joggers may live longer.
252, 528.

Journal of the American Medical Association,

142

Kaplan, R. M., & Atkins, C. J. (1984). Specific efficacy
expectations mediate exercise compliance in patients

with COPD. Health Psychology, 3, 223-242.

 

Kau, M. L., & Fischer, J. (1974). Se1f-modification of
exercise behavior. Journal of Behavioral Therapy and

Experimental Psychiatry, 3, 213-214.

 

Kazdin, A. E. (1980). Covert and overt rehearsal and
elaboration during treatment in the development of

assertive behavior. Behavior Research and Therapy,

 

_33, 191-201.

Keefe, F. J., & Blumenthal, J. A. (1980). The life fitness
program: A behavioral approach to making exercise a
habit. Journal of Behavioral Therapy and Experimental
Psychology, 33, 31-34.

Kenny, D. A. (1979). Correlation and causality. New York:

 

Wiley.

Kerlinger, F. N., & Pedhazur, F. (1973). Multiple
regression in behavioral research. New York: Holt,
Rinehart, & Winston.

Laffrey, S. C., & Isenberg, M. (1983). The relationship of
internal locus of control, value placed on health,
perceived importance of exercise, and participation in
physical activity during leisure. International
Journal of Nursing Studies, 39, 187-196.

Lee, C. (1982). Se1f-efficacy as a predictor of

performance in competitive gymnastics. Journal of

Sport Psychology, 9, 405-409.

143

Lundberg, G. D. (Ed.) (1984). Olympic Issue [Special
Issue]. Journal of American Medical Association, 333,
(4).

Marlatt, G. A., & Gordon, J. R. (1980). Determinants of
relapse: Implications for the maintenance of behavior
chanage. In P. 0. Davidson & S. M. Davidson (Eds.).
Behavioral medicine: Changing health lifestyple
(pp.410—452). New York: Brunner and Mazel.

Marziller, J. & Eastman, C. (1984). Continuing problems
with self-efficacy theory: A reply to Bandura.
Cognitive Therapy and Research, 3, 257-262.

Merkin, G, & Hoffman, M. (1978). The sportsmedicine

 

book. Boston: Little, Brown.
Morgan, W. P. (1979). Anxiety reduction following acute

physical activity. Psychiatric Annals, 9, 141-147.

 

Morgan, W. P. (1981). Psychological benefits of physical
activity. In F. J. Nagell & H. J. Montoye (Eds.),
Exercise in health and disease, (pp-299-314).
Springfield, IL: Charles C. Thomas.

Morgan, W. P. (1985). Affective beneficence of vigorous
physical activity. Medicine and Science in Sports and
Exercise, 33, 94-100.

Muir-Gray, J. A., Young, A., & Ennis, J. R. (1983).
Promotion of exercise at work. British Medical

Journal, 286, 1958-1959.

144

Mustafa, S. D., Garrett, H. L., & Freeman, R. A. (1984).
Physical exercise and health: A review study.

Mississippi State Medical Association, 3, 87-88.

 

Nelson, A., Gold, B., Hutchinson, R., & Benezra, E.
(1975) Drug default among schizophrenic patients

American Journal of Hospital Pharmacy, 33, 1237-1242.

 

 

Olszewska, G. (1982). The relation of a self-image, se1f-
estimation and a tendency to dominate or submit to the
effectiveness of the performance of team players.

International Journal of Sport Psychology, 33, 107-

 

113.

Ott, L. (1977). An introduction to statistical methods

 

and data analysis. North Scituate, MA: Duxburg

 

Press.

Paffenfarger, R. S., Hyde, R. T., Wing, A. L., &
Steinmetz, C. H. (1984). A natural history of
athleticism and cardiovascular health. The Journal of

the American Medical Association, 252, 491-495.

 

Paolone, A. M. (1983). Prescribing exercise programs. In
A. A. Bove & L. C. Lowenthal (Eds.), Exercise

Medicine: Physiological Principles and Clincial

 

Applications (pp. 361-381). New York: Academic
Press.

Parent, C. J., & Whall, A. L. (1984). Are physical
activity, self-esteem, and dependency related?

Journal of Gerontological Nursing, 39(9), 3'11-

145

Pedhazur, E. J. (1982). Multiple regression in behavioral

 

research: Explanation and prediction. New York:
Holt, Rinehart, and Winston.
Riddle, R. K. (1980). Attitudes, beliefs, behavioral
intentions, and behaviors of women and men toward

regular jogging. Research Quarterly for Exercise and

 

§pp£3, 33, 663-674.

Roberts, G. C., Kleiber, D. A., & Duda, J. L. (1981). An
analysis of motivation in children’s sport: The role
of perceived competence in participation. Journal of
Sport Psychology, 3, 206-216.

Ryan, A. J. (1980). The physician and sportsmedicine guide
to running. New York: McGraw-Hill.

Sachs, M. L. (1982). Compliance and addiction to exercise.

In R. C. Cantu (Ed.), The exercising adult (pp. 19-

 

27). Lexington, MA: Collamore Press.

Sachs, M. L. & Pargman, D. (1979). Running addiction: A
depth interview examination. Journal of Sport
Behavior, 3, 143-155.

