Soc. Sci. Med. Vol. 46, No. 9, pp.

1175±1182, 1998
# 1998 Elsevier Science Ltd. All rights reserved
PII: S0277-9536(97)10045-4 Printed in Great Britain
0277-9536/98 $19.00 + 0.00

INTERNAL STRUCTURE OF A MEASURE OF SELF-

EFFICACY IN PHYSICAL ACTIVITY AMONG HIGH
SCHOOL STUDENTS
JOHN J. M. DWYER,1* KENNETH R. ALLISON2 and SUSAN MAKIN1
1
North York Public Health Department, 5100 Yonge Street, North York, Ont., Canada M2N 5V7 and
2
University of Toronto, Toronto, Ont., Canada

AbstractÐThe preliminary development of a measure of self-ecacy to participate in vigorous physical

activity when confronted with speci®c perceived barriers to physical activity is described. Measures used
in previous research ([Hofstetter, C.R., Hovell, M.F., Sallis, J.F., 1990a. Social learning correlates of
exercise self-ecacy: Early experiences with physical activity, Social Science and Medicine, 31, 1169-
1176.]; [Hofstetter, C.R., Sallis, J.F., Hovell, M.F., 1990b. Some health dimensions of self-ecacy:
Analysis of theoretical speci®city, Social Science and Medicine, 31, 1051-1056.]; [Reynolds, K.D.,
Killen, J.D., Bryson, M.S., Maron, D.J., Taylor, C.B., Maccoby, N., Farquhar, J.W., 1990. Psychoso-
cial predictors of physical activity in adolescents, Preventive Medicine, 19, 541-551.]; [Sallis, J.F., Pinski,
R.B., Grossman, R.M., Patterson, T.L., Nader, P.R., 1988. The development of self-ecacy scales for
health-related diet and exercise behaviors, Health Education Research, 3, 283-292.]) were adapted and
original items were developed. The 20-item measure has a 5-point Likert format ranging from not at all
con®dent (1) to very con®dent (5). An earlier pilot study of 200 secondary school students showed that
the measure was free of social desirability. The data were derived from a survey of 1041 secondary
school students from a Metropolitan Toronto board of education. Principal component analysis (PCA)
with oblique rotation of the data yielded two factors: self-ecacy to overcome external barriers and
self-ecacy to overcome internal barriers. The 12-item external barriers subscale had a coecient alpha
of 0.88 and the 8-item internal barriers subscale had a coecient alpha of 0.87. The subscales signi®-
cantly correlated with the frequency of participation in vigorous physical activity. In summary, the
results provide some support for the internal consistency reliability, construct validity, criterion validity,
and discriminant validity of the measure. Though further psychometric research is warranted, the self-
ecacy measure shows promise for physical activity research. # 1998 Elsevier Science Ltd. All rights
reserved

Key wordsÐphysical activity, self-ecacy, factor analysis

INTRODUCTION endorsed by the American College of Sports

Self-ecacy, a central construct of Bandura's social Medicine (1991). This de®nition of vigorous physi-
cognitive theory, represents the con®dence one has cal activity has been used in several previous studies
in being able to enact a certain behavior (Bandura, such as: Hofstetter et al. (1990a,b); Sallis et al.
(1988, 1989) and Sallis et al. (1992a,b).
1977, 1986). The theory emphasizes reciprocal re-
Perceived barriers to physical activity have been
lationships between personal factors (e.g., self-e-
examined in previous research. Respondents in the
cacy), environmental factors (e.g., barriers), and
Campbell's survey on well being (Stephens and
behavior (e.g., vigorous physical activity). In the
Craig, 1990) were asked to identify the importance
current study, one's con®dence to participate in vig-
of various factors (barriers) to being more physi-
orous physical activity when confronted with
cally active. A factor analysis of the 14 original bar-
speci®c perceived barriers was of interest. We
riers yielded 4 factors: lack of support and services;
de®ned vigorous physical activity as activity that
lack of con®dence and ability; lack of time; and
lasts for 20 min or more, makes your heart beat fas-
fear of injury (Wankel and Mummery, 1992).
ter, and makes you breathe a lot faster. For our
Multiple regression analysis indicated that lack of
study, we modi®ed measures of vigorous physical
con®dence and ability and fear of injury were the
activity used in the Campbell's survey (Stephens
best predictors of intention among the entire
and Craig, 1990) and the U.S. National Adolescent
sample. In another study, proximal variables
Student Health Survey (American School Health
(including perceived barriers) were strong correlates
Association, 1989). The measures we used retain the
of self-ecacy (Hofstetter et al., 1990a,b). Multiple
standard description of vigorous physical activity regression analysis revealed that perceived barriers
was one of the most important predictors of self-
*Author for correspondence. ecacy when controlling for the e€ects of other
1175

