Highlights from wave I of the National Survey of Personal
Health Practices and Consequences: United States, 1979.
Vital and Health Statistics, series 15, No. 1. DHHS publica-
tion No. (PHS) 81-1162. National Center for Health Statis-
tics, Hyattsville, MD, June 1981.
Canada Fitness Survey: Fitness and lifestyle in Canada.
Canada Fitness Survey, Ottawa, May 1983.
Bradstock, M. K., et al.: Behavioral Risk Factor Surveil-
lance, 1981-1983. CDC Surveillance Summaries. MMWR
33 (No. ISS): ISS-4SS (1984).
The Miller Lite report on American attitudes toward
sports. Miller Brewing Co., Milwaukee, 1983.
Kannel, W. B.: Habitual level of physical activity and risk
ofcoronary heart disease: the Framingham study. Can Med
Assoc J 96: 811-812 (1967).
Berkman, L. F., and Breslow, L.: Health and ways of
living: the Alameda County study. Oxford University
Press, New York, 1983.
Montoye, H. J.: Physical activity and health: an epidemi-
ologic study of an entire community. Prentice-Hall, Inc.,
Englewood Cliffs, NJ, 1975.
Taylor, H. L., et al.: A questionnaire for the assessment of
leisure time physical activities. J Chronic Dis 31: 741-755
Haskell, W. L., Montoye, H. J., and Orenstein, D. R.:
Physical activity and exercise to achieve health-related
components of physical fitness. Public Health Rep 100:
202-212, March-April 1985.
Chubb, M., and Chubb, H. R.: One third of our time? An
introduction to recreation behavior and services. John
Wiley and Sons, Inc., New York, 1981, p. 95.
Canada Fitness Survey: Canadian youth and physical activ-
ity. Canada Fitness Survey, 506-294 Albert St., Ottawa,
KIP 6E6, October 1983.
Powell, K. E., and Paffenbarger, R. S.: Summary of the
workshop on epidemiologic and public health aspects of
physical activity and exercise. Public Health Rep 100:
118-126, March-April 1985.
The Gallup Poll: Six of 10 adults exercise daily. Los
Angeles Times Syndicate, Los Angeles, May 1984.
18.Contemporary Research Centre, Ltd.: Behaviour and at-
titude toward physical activity among Canadians. Results
of a national survey conducted for PARTICIPaction.
Toronto, Canada M5H2A4, Toronto, May 1982. Mimeo-
Canada Fitness Survey: Exercise and sports: age is not a
barrier. Canada Fitness Survey Highlights No. 14, October
The use of time, edited by A. Szalai. Mouton, The Hague,
A dictionary of epidemiology, edited by J. M. Last. Oxford
University Press, New York, 1983.
Department of Health and Human Services: Promoting
health/preventing disease: objectives for the nation. U.S.
Government Printing Office, Washington, DC, fall 1980, (a)
Paffenbarger, R. S., Hyde, R. T., Wing, A. L., and Stein-
metz, C. H.: A natural history ofathleticism and cardiovas-
cular health. JAMA 252: 491-495, July 27, 1984.
Laporte, R. E., Montoye, H. J., and Caspersen, C. J.:
Assessment of physical activity in epidemiologic research:
problems and prospects. Public Health Rep 100: 131-146,
Montoye, H. J.: Estimation of habitual physical activity by
questionnaire and interview. Am J Clin Nutr 24: 1113-1118
Bouchard, C., et al.: A method to assess energy expenditure
in children and adults. Am J Clin Nutr 37: 461-467 (1983).
Sallis, J. F., et al.: Physical activity assessment methodol-
ogy in the Five-City Project. Am J Epidemiol 121: 91-106
Furrie, A. D., and Stephens, T.: Energy expenditure pat-
terns in the Canadian population. In Health risk estimation,
risk reduction and health promotion, edited by F. Landry.
Proceedings of the 18th annual meeting of the Society for
Prospective Medicine, 1982. The Canadian Public Health
Association, Ottawa, pp. 103-114, 1983.
The Determinants of Physical
Activity and Exercise
ROD K. DISHMAN, PhD
JAMES F. SALLIS, PhD
DIANE R. ORENSTEIN, PhD
Dr. Dishman is an Associate Professor, Department of Physi-
cal Education, University ofGeorgia, Athens, GA 30602. He was
formerly the Associate Director, Adult Fitness and Cardiopul-
monary Rehabilitation Exercise Programs, University of Cali-
fornia at Davis. Dr. Sallis is Assistant Adjunct Professor, Divi-
sion ofGeneral Pediatrics, University of California at San Diego,
La Jolla. Dr. Orenstein is a Research Psychologist in the Behav-
ioral Epidemiology and Evaluation Branch, Division of Health
Education, Center for Health Promotion and Education, Centers
for Disease Control, Atlanta, GA 30333.
Tearsheet requests to Dr. Dishman.
Evaluation and delivery ofphysical activity and
exercise programs appear impeded by the substan-
tial numbers of Americans who are unwilling or
unable to participate regularly in physical activity.
As a step toward identifying effective interventions,
we reviewed available research on determinants re-
lating to the adoption and maintenance ofphysical
activity. We categorized determinants as personal,
environmental, or characteristic ofthe exercise. We
have considered supervised participation
rately from spontaneous activity in the general
A wide variety ofdeterminants, populations, and
settings have been studied within diverse research
traditions and disciplines. This diversity and the
Public Health Reports
varied interpretation of the data hinder our clearly
summarizing the existing knowledge. Although we
provide some directions for future study and pro-
gram evaluation, there is a need for research that
tests hypotheses derived from theoretical models
and that has clear implications for intervention
programs. We still need to explore whether general
theories of health behavior or approaches relating
to specific exercises or activities can be used to
predict adoption and maintenance ofphysical activ-
NATIONAL GOALS call for participation in regular
and vigorous physical activity by 90 percent of
youth and 60 percent of adults by 1990 (1). At this
time, however, best estimates indicate that 41 per-
cent to 51 percent of adults are sedentary (2,3)
while only one-third of all adults participate in exer-
cise on a weekly basis. Just 15 percent are believed
to expend an energy equivalent (1,500 kcal per
week) ofknown epidemiologic significance (3,4). Of
those already regularly engaged in either group or
solitary exercise, about 50 percent will discontinue
activity at some time in the coming year (5-7).
Moreover, less than 10 percent of sedentary adults
are likely to begin a program of regular exercise
within a year (James F. Sallis, unpublished observa-
tions, November 18, 1982, and reference 3).
Estimates (reference 8 and Steven N. Blair, Insti-
tute for Aerobics Research, Dallas, TX, unpub-
lished observations, May 26, 1984) do show recent
increases in participation in activity that develops
However, these increases seem to occur in certain
population segments only-notably young adults,
the well educated, and members of high socioeco-
nomic groups (3,4).