Schunk, D. H. (1984). Enhancing self-efficacy and
achievement through rewards and goals: Motivational

and informational effects. The Journal of Educational

 

Research, 39, 29-34.
Serfass, R. C., & Gerberich, S. C. (1984). Exercise for
optimal health: Strategies and motivational

considerations. Preventive Medicine, 33, 79-99.

 

146

Sharkey, B. J. (1975). Physiology and physical activity.

 

New York: Harper and Row.
Sheehan, G. A. (198). The case for exercise. The American

Journal of Cardiology, 33, 260.

 

Shephard, R. J. (1984). Management of exercise in the
elderly. Canadian Journal of Applied Sport Sciences,
9, 109-120.

Shephard, R. J., Corey, P., & Cox, M. (1982). Health hazard
appraisal: The influence of an employee fitness
program. Canadian Journal of Public Health, 33, 183-
187.

Sinyor, D. Schwartz, S. G. Peronnet, F., Brisson, G., &
Seraganian, P. (1983). Aerobic fitness levels and
reactivity to psychosocial stress: Physiological,
biochemical, and subjective measures. Psychosomatic
Medicine, 3, 205-217.

Slenker, S. D., Price, J. H., Roberts, J. M., & Jurs, S. G.
(1984). Joggers versus nonexercisers: An analysis of
knowledge, attitudes, and beliefs about jogging.
Research Quarterly for Exercise and Sport, 33, 371-
378.

Snyder, E. E., & Spreitzer, E. (1984). Patterns of
adherence to a conditioning program. Sociology of
Sport Journa , 33, 103-116.

Sonstroem, R. J. (1974). Attitude testing examining certain
psychological correlates of physical activity.

Research Quarterly, 93, 93-103.

 

147

Sonstroem, R. J. (1978). Physical estimation and attraction

scales: Rationale and research. Science in Sports,

 

39, 97-103.

Sonstroem, R.J. (1984). Exercise and self-esteem. In R.
L. Terjung (Ed.), Exercise and sport sciences reviews
(Volume 12, pp. 123-155). Lexington, MA: D. C.
Heath.

Spirduso, W. W. (1985). Does exercise contribute to longer

and/or better life? Gerontology Newsletter, 9,(4),

 

1-7.
Stamford, B. (1984). Exercise and longevity. The Physician

and Sportsmedicine, 112, 209.

 

Suinn, R. M. (1972). The Tennessee Self-Concept Sca1e. In

0. K. Buros (Ed.), The seventh mental measurements

 

yearbook (pp.367-369). Highland Park, NJ: Gryphon.
Surwit, R. R., Feinglos, M. N. & Scovern, A. W. (1983).
Diabetes and behavior: A paradigm for health

psychology. American Psychologist, 39, 255-262.

 

Thompson, J. K., & Martin, J. E. (1984). Exercise in health
modification: Assessment and training guidelines.

The Behavior Therapist, 3, 5-8.

 

Truax, C. B., & Carkhuff, R. R. (1965). Client and
therapist transparency in the psychotherapeautic

encounter. Journal of Counseling Psychology, 33, 3-9.

 

 

148

Trujillo, C. M. (1983). The effect of weight training and
running exercise intervention programs on the self-

esteem of college women. International Journal of

 

Sport Psychology, 39, 162-173.
Tu, J., & Rothstein, A. L. (1979). Improvement of jogging

performance through applications of personality

specific motivational techniques. Research Quarterly,

 

39, 97-103.
Tucker, L. A. (1983). Muscular strength: A predictor of

personality in males. Journal of Sportsmedicine and

 

Physical Fitness, 33, 213-219.
Van Andel, G. E., & Austin, D. R. (1984). Physical fitness

and mental health: A review of the literature.

Adapted Physical Activity Quarterly, 3, 207-220.

 

Walsh, W. M. (1983). Exercise for stress management:
Program design and trouble shooting. Postgraduate
Medicine, 39, 245-255.

Weinberg, R., Gould, D., & Jackson, A. (1979).
Expectations and performance: An empirical test of
Bandura's self-efficacy theory. Journal of Sport
Psychology, 3, 320-331.

Weinberg, R. S., Yukelson, D., & Jackson, A. (1980).
Effect of public and private efficacy expectations on

competitive performance. Journal of Sport Psychology,

 

3, 340-349.

 

149

Weinberg, R. S., Gould, D., Yukelson, D., & Jackson, A.
(1981). The effect of preexisting and manipulated
self-efficacy on a competitive muscular endurance
task. Journal of Sport Psychology, 9, 345-354.

Weinberg, R. S., Hughes, H. H., Critelli, J. W., England,
R., & Jackson, A. (1984). Effects of preexisting and
manipulated self-efficacy on weight loss in a self-
control program. Journal of Research in Personality,
_9, 352-358.

Young, R. J., & Ismail, A. H. (1976). Personality
differences in adult men before and after a physical
fitness program. The Research Quarterl , 93, 513-518.

Young, R. J., & Ismail, A. H. (1977). Comparison of
selected physical and personality variables in regular
and nonregular adult male exercisers. The Research

Quarterly, 93, 617-621.

1|1|WHIUUIIIUIIIW1111111111”11111111