Study
Sallis et al. (1988)
Sallis et al. (1989)
Sallis et al. (1992a)
Hofstetter et al. (1990b)
Hofstetter et al. (1990a)
Reynolds et al. (1990)
Hovell et al. (1989)
Desharnais et al. (1986)
McAuley et al. (1991)
McAuley (1991)
McAuley (1993)
Courneya and McAuley (1994)
McAuley and Jacobson (1991)
Table 1. Summary of previous studies
Focus
Development of self-ecacy scales
Determinants of vigorous exercise
Explanation of vigorous exercise
Examined health dimensions of self-ecacy
Examined relationship between early
exercise experience and exercise self-
ecacy in adulthood
743 high school students
Predictors of physical activity in
adolescents
Correlates of walking for exercise
Self-ecacy and outcome expectations as
determinants of exercise adherence
Relationship between self-ecacy,
perceptions of success, and intrinsic
motivation for exercise
Relationship between ecacy cognitions,
causal attributions, and a€ective response
Self-ecacy and maintenance of exercise in
older adults
170 under-graduate students
Determinants of frequency, intensity and
duration of physical activity
58 sedentary adult females
Relationship between self-ecacy and
exercise in adult females
1176
Subjects Self-ecacy measures Findings
171 young adults 12 items regarding degree to which subjects Factor analysis identi®ed 2 factors:
sure they could exercise despite certain resisting relapse (5 items) and making time
factors for exercise (7 items)
2053 adults 3 items rated con®dence in ability to set
aside time to exercise, when feeling sad or Self-ecacy was variable most highly
under stress, or when family or social correlated with vigorous exercise in bi-
demands are great variate analysis
1739 adults Baseline self-ecacy was a signi®cant
As in Sallis et al. (1989) predictor of exercise change
525 adults 3 items assessing degree to which subjects
are sure they could exercise regularly, Exercise self-ecacy shown to be distinct
when family and social life take a lot of dimension of self-ecacy in relation to 4
time, and an outcome variable other factors
2053 adults 3 items assessing degree to which subjects
are sure they can exercise when feeling sad
or highly stressful, when family or social Proximal factors, such as perceived
life takes a lot of time, and when set aside barriers, found to be more predictive of
time for exercise exercise self-ecacy than distal factors
Rated how con®dent respondents felt
about their ability to exercise despite Exercise self-ecacy was signi®cantly
speci®c obstacles (6 items) and in general related to exercise for females at 16
(2 items) months postbaseline, but not for males
2053 adults As in Sallis et al. (1989) and Sallis et al. Exercise self-ecacy was a signi®cant
(1992a,b) predictor of walking for exercise
John J. M. Dwyer et al.
98 adults 1 item asking subjects to indicate to what
extent they expected to be capable of Self-ecacy was a signi®cant predictor of
attending the program until completion exercise adherence
265 young adults
5 items rating con®dence to exercise Self-ecacy accounted for signi®cant
vigorously on a regular basis variance in intrinsic motivation
80 middle-aged adults 10-item exercise self-ecacy scale regarding
subjects' perceived capabilities to exercise 3 Self-ecacy had signi®cant direct and
times a week despite barriers indirect e€ects on a€ect
66 middle-aged adults Self-ecacy signi®cantly predicted exercise
As in McAuley (1991) at 4-month postbaseline
Utilized 3 7-item scales to rate con®dence
in relation to frequency, intensity and Self-ecacy explained unique variance in
duration frequency and intensity, but not duration
7 items rating subjects' capabilities to
continue in an exercise program despite Self-ecacy was a signi®cant predictor of
barriers overall exercise level
Continued opposite
 Self-eciency in physical activity 1177
able to exercise at progressively higher perceptions of e€ort expenditure and more

items relating to subjects' con®dence to Subjects with higher self-ecacy regarding

An unspeci®ed number of items assessing Exercise ecacy was predictive of exercise

variables. In other research, perceived barriers was

Higher levels of self-ecacy found among

High exercise self-ecacy related to lower

con®dence in ability to persist with exercise Higher levels of exercise self-ecacy were
barriers had higher levels of participation

found among those currently exercising

Highly ecacious subjects had lower

the strongest predictor of jogging versus non-exer-

intention but not of frequency of

positive a€ect during exercise

cising (Slenker et al., 1984) and of vigorous exercise

those currently exercising

health risk behavior
(Sallis et al., 1989).