These findings are similar to recent Canadian es-
timates (9); however, the U.S. increases are not as
high as the Canadian increases. According to avail-
able figures, our nationwide participation in all
types ofphysical activity has increased only slightly
(from 4 percent to 14 percent) during the past de-
cade (3,9,10). Although we cannot precisely iden-
tify the current nationwide rate (8), it seems un-
likely that the 1990 goals for the nation for participa-
tion in physical activity and exercise can be met
One barrier to developing effective methods to
encourage physical activity among all segments of
the population is lack of knowledge of the determi-
nants of regular physical activity. It appears that the
public health potential ofphysical activity and exer-
cise cannot be defined or fulfilled until the behav-
ioral determinants ofparticipation are identified and
subsequently managed; yet these determinants re-
main poorly understood.
Goals of This Review
The first goal of this paper is to review the scien-
tific literature on known determinants of regular
exercise and physical
activity. We categorized
these determinants by focusing on (a) characteris-
tics of the person and his or her lifestyle habits, (b)
characteristics of environments, and (c) charac-
teristics of the activity itself. This approach helped
to organize the review and to identify domains that
may account for the wide range of determinants
contributing to participation in physical activity.
Identifying these areas may also help specify seg-
ments of the population we need to target and im-
portant variables for future interventions.
The second goal is to identify what appear to be
the most important determinants in each of the
three categories previously described. Because any
single factor can be influential under certain condi-
tions, we present each individually. It is important
to note, however, that these factors probably inter-
act in complex ways and that their relative impor-
tance can vary. Few factors have shown behavioral
uniformity across settings, populations, and time
Our final goal is to recommend specific study
areas that could help us understand what motivates
people to become physically active and help us de-
velop ways to increase activity. Effective exercise
interventions will probably require that both ab-
stract (for example, beliefs) and concrete (for
example, disability) determinants be addressed in
complementary ways to (a) diminish or compensate
for psychological and physical or environmental
barriers to activity; (b) provide knowledge, skills,
and reinforcements that augment the willingness
and ability to be active; and (c) permit selection of
appropriate forms and intensities ofactivity. Under-
lying our recommendations is the need to integrate
the efforts of epidemiology, behavioral medicine,
health psychology, and exercise science under a
public health umbrella. In the past, the purpose,
methods, and scope of the various study ap-
proaches have not assured an orderly progression of
theoretical and practical knowledge. The disparities
March-April 1985, Vol. 100, No. 2
Table 1. Studies that have investigated determinants of physical activity and exercise
Past extra-program activity ...........
School athletics, 1 sport .............
Blue-collar occupation .................................................................................
Higher risks for coronary heart disease ............................
Health exercise knowledge, attitudes ..
Enjoyment of activity ........
Perceived health ..............................
Mood disturbance .....................................................................................
Perceived physical competence
Spouse support .
Perceived available time .............
Access to facilities ..................
Disruptions in routine ..................................................................................
Social reinforcement (staff, exercise partner) .........
Activity intensity .....
Perceived exertion .....................................................................................
Past participation .
Spontaneous physical activit,
Youth sport (organized) ..............
School athletics, 1 sport ............................................................
School athletics, >1 sport ............
Blue-collar occupation .............................................................
Health exercise knowledge ............
Health exercise attitudes ............
Perceived available time ..............
Access to facilities ....................
Family influences .
Peer influences .
in methods and standards (paradigms) need to be
diminished or reconciled in order to generate more
Organization of the review. This research review is
organized into separate discussions of studies of
"supervised" exercise programs and studies of
"spontaneous" changes in levels of physical activ-
ity within a population base. This distinction accu-
rately portrays the origin of most available data. It
also provides compatible data for future compari-
sons of the impact of small group interventions and
population trends. However, the behavioral sig-
nificance of supervision per se, while seemingly im-
portant for some people, is yet to be determined.
Within both sections-supervised exercise pro-
grams and spontaneous physical activity-personal,
environmental, and activity characteristics are con-
sidered. In addition, distinctions are made between
historical and contemporary influences, and be-
tween determinants of adoption or initiation of exer-
cise and maintenance of exercise habits. Each of
these distinctions can help us understand both theo-
retical and practical views of existing evidence.
Public Health Reports
review. Because this is a new area of study, we have
included both cross-sectional and correlational, as
well as experimental, data when they appear to add
to existing knowledge.
1 and 2 summarize the findings of our
Caveats. Because most available research has been
pragmatic, not theoretical, in origin, data have been
generated by different methods, from different pop-
ulations, with somewhat different outcomes
mind. Therefore, there is little standardization in
defining and assessing determinants and physical
activity. Also, the variables studied and time frames
sampled are inconsistent across population seg-
ments and settings. Thus, it is difficult to make
comparisons among studies.
Population surveys have the potential to be gen-
eralized to a larger population group. However,
these surveys rely on subjective, possibly inaccu-
rate, estimates of activity and a narrow set of possi-
ble determinants. These surveys often use cross-
sectional, retrospective designs, yield seemingly in-
congruous results, and do not permit a weighting of
relative influence between variables.
Studies of clinical settings and specific commu-
nity samples have used more precise measures of
activity and determinants, and they frequently ex-
amine variable interactions. However, their ability
to be generalized to other populations, settings, and
activities usually has not been tested and is, there-
fore, restricted. Although prospective studies are
common, there are relatively few controlled exper-
iments. Thus, there is some imbalance in methods
and knowledge across populations, settings, and ac-
tivity (see tables 1 and 2).
Each study has specific limitations because of
inadequacies in either measurement ofdeterminants
or activity patterns, sample size, and representa-
tiveness, or because of inadequate control or quan-
tification of possible confounding variables. All
these factors further limit the ability of the studies
to be generalized to other population groups. Also,
most studies have been descriptive, relying on cor-
data rather than
Hence, in most instances our use of the term "de-
terminant" indicates a reliable association or pre-
dictive relationship, not causation. Because of this
diversity, we are more confident when multiple
studies converge toward one result.
Although it is likely that different determinants
may exist for specific circumstances (for example,
cardiac rehabilitation versus worksite fitness) or for
different behaviors (for example, adopting versus
maintaining regular participation), it is noteworthy
Table 2. Summary of variables that may determine the proba-
bility of exercise
Changes in probability
Past program participation ..........
Past extra-program activity ..........
School athletics, 1 sport ............
School athletics, >1 sport ...........
Blue-collar occupation .............
High risk for coronary heart disease..
Type A behavior ....................
Health, exercise knowledge .........
Enjoyment of activity ...............
Perceived health ...................
Mood disturbance ..................
Expect personal health benefit ......
Self-efficacy for exercise ............
Intention to adhere .................
Perceived physical competence .....
Evaluating costs and benefits .......
Behavioral skills ....................
Spouse support ....................
Perceived available time ............
Access to facilities ..................
Disruptions in routine ...............
Social reinforcement (staff, exercise
Family influences ..................
Peer influence .....................
Physical influences .................
Medical screening ..................
Activity intensity ....................
Perceived discomfort ...............
KEY: ++ = repeatedly documented increased probability; + = weak or mixed
documentation of increased probability; 00 = repeatedly documented that there
is no change in probability; 0 = weak or mixed documentation of no change in
probability; - = weak or mixed documentation of decreased probability; - - =
repeatedly documented decreased probability. Blank spaces indicate no data.
that several factors (such as work status, smoking,
self-motivation, and social reinforcement) have
shown a fairly generalized relationship with activity
Supervised Exercise Programs
Influence of personal characteristics. Personal char-
acteristics are defined here as past or present
knowledge, attitudes, behaviors, personality char-
factors that may influence exercise habits.
traits, and demographic
March-April 1985, Vol. 100, No. 2
Compliance rates in two long-term clinical trials of exercise and
rehabilitation following myocardial infarction, Ontario Exercise
Heart Collaborative Study (OEHCS) and Goteborg, Sweden
SOURCE: Oldridge, N.B.: Compliance and exercise in primary and secondary
prevention of coronary heart disease: a review. Prev Med 11: 62, January 1982.
Copyright © 1982 by Academic Press, Inc. Used with permission.
In supervised programs where activity can be
directly observed, past participation in the program
is the most reliable correlate ofcurrent participation
(5-7,13), and this may account for 30 percent to 50
percent of the variance in participation in activity
across the first few months (Herman M. Frankel,
Kaiser-Permanente, Center for Health Research,
Portland, OR, unpublished observations, January
27, 1984). This variance holds for men and women
alike in adult fitness programs and is consistent with
observations in treatment programs for patients
with coronary heart disease and obesity. As shown
in the figure, the rate ofparticipation typically drops
within the initial 3 to 6 months, then plateaus and
continues a gradually decreasing but linear pattern
across the next 12 to 24 months. Individuals who
are still active after 6 months are likely to remain
active a year later (6,15).
The impact of previous activity not performed in
a supervised situation is less clear. In programs for
School of Public Health, University of Pittsburgh,
unpublished observations, May 23, 1984) and male
cardiac rehabilitation patients (6), routine walking
and active leisure predict participation in super-
vised programs, but the intensity, duration, and fre-
quency ofreported preprogram activity among male
cardiac patients has not been predictive of atten-
dance in a supervised program (16).
Although one cross-sectional study shows that
active male participants in adult fitness programs
are likely to have had a background in sports (17),
no prospective study has shown a relationship be-
tween adherence to cardiac rehabilitation exercise
programs and participation in interscholastic or in-
tercollegiate athletics (5,13,16). This illustrates the
need for cautious interpretation of studies using
retrospective designs. One study found that an en-
joyable elementary physical education program
predicts adherence to a supervised running program
in adult men (Ping Ho, Stanford University School
of Medicine, unpublished observations, June 7,
Current personal characteristics are strong de-
terminants of participation in clinical exercise pro-
grams. Blue-collar workers and smokers are likely
dropouts from cardiac rehabilitation exercise pro-
grams (6) and corporate exercise programs (12, 13).
Overweight persons are less likely to continue a
fitness program (15,18); even in gentle walking pro-
grams, up to 70 percent of obese people stop within
a year (Herman M. Frankel, unpublished observa-
tions, January 27, 1984) (19). Obese persons are
also less likely to respond to alternative activity
While neither circulatory disability nor a high de-
gree of aerobic fitness reliably predicts adherence to
supervised programs (5,6,13), some studies have
found positive relationships (6,7,15 and Ping Ho,
unpublished observations, June 7, 1984). However,
other studies show that men at risk for coronary
heart disease are not likely to enter an exercise
program without referral or are not likely to main-
tain the activity (6,12). Paradoxically, self-report of
the Type A, coronary-prone behavior pattern was
positively related to fitness gains in one study (21),
but people who show Type A behavior have been
early dropouts from cardiac rehabilitation (6) and
infrequent participants in corporate exercise pro-
grams (22). These findings collectively suggest that
those who could benefit most are most resistant to
increased activity in program settings.
Knowledge of and belief in the health benefits of
physical activity may motivate initial involvement
(23), but feelings of enjoyment and well-being seem
to be stronger motives for continued participation in
corporate programs (24). Patients who perceive
their health as poor are unlikely to enter or adhere
to an exercise program, and, if they do, they are
likely to perform little exercise (25). Those who
believe exercise has little value for health and
fitness and also believe health outcomes are out of
their control have been found to exercise less fre-
quently and to drop out sooner in fitness-related
programs than cohorts holding opposite views (7).
However, because most entrants into supervised
programs share similarly positive attitudes and be-
liefs about exercise, their self-perceptions of exer-
Public Health Reports
cise ability, feelings of health responsibility, and
attitude toward exercise do not predict who will
adhere to the program (13,26). In fact, there are
conflicting findings about the importance of health
beliefs concerning exercise (5,13,27 and Ping Ho,
unpublished observations, June 7, 1984). Health be-
liefs can influence the intention to be active, but
intentions have also failed to predict subsequent
participation (13,28). The roles of belief in personal
ability to exercise and control health outcomes, be-
lief in the value of exercise, anticipation of benefits
from it, and self-expectations to be active remain to
Although the importance of personality has not
been systematically studied, several studies show
that a self-motivation trait is related to program
adherence (11,18) and can help predict behavior
when combined with biological traits such as body
weight and composition (18,29). Also, self-per-
ceived mood disturbance is related to early drop-
out from adult fitness (29) and cardiac rehabili-
tation programs (30), and a depressive personality
(31) and somatization (32) have also been asso-
ciated with inactivity. The trait of extroversion has
been both positively (30) and negatively (12) related
to adherence, and ego-strength has shown both a
positive (30) and a neutral (18) influence.
Although these findings are too sparse to interpret
clearly, they collectively support the principle that
individual behavioral differences must be accom-
modated in the planning of supervised programs.
Concern for physical tolerance alone appears in-
adequate. From the standpoint of health risk, those
who may benefit the most from an exercise regimen
seem most resistive to adopting or maintaining one.
Thus, interventions aimed at personal change may
be more effective ifthey help people feel good about
themselves than if they focus exclusively on knowl-
edge of the health benefits of physical activity and
Influence of environmental factors. Environmental
factors can help or hinder physical activity. The
influences of the social environment on physical
activity habits include the attitudes offamily, peers,
and health professionals. Aspects of the physical
environment that may influence exercise include
weather, distance from facilities, and time pres-
sures. Also, there can be a difference in the actual
environment and the way people perceive the en-
vironment, and this needs to be assessed as well.
Support by a spouse is a consistent influence on
adherence to clinical exercise programs, and the
spouse's attitude can be more important than the
participant's (6,7,26,33). This finding illustrates the
power of the social environment to shape exercise
patterns. Personalized social reinforcement from
program staff or an activity partner has also been
found to be a potent determinant of adherence to
clinical programs in several studies (34-36) but not
in all studies (26,37).
The physical environment also influences adher-
ence to clinical programs. The most common reason
given for dropping out of programs is lack of time
(6,7,14). However, dropping out may reflect a lack
of interest, intention, or commitment, since regular
exercisers are as likely as (3), or more likely than
(9), the sedentary to view time as a barrier to activ-
ity. Both perceived convenience ofthe exercise set-
ting and actual geographic proximity to home or
place of employment are consistent predictors of
entry and continued participation in clinical pro-
Even among the active who are well-intentioned
and who value benefits of their participation in
exercise, unexpected disruption in the routine ofthe
activity or its setting can interrupt or conclude a
previously continuous exercise program (6,13). Life
occurrences such as relocation, medical events, and
periodic travel can impede the continuity of activity
and create new barriers.