participation
There is substantial indication in the research lit-
erature that self-ecacy is an important correlate of
physical activity participation (Dishman et al.,
1985; Duncan and McAuley, 1993; Hofstetter et al.,
1990a,b; Hovell et al., 1989; McAuley, 1993;
McAuley and Jacobson, 1991; Reynolds et al.,
1990; Sallis and Hovell, 1990; Sallis et al., 1989).
Several extensive reviews addressed this topic
higher levels; and an unspeci®ed number of

regarding exercising regularly in next year

5-item self-ecacy measure used to assess

5-item self-ecacy measure used to assess

(Dishman et al., 1985; Godin and Shephard, 1990;

Subjects rated their con®dence in being

1 item based on how sure subjects are

con®dence to exercise at progressively
An unspeci®ed number of items used.

subjects' perceived con®dence in their

Two measures used: a scale assessing

con®dence in ability to participate in

abilities to exercise despite obstacles

Sallis and Hovell, 1990). A summary of previous

exercise despite speci®c barriers

exercise in speci®c situations

relevant studies is presented in Table 1.

in speci®c situations

Several existing self-ecacy measures have been

used in physical activity research but the measures
levels

do not address the full range of perceived barriers

to physical activity participation. For example, a
self-ecacy to exercise measure comprised of three
items was used in previous research (Hofstetter et
al., 1990a,b; Sallis et al., 1992a,b). Further, two stu-
dies were conducted to develop self-ecacy
measures for health behaviors including exercise
(Hofstetter et al., 1990a,b; Sallis et al., 1988).
1063 government employees and 429

1093 employees and 801 additional

85 sedentary middle-aged adults

Adults in one study were asked about their ability

to perform activities and the consequences of their
88 sedentary adults

hospital employees
2524 adult seniors

activities related to health and political issues

76 female adults

employees

(Hofstetter et al., 1990a,b). A factor analysis of the

data produced ®ve factors: ``political self-ecacy'',
``medical care self-ecacy'', ``exercise self-ecacy'',
``diet self-ecacy'', and ``outcome ecacy''. Three
items loaded signi®cantly on the exercise self-e-
cacy factor but only one of the items referred to
barriers.
In another study, university students were asked
Self-ecacy relationships with a€ective and

readiness, self-ecacy and decision-making

Relationship between exercise self-ecacy
Ecacy cognitions and social support in

Relationship between goal-setting, self-

Relationship between self-ecacy and

about their con®dence to motivate themselves to

Relationship between motivational

engage in diet and exercise behaviors (Sallis et al.,

exertion responses to exercise

1988). A factor analysis of the exercise items yielded

and stages of change
ecacy and exercise
exercise adherence

health behaviors

two factors: ``resisting relapse'' and ``making time

for exercise

for exercise''. However, the subscales consisted of

items that referred to an exercise program whereas
the current study required a measure that relates to
self-ecacy to participate in vigorous physical ac-
tivity, which is more inclusive than a formal exer-
cise program. Also, the subscale items addressed a
number of barriers of interest (e.g., time, work,
mood, stress) but the measure developed for the
current study addressed a wider range of barriers
including lack of athletic ability, facilities, support,
money, and fun.
McAuley and Courneya (1992)

Duncan and McAuley (1993)

There are few measures of physical activity self-

Poag and McAuley (1992)

Marcus and Owen (1992)

Grembowski et al. (1993)

ecacy which have been utilized with youth or ado-

lescents. Reynolds et al. (1990) report utilizing an
Marcus et al. (1992)

eight-item unidimensional scale to assess physical

activity self-ecacy among a cohort of 10th grade
students. Only a few of these items pertained to the
con®dence students felt in being able to overcome
speci®c obstacles to physical activity.
1178 John J. M. Dwyer et al.
The purpose of this article is to describe the pre-
liminary development of a measure of student self-
ecacy to participate in vigorous physical activity
when confronted with a broad range of perceived
barriers. This study expanded on a pilot study in
which 200 students (112 9th-graders and 88 11th-
graders) in a Metropolitan Toronto secondary
school completed the self-ecacy measure and
other measures such as a social desirability scale
(Crowne and Marlowe, 1960, 1964). The self-e-
cacy items were found to be free of social desirabil-
ity in that none of the self-ecacy items correlated
signi®cantly with the social desirability scale (range
from r = ÿ 0.09, p>0.05 to r = 0.14, p>0.05), an
indication of discriminant validity.
METHOD
Participants
The data presented in this paper were derived
from a sample of 1041 students, aged 20 years or
less, from eight secondary schools in a large metro-
politan Toronto board of education. The sample
included 688 students from grade 9 and 353 stu-
dents from grade 11, and the gender distribution
consisted of 529 males and 511 females (1 respon-
dent did not indicate his/her sex). The study
received ethical approval from the board of edu-
cation, the university, and the public health depart-
ment.
Sample design
The survey used a two-stage cluster sample design
(Babbie, 1973; Green and Lewis, 1986; Warwick
and Lininger, 1975). In the ®rst stage of sampling,
12 of 18 eligible secondary schools containing both
grade 9 and 11 students were randomly selected.
Principals from 8 of the 12 selected schools agreed
to participate in the study. In the second stage,
classrooms were randomly selected from the 8
schools. Students completed the questionnaire
during regular class time. The overall student re-
sponse rate was 81% (i.e. 1041 eligible completions/
(1041 eligible completions +217 students absent
+21 students refusing))*. Thus, we believe the qual-
ity of the sample to be high.
Measures
A measure of self-ecacy for participation in vig-
orous physical activity was developed by adapting
measures used in previous research (Hofstetter et
al., 1990a,b; Reynolds et al., 1990; Sallis et al.,
1988) as well as developing original items. The cur-
rent measure was partly derived from instruments
that have good psychometric properties. For
example, Reynolds et al. (1990) reported an alpha
*Eligible students were grade 9 and grade 11 students
aged 20 years or less.
reliability of 0.89 for their self-ecacy scale and
Sallis et al. (1988) reported alpha reliability coe-
cients of 0.85 for the resisting relapse factor and
0.83 for the making time for exercise factor and a
0.68 test±retest reliability for both self-ecacy for
exercise factors. The current measure is comprised
of 20 items that assess the degree of con®dence on
a ®ve-point Likert scale, ranging from not at all
con®dent (1) to very con®dent (5), to participate in
vigorous physical activity if speci®c perceived bar-
riers to physical activity are present (e.g., lack of
time due to school work; lack of support from
family; do not have fun). In addition, there is an
item measuring global self-ecacy for participation
in vigorous physical activity (i.e. degree of con®-
dence to participate in vigorous physical activity at
least three days per week) but this item was not
included in the analysis of the situation-speci®c
items. The items are listed in Appendix A.
Frequency of participation in vigorous physical
activity was measured by asking students to report
how many days in an average week they normally
participate in vigorous physical activity, at the pre-
sent time, in (a) physical education class, (b) other
school settings (e.g., intramurals; school teams),
and (c) outside of school.
Also, there were questions to determine the stu-
dents' age and sex and the highest level of edu-
cation that their father and mother completed, as a
measure of socioeconomic status.
RESULTS
Factor structure of measure
SPSS/PC+ (Norusis, 1990) was used to analyze
the data. The sample size of 1041 was sucient to
produce reliable correlation coecients considering
the Comrey (1973) and Gorsuch (1983) rec-
ommended sample size for factor analysis. An
inspection of the correlation matrix for the self-e-
cacy items revealed that most of the correlations
were greater than 0.30. Therefore, some clustering
of items was expected and exploratory factor analy-
sis was deemed appropriate in this early stage of
research.
PCA with oblique rotation was performed on the
students' responses to the 20 self-ecacy items.