It is believed that the impact of stressful events is
diminished as the activity habit becomes more es-
tablished (7). Interventions appear to help people
anticipate and plan for stressful events, recognize
these events as temporary impediments, and de-
velop appropriate self-regulatory skills (37,38).
In summary, there are several important envi-
ronmental influences on activity that exert their
influence outside the program setting. Thus, effec-
tive planning for adherence cannot rely solely on
management ofthe exercise environment. Although
proximity and convenience of the program are im-
portant, perceived lack of time and disruptions in
March-April 195, Vol. 100, No. 2
daily routines also interfere with participation.
These factors may require behavioral planning on
the part of the exerciser. Behavioral and cognitive-
behavioral strategies may be useful adjuncts to
existing exercise programs, but their specific impact
on increasing vigorous activity has not yet been
demonstrated. The influence of historical environ-
mental factors, such as social support during child-
hood, has not been studied, but adherence can be
fostered by individualized support by program staff
or exercise partners. Neither type of individualized
support, however, seems as influential as support
by a spouse, suggesting again that program adher-
ence cannot be understood by viewing the super-
vised setting alone.
Relatively few experimental studies of changes in
physical activity have been attempted. Those con-
ducted have applied general principles and tech-
niques of behavior modification to alter exercise
behavior rather than to manipulate known corre-
lates of exercise behavior. This has been partly a
pragmatic decision, because several known exer-
(for example, smoking, work
status, and overweight) are themselves difficult to
change. However, these determinants have not
been controlled, and therefore they may confound
The techniques and principles of behavior mod-
ification can be viewed either as reinforcement and
havioral and self-regulation skills. Behavioral ap-
proaches, including written agreements (39), behav-
ior contracts and lotteries (40,41), stimulus control
(42), and contingency incentives (36), have been
used successfully in case-control studies. Cognitive
approaches, including self-monitoring (39), sensory
distraction (34), goal setting (34,37,42), and de-
cisionmaking (35), have appeared equally effective
when used alone or when combined in intervention
Behavioral techniques are collectively associated
with a 10- to 25-percent increase in frequency of
physical activity, but we do not know their impact
on changes in intensity and duration of activity.
With few exceptions (43), studies have not fo-
cused on health advantages. Likewise, most inter-
ventions have lasted only 3 to 10 weeks, and only a
few show maintenance of activity in followup as-
sessments. Also, placebo control comparisons have
been infrequent. Therefore, we cannot make gener-
alizations about specific components of the inter-
ventions that are effective for specific populations.
Strategies that effect a change have in common a
dimension of social reinforcement (36). And they
appear more successful when carried out in groups
rather than in outside supervised settings (13,44).
Influence of the activity itself. Because the mode and
intensity of activity examined have not systemat-
ically differed across populations or settings, the
influences of activity characteristics on supervised
and spontaneous activity participation will be dis-
cussed together in the following section.
Spontaneous Physical Activity
Influence of personal characteristics. In one rep-
resentative population survey, about two-thirds of
adults with a history of participation in two or more
sports in their youth were physically active, and
they were two to three times more likely to engage
in vigorous activities than people who had not par-
ticipated in sports in their youth (10). Former par-
ticipants in only one sport had the same likelihood
of regular exercise as those with no history of par-
in sports. However, by middle age,
former male college athletes may be less active than
men who did not participate in sports (45). These
findings reinforce the need to distinguish between
sport, recreation, and fitness-related exercise when
describing activity patterns.
There appears to be a relatively strong relation-
ship between sport play in youth and involvement in
organized sports as an adult (3,45,46), especially
continuity in recreational sport activity between
adolescence and adulthood (45,46) than there is in
competitive sport activity. It appears that while
exercise or sport experience in youth can be a
strong agent in influencing exercise behavior in
adults, its influence is frequently overridden by
other personal and environmental influences. While
attitudes about exercise are related to sport partici-
pation in youth (48,49) and adulthood, they are
unrelated to adherence to leisure activity or fitness
programs in adults (13,26,50 and W. P. Morgan,
Sports Psychology Laboratory, University of Wis-
consin, Madison, unpublished observations, April
Public Health Reports
14, 1983). It is not known if factors related to the
decision to discontinue sports participation in youth
influence adult activity patterns. Also, there is little
information on activity patterns that do not relate
to sports in youth and future adult exercise habits.
Some current personal characteristics are consis-
tent in studies of supervised activities and in general
population studies. These findings indicate that
well-educated persons are more likely to exercise
(3,4,9,17,51 and Ping Ho, unpublished observa-
tions, June 7, 1984). There is also strong evidence
Blue-collar workers are less likely to engage in
either leisure or supervised exercise (6,12). Surpris-
ingly, in one community sample (51), a person's
body mass was not related to measures of activity.
Working women and single parents are more likely
to exercise than other groups (53), but this relation-
ship may be confounded with age.
There is little relationship between improving
knowledge about or attitudes toward exercise and
increased adherence (7,13,54). While both active
and inactive people view exercise as a positive
health behavior (3,9,26,49), those who strongly
value exercise, who believe they have control over
health outcomes, and who expect personal health
benefits from exercise are likely to engage in much
exercise (7,13). Knoweldge of health and exercise
was associated with improved maintenance of life-
style activities (routine leisure activities such as
walking) for men and women in one community
sample (James F. Sallis, unpublished observations,
November 18, 1982), but not with participation in
vigorous activity (exercise for fitness) (3,9). Thus,
no evidence supports the idea that increased knowl-
edge about exercise leads to enhanced participa-
tion. In fact, less than 5 percent of the population
believe that more information on fitness benefits
would be likely to increase their participation (3).
In one survey, more than 80 percent of both ac-
tive and inactive respondents felt that they should
exercise more than they do (4). Estimates from
several countries (9,55) indicate that more than half
the public is aware of fitness promotion programs,
but less than 20 percent of the active respondents in
one survey felt that they were influenced by such
programs (55). Thus, while the active are more
knowledgeable about exercise, it is unclear whether
such knowledge is an antecedent or a consequence
of involvement. Active people also believe in the
benefits of activity, but only one study has indicated
(James F. Sallis, unpublished observations, No-
vember 18, 1982). There are also mixed results on
the association between attitudes toward other
health behaviors and probability of regular exercise
(4,56,57). In one survey (9), regular activity was
rated seventh most important among health-related
behaviors. However, the active were twice as likely
(55 percent versus 26 percent) to view it as very
Perceived self-efficacy, or confidence in one's
ability to exercise (James F. Sallis, unpublished
estimates of the likelihood of adherence (Ping Ho,
unpublished observations, June 7, 1984) have pre-
dicted future activity, while perception of one's
overall physical competence has not (5,13,50).