Oblique rotation, which allows the factors to be
statistically related (Diekho€, 1992; Tabachnick and
Fidell, 1995), was used because it was expected that
the factors underlying self-ecacy would be corre-
lated in reality.
An initial analysis with PCA was conducted to
identify the number of factors with eigenvalues of
1.0 or greater, which is an estimate of the maximum
number of stable factors (Tabachnick and Fidell,
1995). This analysis yielded three factors with eigen-
values of 1.0 or greater. However, Comrey (1978)
contends that this factor extraction technique fre-
 Self-eciency in physical activity
Table 2. Factor loadings for principal component analysis with
oblique rotation of self-ecacy in physical activity items: total
sample (n = 1041)
Item Factor
external
barriers
Lack of support from friends 0.74
Lack of support from family 0.72
Lack of time due to family 0.72
responsibilities
Lack of time due to part-time work 0.70
Lack a partner 0.68
Lack of facilities 0.61
Participation costs money 0.59
Lack of programs 0.58
Lack of time due to school work 0.56
Lack athletic ability 0.53
Lack self-discipline or willpower 0.52
Lack of time due to other interests 0.49
Feel discomfort
Do not feel in the mood
Feel stressed
Fear of injury
Long-term illness, disability, or injury
Do not have fun
Self-conscious
Lack energy 0.30
Eigenvalue 7.89
Percent of variance 39.4
quently generates too many factors. The scree test
(Cattell, 1966), another factor extraction technique,
suggested the existence of two factors. There
appeared to be a break between the steep slope that
joined factors 1 and 2 and the gentle slope that
joined the remaining factors. Tabachnick and Fidell
(1995) recommend exploring factor solutions in
which one or two factors less than or more than the
number of factors suggested by the scree test are
speci®ed. Factor solutions with one, three, and four
factors speci®ed were subsequently examined.
However, the two-factor solution was deemed to be
most appropriate based on theoretical grounds and
interpretability.
Table 2 shows the factor loadings from the pat-
tern matrix for the two-factor solution. A loading
of 0.30 or greater, which represents at least a 9%
variance overlap between the item and factor, was
selected as the cuto€. Loadings less than 0.30 are
not depicted in the table. None of the items had
substantial loadings on both factors. Factor 1 was
comprised of items relating to lack of support
(from friends and family), lack of time (due to
family responsibilities, part-time work, school work,
and other interests), and lack of resources (partner,
facilities, programs, and money). This factor
appears to represent con®dence to participate in
vigorous physical activity when confronting external
barriers. Only two items, speci®cally the items con-
cerned with athletic ability and self-discipline, were
less relevant to con®dence to overcome external
barriers. Both items had lower loadings. The exter-
nal barriers factor accounted for 39% of the var-
iance in the set of 20 items. Factor 2 consisted of
items relating to a€ect (discomfort, mood, stress,
1179
fear of injury, not having fun, and self-conscious)
and physical status (illness and lack energy). This
factor, which accounted for an additional 8% of
internal
the variance, appears to represent con®dence to
barriers participate in vigorous physical activity when facing
internal barriers. The ÿ0.61 correlation between
both factors indicated a relationship between the
external barriers and internal barriers dimensions
and supported the decision to utilize oblique ro-
tation.
A factor solution is stable if di€erent extraction
techniques produce a similar solution (Tabachnick
and Fidell, 1995). To examine stability, principal
axis factoring (PAF) with oblique rotation for a
two-factor solution was performed on the students'
ÿ0.83 responses to the self-ecacy items. PAF yielded a
ÿ0.81
ÿ0.77
factor solution that was very similar to the solution
ÿ0.74 generated by PCA. PAF generated external barriers
ÿ0.69 and internal barriers factors. The same items loaded
ÿ0.63
ÿ0.62 on each factor and the size of their loadings were
ÿ0.42 similar.
1.54
7.7 Internal consistency reliability of measure
The internal consistency of the factor-derived
subscales was calculated. The 12-item external bar-
riers subscale had a coecient alpha of 0.88 and
the 8-item internal barriers subscale had a 0.87
coecient alpha which indicate high internal con-
sistency reliability.
Validity of measure
Pearson correlations (two tailed) between the
physical activity self-ecacy subscales and other
measures were computed to establish the validity of
the self-ecacy measure. First, both subscales were
signi®cantly related to the global measure of self-
ecacy. Global self-ecacy correlated.40
(p < 0.0001) with the external barriers subscale and
ÿ0.28 (p < 0.0001) with the internal barriers sub-
scale.
Also, the results showed that the self-ecacy
measure predicts participation in vigorous physical
activity, an indication of criterion validity. The
external barriers subscale correlated signi®cantly
and positively with frequency of participation in
vigorous physical activity in physical education
class (r = 0.10, p < 0.01), in other school settings
(r = 0.22, p < 0.0001), and outside of school
(r = 0.28, p < 0.0001). The internal barriers sub-
scale correlated signi®cantly and negatively with fre-
quency of participation in vigorous physical activity
in physical education class (r = ÿ 0.11, p < 0.001),
in other school settings (r = ÿ 0.17, p < 0.0001),
and outside of school (r = ÿ 0.24, p < 0.0001).
With regard to the relationship between the self-
ecacy subscales and socio-demographic variables,
the subscales were not signi®cantly related to age.
However, they were signi®cantly related to sex
(external barriers subscale, r = ÿ 0.09, p < 0.01; in-
ternal barriers subscale, r = 0.11, p < 0.001).
1180 John J. M. Dwyer et al.
Finally, the relationship between the self-ecacy
subscales and parents' education was not signi®-
cant.
DISCUSSION
This study examined the factor structure of physi-
cal activity self-ecacy among adolescents, a group
for which little research has been conducted on self-
ecacy or the determinants of physical activity
more generally (Dishman et al., 1985; Sallis et al.,
1992a,b). The most closely related previous study
(Reynolds et al., 1990) reported evidence in support
of a unidimensional scale of exercise self-ecacy,
while the analysis reported here suggests that physi-
cal activity self-ecacy is multidimensional. This
di€erence may be partly explained by the former
study's use of the term ``motivate myself'' in the
wording of several scale items. For example, one
item reads ``I am con®dent that I can motivate
myself to exercise no matter how stressed I feel''. In
contrast, the self-ecacy items in the current study
do not use the term ``motivate myself'' and are
more focused on the individual's perceived capa-
bility to participate despite barriers.
Previous research on the factor structure of exer-
cise self-ecacy among adults also indicates that it
is a multidimensional construct (McAuley and
Jacobson, 1991; Sallis et al., 1988). While these stu-
dies included several items concerning the perceived
con®dence of individuals to engage in physical ac-
tivity despite barriers, none of them identi®ed the
factors as self-ecacy regarding external barriers
and self-ecacy regarding internal barriers. The fac-
tors identi®ed by Sallis et al. (1988) were called
``resisting relapse'' and ``making time for exercise''.
While not labelled as such, many of the ``resisting
relapse'' items deal with conditions or situations
(external barriers), such as family demands, house-
hold chores, excessive demands at work, and social
obligations. Similarly, some of the ``making time
for exercise'' items could be considered to be in-
ternal barriers, e.g., a tiring day at work and feeling
depressed.
Self-ecacy factors identi®ed by McAuley and
Jacobson (1991) as barriers to participation in an
exercise program included ``lack of progress'' and
``other commitments''. Again, there is some corre-
spondence in the factor groupings of these items
and those from the current study. For example,
some of the ``lack of progress'' items included dis-
comfort and progressing at a slower rate than
others (internal barriers). Some of the ``other com-
mitments'' items correspond to external barriers,
such as a family member becoming ill or an event
con¯icting with the time of the exercise class. Thus,
while the speci®c items used in the Sallis et al.
(1988) and McAuley and Jacobson (1991) studies
di€er from those used in the current study, there
are hints of common underlying factor structures
across these measures. That is, the derived factor
structures in these studies could also be conceptual-
ized as distinguishing between self-ecacy regarding
dealing with internal barriers and self-ecacy
regarding dealing with external barriers.
Based on the ®ndings, levels of self-ecacy di€er
according to whether potential obstacles to physical
activity participation are attached more to external
circumstances and in¯uences or to internal issues or
motivation. These distinctions are similar, concep-
tually, to the Dishman et al. (1985) reference to
``environmental'' and ``personal'' characteristics in
their review of the determinants of physical activity
and exercise. However, they considered self-ecacy
to be a ``personal characteristic'' and listed a large
number of potential barriers (cost, climate, family
in¯uences, etc.) as ``environmental characteristics''.