These findings collectively suggest that if psycho-
logical factors are to be successfully used to predict
future activity, they should be directed toward
specific types or intensities of physical activity or
single exercise behaviors and time frames (for ex-
ample, adopting versus maintaining involvement)
rather than toward a broad and diffuse concept of
On the other hand, additional psychological data
indicate that broad behavioral traits may indeed be
linked with overall adherence to an exercise rou-
tine. If so, standardization ofvariables across future
studies would be enhanced. For example, those
who are self-motivated or have a generalized ten-
dency to follow through with behavioral decisions
(18) are more likely to continue exercise programs
corporate, and community
(7,11,29,56). This characteristic does not, however,
reliably predict daily participation or whether the
(11,29,35), suggesting that persons with high and
low self-motivation are equally likely to select activ-
ity environments that do not rely on professional or
social support. Self-motivated persons also appear
less sensitive to activity barriers, such as incon-
venience or competing lifestyle behaviors (7,11).
March-April 1985, Vol. 100, No. 2 165
In both clinical and community programs, the act
of rationally evaluating the anticipated benefits and
costs of being active has consistently facilitated in-
creased participation or a return to activity among
people who had been active but subsequently be-
came inactive (35). This increase in or return to
activity has been demonstrated only for short pe-
riods (4 to 10 weeks) among the already motivated.
Thus, this evaluation appears to prompt an existing
behavioral skill (20), or it may strengthen an earlier
decision to exercise.
It is believed that many who intend to be active
but remain sedentary (28) lack the self-regulatory
skills necessary to engage in the complex sets of
behaviors referred to as exercise habits. Short-term
studies suggest that interventions that teach goal
setting, planning, self-monitoring, and self-reward
skills can increase participation among people who
do intend to exercise (34). Long-term exercise goals
related to health and fitness are more predictive of
continued involvement than are short-range expec-
tations (3,34), although flexible daily goals can in-
crease maintenance of involvement. Feelings re-
lated to well-being and enjoyment seem more im-
portant to maintaining activity than concerns about
health (7,9,24). Also, the ability to make specific
plans to avoid relapse may be important (37,38).
Although dropouts from supervised exercise pro-
grams are more likely than adherents of these pro-
grams to view time conflicts and inconvenience of
the setting as barriers to participation, perceived
barriers to exercise were similar for active and inac-
tive respondents in one nationwide survey (4). The
principal barriers were lack of time (43 percent),
lack of willpower (16 percent), "just don't feel like
it" (12 percent), medical problems (9 percent), and
lack of energy (8 percent). This similarity suggests
that perceived barriers may often represent logical
post hoc explanations rather than true determinants
of inactivity. That nearly twice as many sedentary
(13 percent) as active people (7 percent) observed
that they "just don't feel like it" is consistent with
the view that perceived barriers may frequently re-
flect inadequate motivation to be active rather than
reasons for inactivity (9). This can be a critical
distinction, because no data support the notion that
removing stated barriers leads to increased activity.
In summary, the people most likely to engage
regularly in spontaneous exercise are well-edu-
cated, are self-motivated, and have the behavioral
skills to plan an exercise program and prepare
for relapses. But we need to know more about the
development of self-motivation and exercise behav-
ior skills. Active people tend to expect and believe
that they receive personal health benefits from
exercise, but positive feelings from activity seem
more important than beliefs. Perceived barriers to
exercise do not preclude participation. Knowledge
of the health benefits of exercise predicts lifestyle
activity but not fitness-related exercise; a history of
involvement in sports during youth may predict
physical activity in young adulthood, but not in
through the life cycle is poorly understood. It is not
known how or why people decide to become active
and to what degree they will be active, why a per-
son's activity often declines with age, or what might
be changed to prevent or diminish this decline.
continuity of activity
Influence of environmental factors. During child-
hood, family influence on exercise behavior is prob-
ably based on modeling of interests and skills, rein-
forcing behavior, and providing activity prompts
and settings. Activity modeling and support by the
mother are particularly strong predictors of later
exercise participation by daughters (47,48). How-
ever, early peer and family influences on later be-
havior have not been well studied, though it is rea-
sonable to expect some indirect effects. Although
the family seems to influence and be influenced by
the physical activity habits of its members (3,4),
most of the studies in this area are retrospective
surveys. Among parents who participate in sport,
nearly 60 percent spend one-half of their involve-
ment participating with the family (3). Among those
who choose exercise as leisure activity, 37 percent
have spouses who share their preference for exer-
cise over sedentary leisure pursuits (4).
Children's participation in physical activity out-
side structured programs
more by peers than by family (9,45). Peer influences
are sex-linked through adolescence but can become
independent ofgender in adulthood. Peer influences
appear to strengthen with age, moving from neigh-
borhood influences in childhood to people in the
place of employment in adulthood. Peers can exert
is perhaps influenced
Public Health Reports
the social influence to exercise more or less, and
they provide models for exercise-related attitudes
and behaviors (47). It is not clearly known, how-
ever, to what degree peers influence behavior change
or are chosen to match existing interests and skills.
As Iverson and colleagues (58) point out, physi-
cians represent potentially effective change agents
for increasing physical activity, but evidence on
their impact has been mixed. Estimates suggest
that among regular exercisers, one-fourth have been
advised to exercise by their physicians; however, in
one survey only 3 percent reported that this was the
reason for their activity (10). In more recent survey
data (9), 23 percent of adults cited doctor's orders
to be active as a very important reason for activity;
however, this was the seventh-ranked reason, and
the ranking did not differ between active and seden-
tary persons. This observation is interesting in light
of findings that 75 percent of family members ex-
press confidence in their physicians (4) and that
substantial proportions of both the inactive (43 per-
cent) and the active (32 percent) state that a physi-
cian's recommendation would likely increase their
involvement in sport (3).
Survey results indicate that enrollment fees are
not perceived as barriers to participation in com-
munity, corporate, or clinical exercise programs
(9,58). In one sample, just 10 percent of the public
reported that less expensive facilities would likely
increase their sports involvement. The already ac-
tive were paradoxically twice as likely as the inac-
tive (13 percent versus 6 percent) to hold this view
(3). While adherence rates are similar in commer-
cial programs and free corporate and research pro-
grams (12,34,35,59), the impact of cost on program
entry has not been determined. However, 75 per-
cent offamily members feel that medical checkups,
often recommended as a prerequisite to exercise
participation, are a financial burden (4).
In a survey of habitual runners, only 10 percent
reported that weather conditions had no impact on
their activity patterns
choices and seasonal feasibility of outdoor leisure
activities (9,46), but it is unknown to what extent
climate affects overall activity level. Very active
adults are more likely to reside in the West and
Midwest, while inactive adults are more likely to
reside in the South and East (3,8). This distribution
is confounded by age and socioeconomic factors,
however. Although 27 percent of the public state
that more favorable weather would probably in-
crease their sport involvement, nearly twice as
many active persons (33 percent) as inactive per-
sons (18 percent) believe this to be the case.
(60). Climate influences
Access to facilities is a necessary but insufficient
facilitator of community
(3,9). Contrary to findings concerning clinical exer-
cise programs, participants in unsupervised activity
who live closer to the exercise setting have been
shown to be more likely to drop out, stating that
they perceive inconvenience as a factor leading to
their return to inactivity (38). Also, the already
active are twice as likely as the inactive to feel that
greater availability of resources would increase
their participation (3,9). The already active are also
twice as likely to believe that a more flexible work
schedule would increase their participation (3,9).