While their distinction between environmental and
personal characteristics is compatible with the
Bandura (1977, 1986) notion of the reciprocal re-
lationship between environmental factors, personal
factors, and behavior in his social learning theory,
it does not recognize the likelihood that one's sense
of ecaciousness regarding a particular behavior
itself varies according to both external events and
in¯uences and more personal, internal features of
one's experience.
The loading of speci®c self-ecacy items on the
two factors (self-ecacy regarding external barriers
and self-ecacy regarding internal barriers) makes
sense from a theoretical perspective. One would
expect ecaciousness to vary according to the
degree to which individuals believe in their capabili-
ties to enact a behavior in speci®c situations. The
current ®ndings suggest that a sense of personal e-
cacy varies depending on whether events and situ-
ations are external or internal to the individual.
Theory would suggest that ecaciousness regarding
external barriers would be lower than ecacious-
ness regarding internal barriers. But internal bar-
riers are not necessarily a product of lack of
motivation. For example, pain and discomfort, feel-
ing sad, or under stress are treated conceptually as
internal barriers. Yet, it could be argued, physical
activity self-ecacy may be lower for those situ-
ations that have more a€ective characteristics.
Thus, individuals may actually feel more ecacious
regarding lack of time, lack of family or friends'
support (external barriers) than they do for feelings
of sadness, stress, or discomfort (internal barriers).
While the primary focus of this preliminary devel-
opment of the self-ecacy measure was to examine
its internal structure, we also wanted to examine
the issues of the reliability and validity of the
measure. Both subscales, which are comprised of
multiple items to adequately represent the two
speci®c domains, have high internal consistency re-
liability. The study demonstrated some support for
the internal consistency, criterion validity, and dis-
criminant validity of the self-ecacy measure.
 Self-eciency in physical activity
In terms of criterion validity, the two self-e-
cacy factors were shown to be predictive of par-
ticipation in vigorous physical activity in three
settings, a relationship supported in several other
studies (Hovell et al., 1989; Sallis et al., 1988,
1989, 1992a,b). While self-ecacy regarding exter-
nal barriers was found to be positively correlated
with participation, self-ecacy regarding internal
barriers was found to be negatively correlated.
Further analysis of these relationships will examine
possible explanations for these di€erences, since
self-ecacy theory suggests that both self-ecacy
factors would be positively related to participation.
While our analysis provided some evidence in sup-
port of the discriminant validity of the physical ac-
tivity self-ecacy factors, further research should
examine the relationship between these factors and
additional constructs such as the Health Locus of
Control scales (Wallston and Wallston, 1981).
In summary, the analysis reported here provides
further empirical support for the relevance of
Bandura's social cognitive theory to studies of the
factors in¯uencing health-related behavior.
Speci®cally, the analysis indicates that self-ecacy
for vigorous physical activity by teenagers is a mul-
tidimensional construct comprised of two meaning-
ful and interpretable factors, self-ecacy regarding
participating despite external barriers and self-e-
cacy for participating despite internal barriers.
While con®rmatory factor analysis is needed to sub-
stantiate these initial ®ndings, our analysis suggests
that physical activity self-ecacy is situation-
speci®c, adding support to existing theory on this
subject.
AcknowledgementsÐThe authors were project investi-
gators of the North York Community Health Promotion
Research Unit, supported by a Health System-Linked
Research Unit grant from the Ontario Ministry of Health.
Dr. K.R.A. is a Career Scientist of the Ontario Ministry
of Health, Health Research and Development Program.
This research was supported by a grant from the
Canadian Fitness and Lifestyle Research Institute (project
#942R017). Appreciation is extended to Patricia McNair
for her assistance in data collection and data processing
and to Dr. Edward Adlaf and two anonymous reviewers
for their comments on this paper.
REFERENCES
American College of Sports Medicine (1991) Guidelines for
Exercise Testing and Prescription (4th ed.). Lea and
Febiger, Philadelphia.
American School Health Association (1989) The National
Adolescent Student Health Survey: A Report on the
Health of America's Youth. Third Party Publishing
Company, Oakland, CA.