The available evidence on environmental factors
suggests that the influence of the family on current
activity habits is strong but that family variables
have been insufficiently explored. Similarly, the ef-
fect ofpeers from childhood into adulthood appears
to be important but is poorly defined. While the
impact of physicians' recommendations of exercise
participation is potentially great, little effect has
been documented. Weather has a direct effect on
participation, but the effects of cost and conveni-
ence are complex and uncertain. One interpretation
of the effects of physical environments on activity
habits is that such considerations as time, money,
and convenience are irrelevant to those who have
not made the decision to exercise, while active
people are attuned to environmental barriers. The
long-term effects of the social environment and the
influences of past physical environments are largely
unknown, but they are likely to be minor in relation
to contemporary pressures.
Influence of the activity itself. Different activities
require different skills for maintained participation.
The commitment required to plan 30 minutes for a
walk before supper is probably less than that
needed tojoin a health club and make the necessary
preparations to swim or run for 30 minutes several
times per week. The behavioral differences between
March-April 1985, Vol. 100, No. 2 167
lifestyle and programmed activities have been dis-
cussed previously (49), and they have different
prevalence patterns (51).
Lifestyle activity patterns (that is, routine physi-
cal activity) do not differ greatly by age or gender,
but men and younger adults are much more likely to
engage in vigorous activities (that is, fitness-related
activities) (51). While more men than women will
adopt vigorous exercise in the period of a year, a
comparatively large proportion of women will in-
crease routine or lifestyle activities. Furthermore,
lifestyle activities show a dropout rate roughly
one-half of that typically seen for vigorous exercise
(James F. Sallis, unpublished observations, No-
1982). And in high-intensity, high-
frequency running programs, a substantial number
of participants can be expected to drop out because
of stress-induced injury (61). However, in a 4-year
study of a clinical exercise program
amount of exercise was unrelated to adherence.
Physical activity produces results that can en-
courage or discourage subsequent participation.
For example, perceived discomfort during an exer-
cise program, regardless of exertion, has been re-
ported among women who drop out (62). Some
highly committed exercisers may use activity as a
coping strategy for mental stress (13), and positive
feelings associated with activity may be associated
with an excessive dependence in certain types of
people (63). Numerous studies show that exercise is
related to positive mood and psychological func-
tioning (64,65), but the behavioral meaning of this
relationship is not yet known.
A daily routine that helps reinforce participation
and minimize impediments should include a conve-
nient and consistent time and place that is flexible
enough to accommodate a person's existing activity
preferences and daily fluctuations in motivation and
yet is planned to achieve tangible, long-term objec-
tives within a reasonable time (for example, 6-10
The available evidence indicates that lifestyle ac-
tivities have higher prevalence and adherence rates
than fitness-related (particularly aerobic) exercises.
But little is now known about the amount and mode
of activity needed from a behavioral health perspec-
tive, as opposed to an exclusively fitness-oriented
perspective. Likewise, we do not know the impact
of how people choose their activities and what role
this has in regular participation. However, it is clear
that the current practice of designing activity plans
exclusively on the basis of initial fitness and its
biological dose-response rate neglects important
perceptions of the effects of activity strain.
Regular participation in physical activity and
exercise must be viewed as a dynamic process in
which adoption and maintenance of involvement
are key outcomes. Intention, personal capabilities,
behavioral skills, commitment, and reinforcement
emerge as determining factors that appear constant
across populations, settings, and modes of activity.
Knowledge of, attitudes toward,,and beliefs about
health and activity; perceived needs and abilities;
and outcome expectations interact with biomedical
and personality traits, feelings, lifestyle behaviors,
and environments to influence a person's disposi-
tion to adopt or maintain involvement in physical
activity. This disposition is shaped by a history of
activity; normative modeling; reinforcement by
health care agents, family, and peers; environmen-
tal prompts; and accessibility of facilities. Although
the decision to be active or sedentary ultimately
resides in the individual person, evidence indicates
that this is not exclusively a reasoned decision.
Critical behavioral determinants may not be known
by the individual or may be outside the control of
the person's abilities or skills. Environmental bar-
riers can outweigh personal intention. Thus, physi-
cal activity and exercise are at once socially and
It is not possible to specify important interactions
among known determinants at this time. However,
it appears that some determinants are direct in their
influence while others operate indirectly, through
mediators. Some describe dynamic behavioral pro-
cesses but many mark personal and environmental
traits or behavioral products. It is necessary to rec-
ognize these distinctions because effective interven-
tions must allow for stable influences while attempt-
ing to modify those that are dynamic.
Recommendations for Future Research
The following recommendations are based upon a
public health perspective. While knowledge of de-
terminants of physical activity habits is a fruitful
area in which to test scientific theories and gather
basic data, the major application to public health is
that knowledge of determinants can guide interven-
tions. The questions below are those that we deem
important for future investigations to address. They
can be clustered in three major groupings, accord-
ing to need and outcome.
First, there remains a need to conceptualize and
in a general way rank determinants according to
priority. Our knowledge will continue to benefit
Public Health Reports
from replication, extension, and direct comparison of
factors implicated by previous studies. Therefore,
we need to:
1. Determine factors that lead to the decision or
intention to begin a physical activity program.
2. Specify the cognitive and behavioral skills or
physical abilities needed to initiate and maintain a
physical activity program.
3. Identify and put in priority the critical interac-
tions, within and among personal and environmen-
tal factors, that determine a person's willingness
and ability to be active.
4. Determine the degree to which influences on par-
ticipation may vary for different activity behaviors.
5. Determine the behavioral significance of per-
ceived barriers to activity and, likewise, excuses for
6. Examine the degree to which known health-risk
factors precede or follow inactivity, and why.
7. Determine how perceived exertion during and
after activity influences future activity.
Second, as our general knowledge grows, it will
also be necessary to specify major activity deter-
minants for certain populations and settings. There-
fore, we need to:
8. Study how activity determinants differ according
to a person's age, gender, ethnicity, socioeconomic
level, and health or fitness status.
9. Investigate possible differences in determinants
of lifestyle (routine leisure) and vigorous (exercise
for fitness) activities.
10. Establish whether determinants of participation
in supervised and unsupervised programs differ.
11. Determine if history of previous activity or
sports, family or peer influences, socioeconomic
status, education level, and age represent true in-
fluences on activity or if these represent a selec-
12. Determine who is most likely to follow and
benefit from programs of vigorous exercise, from
routine physical activity, and from activity modified
for disabling conditions (66).
13. Investigate the extent to which physical activity
and other health behaviors may reinforce or negate
Third, advancing age and elapsed time after initial
adoption of an activity are among the most reliable
predictors of inactivity. Thus, it seems likely that
past activity environments and experiences are
strong influences on present and future participa-
tion. Yet, little is now known in these areas. There-
fore, we need to:
14. Examine when and how preferences for types
and intensities of activity are formed and how they
influence future activity.
15. Establish what determinants at one age (for ex-
ample, childhood) can be altered to increase the
likelihood that the person will be active at another
age (for example, adulthood).
16. Determine if certain types of individuals are
predisposed to activity or inactivity and if this
changes at definable stages across the lifespan.
17. Examine why activity seems to diminish with
advancing age and what might be done to retard this
18. Investigate how self-motivation and intrinsic
reinforcements for regular participation develop,
what the time course
whether it is realistic to change them to increase
19. Determine if definable stages or patterns of par-
ticipation exist in which different determinants op-
erate or vary in their influence, making specific in-
terventions uniquely effective at different times.
is for their impact, and
To address the preceding questions and to pro-
vide guidance to those planning intervention pro-
grams, the following recommendations for research
and program design are offered.
1. Most population studies of determinants are
retrospective and cross-sectional. Current knowl-
edge suggests that a longitudinal propulation study
should be conducted. A representative sample (in-
cluding all sex, age, ethnic, socioeconomic, health,
and regional groups) should be assessed at intervals
of 3 to 4 years. A natural history of activity and fac-
tors associated with changing patterns will be use-
ful, but measures should be chosen based on prob-
able relevance to interventions targeted for people
2. Small descriptive and experimental studies, test-
ing the application of behavioral science theories to
determinants of activity habits, should be carried on
concurrently. These studies should focus on se-
lected groups (such as children, blue-collar work-
ers, working mothers, Hispanics, blacks, Asians,
those with high-risk health profiles, and the habitu-
ally active), or should test intervention hypotheses,
assess the interaction of selected determinants, or
validate survey reports of activity and its determi-
3. Study questions, variables, and measurement
methods should be standardized, or their disparities
March-April 1985, Vol. 100, No. 2
quantified and reconciled, to allow researchers to
examine whether the results of studies ofbehavioral
patterns and determinants can be generalized across
settings, activities, and population segments. Stan-
dardized questions concerning determinants should
be added to population surveys.
4. Most studies have been guided by applied con-
cerns rather than by theory. Although some at-
tempts to conceptualize existing evidence (7,13,14)
have helped form predictive hypotheses, there is a
need to continue bridging applied questions with
theory. This is important if physical activity, exer-
cise, and fitness are to be examined in relation to
other health behaviors and outcomes. Standardized
theories and technologies will allow us to determine
ifcommon determinants exist or ifmodels unique to
physical activity are needed. Although previous
tests ofgeneralized behavior models (such as health
locus of control, the health belief model, theory of
reasoned action, and self-efficacy or competence
motivation) have not been encouraging (18,27,
52,53,67,68), further efforts are warranted. This con-
clusion seems timely, since recent public health re-
views ofbehavioral change have not discussed theo-
ries within which exercise and physical activity are
target behaviors (69-72).
Department of Health and Human Services, Office of Dis-
ease Prevention and Health Promotion: Prevention '82.
DHHS Publication No. (PHS) 82-50157. U.S. Government
Printing Office, Washington, DC, 1982.
National Center for Health
Washington, DC, 1980.
The Perrier study: fitness in America. Perrier-Great Waters
of France, Inc., New York, 1979.
The General Mills American family report, 1978-1979:
Family health in an era of stress. Yankelovich, Skelly and
White, Inc., New York, 1979.
Morgan, W. P.: Involvement in vigorous physical activity
with special reference to adherence. In National College of
G. I. Gedvilas and M. E. Kneer. University ofIllinois Press,
Oldridge, N. B.: Compliance and exercise in primary and
secondary prevention of coronary heart disease: a review.
Prev Med 11: 56-70 (1982).
Dishman, R. K.: Compliance/adherence in health-related.
exercise. Health Psychol 1: 237-267 (1982).
Stephens, T., Jacobs, D. R., and White, C. C.: The descrip-
tive epidemiology of leisure-time physical activity. Public
Health Rep 100: 147-158, March-April 1985.
Canada Fitness Survey. Fitness and lifestyle in Canada.
Fitness Canada, Ottawa, 1983.
Clarke, H. H., editor: National adult physical fitness sur-
vey. President's Council on Physical Fitness and Sports
Newsletter, Washington, DC, special edition, May 1973.
Statistics: Health, United
Dishman, R. K.: Motivation and exercise adherence. In
Psychological foundations of sport, edited by J. Silva and
R. Weinberg. Human Kinetics Publications, Champaign,
IL, 1984, pp. 416-430.
Cox, M. H.: Fitness and lifestyle programs for business and
industry: problems in recruitment and retention. J Cardiac
Rehabil 4: 136-142 (1984).
Dishman, R. K.: Exercise adherence and habitual physical
activity. In Coping with mental stress: the potential and
limits ofexercise intervention, edited by W. P. Morgan and
S. N. Goldston. National Institute of Mental Health,
Rockville, MD. In press.
Martin, J. E., and Dubbert, P. M: Exercise applications and
promotion in behavioral medicine: current status and future
directions. J Consult Clin Psychol 50: 1004-1017 (1982).
Dishman, R. K.: Biologic influences on exercise adherence.
Res Q Exerc Sport 52: 143-159 (1981).
Dishman, R. K.: Prediction of adherence to habitual physi-
cal activity. In Exercise in health and disease, edited by
F. J. Nagle and H. J. Montoye, Charles C Thomas, Spring-
field, IL, 1981.
Harris, D. V.: Physical activity history and attitudes of
middle-aged men. Med Sci Sports 2: 203-208 (1970).
Dishman, R. K., and Ickes, W.: Self-motivation and adher-
ence to therapeutic exercise. J Behav Med 4: 421-438
Brownell, K., and Stunkard, A. J.: Physical activity in the
development and control of obesity. In Obesity, edited by
A. J. Stunkard. W. B. Saunders and Company, Inc.,
Philadelphia, PA, 1980.
Brownell, K., Stunkard, A. J., and Albaum, J.: Evaluation
and modification of exercise patterns in the natural envi-
ronment. Am J Psychol 137: 1540-1545 (1980).
Rejeski, W. J., Morley, D., and Miller, H. S.: The Jenkins
Activity Survey: exploring its relationship with compliance
to exercise prescription and MET gain within a cardiac
rehabilitation setting. J Cardiac Rehabil 4: 90-94 (1984).
Shephard, R. J., and Cox, M.: Some characteristics of
participants in an industrial fitness programme. Can J Appl
Sport Sci 5: 69-76 (1980).
Teraslinna, P., et al.: Characteristics affecting willingness
of executives to participate in an activity program aimed at
coronary heart disease prevention. J Sports Med Phys Fit-
ness 9: 224-229 (1969).
Morgan, P. P., Shephard, R. J., and Finucane, R.: Health
beliefs and exercise habits in an employee fitness pro-
gramme. Can J Appl Sport Sci 9: 87-93 (1984).
Shephard, R. J.: Physical activity and aging. Croom Helm,
Andrew, G. M., et al.: Reasons for dropout from exercise
programs in post coronary patients. Med Sci Sports Exerc
13: 164-168 (1981).
Lindsay-Reid, E., and Osborn, R. W.: Readiness for exer-
cise adoption. Soc Sci Med 14: 139-146 (1980).
MacKeen, P. C., Franklin, B. A., and Nicholas, W. C.:
Body composition, physical work capacity and physical
18-month follow-up of middle-aged
women participating in an exercise intervention program.
Int J Obes 7: 61-71 (1983).
Ward, A., and Morgan, W. P.: Adherence patterns of
healthy men and women enrolled in an adult exercise pro-
gram. J Cardiac Rehabil 4: 143-152 (1984).
Blumenthal, J. A., et al.: Physiological and psychological
variables predict compliance to prescribed exercise therapy
Public Health Reports
in patients recovering from myocardial infarction. Psychol
Med 6: 519-527 (1982).
Lobstein, D. D., Mosbacher, B. J., and Ismail, A. H.:
Depression as a powerful discriminator between physically
active and sedentary middle-aged men. J Psychosom Res
27: 69-76 (1983).
Collingwood, T. R., et al.: Caronical correlation analysis of
clinical and psychological data in 4,351 men and women. J
Cardiac Rehabil 3: 706-711 (1983).
Heinzelmann, F., and Bagley, R. W.: Response to physical
activity programs and their effects on health behavior. Pub-
lic Health Rep 86: 905-911 (1970).
Martin, J. E., et al.: The behavioral control of exercise in
sedentary adults: studies
Psychol. In press.
Wankel, L. M.: Decision-making and social-support strate-
gies for increasing exercise involvement. J Cardiac Rehabil
4: 124-135 (1984).
Stalonas, P. M., Johnson, W. G., and Christ, M.: Behavior
modification for obesity: the evaluation of exercise, con-
tingency management, and program adherence. J Consult
Clin Psychol 46: 463-469 (1978).
King, A. L., and Frederiksen, L. W.: Low-cost strategies
for increasing exercise behavior: relapse preparation train-
ing and social support. Behav Mod. In press.
Marlatt, G. A., and Gordon, J. R.: Determinants ofrelapse:
implications for the maintenance of behavior change. In
Behavioral medicine: changing health lifestyles, edited by
P. Davidson and S. Davidson. Brunner-Mazel, New York,
1980, pp. 410-452.
Oldridge, N. B., and Jones, N. L.: Improving patient com-
pliance in cardiac rehabilitation: effects of written agreement
and self-monitoring. J Cardiac Rehabil 3: 257-262 (1983).
Epstein, L. H., et al.: Attendance and fitness in aerobics
exercise: the effects of contract and lottery procedures.
Behav Mod 4: 465-479 (1980).
Wysocki, T., et al.: Behavioral management of exercise:
contracting for aerobic points. J Appl Behav Anal 12: 55-64
Keefe, F. J., and Blumenthal, J. A.: The life fitness pro-
gram: a behavioral approach to making exercise a habit. J
Behav Ther Exp Psychiatry 11: 31-34 (1980).
Epstein, L. H., Koeske, R., and Wing, R. R.: Adherence to
exercise in obese children. J Cardiac Rehabil 4: 185-195
Wilhelmson, L., et al.: A controlled trial of physical train-
ing after myocardial infarction. Prev Med 4: 491-508
Loy, J. W., McPherson, B., and Kenyon, G.: Sport and
social systems. W. B. Saunders and Company,
Philadelphia, PA, 1978.
The Miller Lite report on American attitudes toward sports.
Research & Forecasts, Inc., New York, 1983.
Greendorfer, S. L.: Shaping the female athlete: the impact
of the family. In The sporting woman, edited by M.
Boutslier and L. San Giovani. Human Kinetics Publishers,
Champaign, IL, 1983, 135-155.
Butcher, J.: Socialization of adolescent girls into physical
activity. Adolescence 18: 753-766 (1983).
Engstrom, L. M.: Physical activity of children and youth.
Acta Paediatr Scand [Suppl] 283: 101-105 (1980).
Sonstroem, R. J.: Attitudes and beliefs in the prediction of
exercise participation. In Sports medicine, sports science:
bridging the gap, edited by R. L. Cantu and W. J. Gillespie.
The Collamore Press, Lexington, MA, 1982, 3-16.
1 through 6. J Consult Clin
Sallis, J. F., et al.: Physical activity assessment methodol-
ogy in the Five City Project. Am J Epidemiol. In press.
demiological study of an entire community. Prentice-
Hall, Inc., Englewood Cliffs, NJ, 1975.
Gettman, L. R., Pollock, M. L., and Ward, A.: Adherence
exercise. Phys Sportsmed
Meyer, A., et al.: Skills training in a cardiovascular health
education campaign. J Consult Clin Psychol 48: 129-142
McIntosh, P.: "Sport for All" programs throughout the
world. Report prepared for UNESCO (Contract No.
207604). UNESCO, New York, 1980.
Fitness Ontario: The relationship between physical activity
and other health-related lifestyle behaviors. A research re-
port from the Ministry of Culture and Recreation, Sports
and Fitness Branch, Government of Ontario, Toronto,
Blair, S. N., Jacobs, D. R., and Powell, K. E.: Relation-
ships between exercise or physical activity and other health
behaviors. Public Health Rep 100: 172-180, March-April
Iverson, D. C., Fielding, J. E., Crow, R. S., and Christen-
son, G. M.: The promotion of physical activity in the U.S.
population: the status of programs in medical, worksite,
community, and school settings. Public Health Rep 100:
212-224, March-April 1985.
Fielding, J. E.: Effectiveness of employee health improve-
ment programs. J Occup Med 24: 907-916 (1982).
Sacks, M. H., and Sachs, M. L., editors: Psychology of
running. Human Kinetics Publishers, Champaign, IL, 1981.
Pollock, M. L., et al.: Effects offrequency and duration of
training on attrition and incidence of injury. Med Sci Sports
9: 31-36 (1977).
Oldridge, N. B., et al.: Predictive indices for dropout: the
Ontario Exercise Heart Collaborative Study Experience.
Am J Cardiol 51: 70-74 (1983).
Inger, F., and Dahl, H. A.: Dropouts from an endurance
training program. Scand J Sports Sci 1: 20-22 (1979).
Morgan, W. P.: Negative addiction in runners. Physician
Sports Med 7: 57-70 (1979).
Taylor, C. B., Sallis, J. F., and Needle, R.: The relationship
between physical activity and exercise and mental health.
Public Health Rep 100: 195-202, March-April 1985.
Dishman, R. K.: Medical psychology in exercise and sport.
Med Clin North Am. In press.
LaPorte, R. E., et al.: The spectrum of physical activity,
perspective. Am J Epidemiol 120: 507-517 (1984).
Godin, G., and Shephard, R. J.: Physical fitness promotion
programmes: effectiveness in modifying exercise behavior.
Can J Appl Sport Sci 8: 104-113 (1983).
Owen, N., and Lee, C.: Why people do and do not exer-
cise: recommendations for initiatives to promote regular,
vigorous physical activity in Australia. Department of Rec-
reation and Sport, South Australia, Sport and Recreation
Minister's Council, Adelaide, 1984.
Benfari, R. C., Eaker, E., and Stoll, J. G.: Behavioral
interventions and compliance to treatment regimes. Annu
Rev Public Health 2: 421-471 (1981).
Green, L. W.: Modifying and developing health behavior.
Annu Rev Public Health 5: 215-236 (1984).
Fielding, J. E.: Health promotion and disease prevention at
the worksite. Annu Rev Public Health 5: 237-265 (1984).
J.: Physical activity and health: an epi-
March-April 1985, Vol. 100, No. 2