Babbie, E. R. (1973) Survey Research Methods.
Wadsworth Publishing Company, Belmont, CA.
Bandura, A. (1977) Social Learning Theory. Prentice-Hall,
Englewood Cli€s, NJ.
1181
Bandura, A. (1986) Social Foundations of Thought and
Action: Social Cognitive Theory. Prentice-Hall,
Englewood Cli€s, NJ.
Cattell, R.B. (1966) The scree test for the number of fac-
tors. Multivariate Behavioral Research 1, 245±276.
Comrey, A. L. (1973) A First Course in Factor Analysis.
Academic Press, New York.
Comrey, A.L. (1978) Common methodological problems
in factor analytic studies. Journal of Consulting and
Clinical Psychology 46, 648±659.
Courneya, K. and McAuley, E. (1994) Are there di€erent
determinants of the frequency, intensity, and duration
of physical activity? Behavioral Medicine: An
Interdisciplinary Journal of Research and Practice 20,
84±90.
Crowne, D.P. and Marlowe, D. (1960) A new scale of
social desirability independent of psychopathology.
Journal of Consulting Psychology 24, 349±354.
Crowne, D. P. and Marlowe, D. (1964) The Approval
Motive. Wiley, New York.
Desharnais, R., Bouillon, J. and Gaston, G. (1986) Self-
ecacy and outcome expectations as determinants of
exercise adherence. Psychological Reports 59, 1155±
1159.
Diekho€, G. (1992) Statistics For the Social and
Behavioral Sciences: Univariate, Bivariate, Multivariate.
Wm. C. Brown Publishers, Dubuque, IA.
Dishman, R.K., Sallis, J.F. and Orenstein, D.R. (1985)
The determinants of physical activity and exercise.
Public Health Reports 100, 158±171.
Duncan, T.E. and McAuley, E. (1993) Social support and
ecacy cognitions in exercise adherence: A latent
growth curve analysis. Journal of Behavioral Medicine
16, 199±218.
Godin, G. and Shephard, R.J. (1990) Use of attitude-
behaviour models in exercise promotion. Sports
Medicine 10, 103±121.
Gorsuch, R. L. (1983) Factor Analysis. Lawrence
Erlbaum, Hillsdale, NJ.
Green, L. W. and Lewis, F. M. (1986) Measurement and
Evaluation in Health Education and Health Promotion.
May®eld Publishing Company, Palo Alto, CA.
Grembowski, D., Patrick, D., Diehr, P., Durham, M.,
Beresford, S., Kay, E. and Hecht, J. (1993) Self-ecacy
and health behavior among older adults. Journal of
Health and Social Behavior 34, 89±104.
Hofstetter, C.R., Hovell, M.F. and Sallis, J.F. (1990a)
Social learning correlates of exercise self-ecacy: Early
experiences with physical activity. Social Science and
Medicine 31, 1169±1176.
Hofstetter, C.R., Sallis, J.F. and Hovell, M.F. (1990b)
Some health dimensions of self-ecacy: Analysis of
theoretical speci®city. Social Science and Medicine 31,
1051±1056.
Hovell, M.F., Sallis, J.F., Hofstetter, C.R., Spry, V.M.,
Faucher, P. and Caspersen, C.J. (1989) Identifying cor-
relates of walking for exercise: An epidemiologic prere-
quisite for physical activity promotion. Preventive
Medicine 18, 856±866.
Marcus, B. and Owen, N. (1992) Motivational readiness,
self-ecacy and decision-making for exercise. Journal of
Applied Social Psychology 22, 3±16.
Marcus, B., Selby, V., Niaura, R. and Rossi, J. (1992)
Self-ecacy and the stages of exercise behavior change.
Research Quarterly for Exercise and Sport 63, 60±66.
McAuley, E. (1991) Ecacy, attributional, and a€ective
responses to exercise participation. Journal of Sport and
Exercise Psychology 13, 382±393.
McAuley, E. (1993) Self-ecacy and the maintenance of
exercise participation in older adults. Journal of
Behavioral Medicine 16, 103±113.
1182 John J. M. Dwyer et al.
McAuley, E. and Courneya, K. (1992) Self-ecacy re-
lationships with a€ective and exertion responses to exer-
cise. Journal of Applied Social Psychology 22, 312±326.
McAuley, E. and Jacobson, L. (1991) Self-ecacy and
exercise participation in sedentary adult females.
American Journal of Health Promotion 5, 185±191.
McAuley, E., Wraith, S. and Duncan, T. (1991) Self-e-
cacy, perceptions of stress, and intrinsic motivation for
exercise. Journal of Applied Social Psychology 21, 139±
155.
Norusis, M. J. (1990) SPSS/PC+ StatisticsTM 4.0
[Computer software]. SPSS Inc., Chicago, IL.
Poag, K. and McAuley, E. (1992) Goal setting, self-e-
cacy, and exercise behavior. Journal of Sport and
Exercise Psychology 14, 352±360.
Reynolds, K.D., Killen, J.D., Bryson, M.S., Maron, D.J.,
Taylor, C.B., Maccoby, N. and Farquhar, J.W. (1990)
Psychosocial predictors of physical activity in adoles-
cents. Preventive Medicine 19, 541±551.
Sallis, J.F. and Hovell, M.F. (1990) Determinants of exer-
cise behavior. Exercise and Sport Sciences Reviews 18,
307±330.
Sallis, J.F., Pinski, R.B., Grossman, R.M., Patterson, T.L.
Nader, P.R. (1988) The development of self-ecacy
scales for health-related diet and exercise behaviours.
Health Education Research 3, 283±292.
Sallis, J.F., Hovell, M.F., Hofstetter, C.R., Faucher, P.,
Elder, J.P., Blanchard, J., Caspersen, C.J., Powell, K.E.
Christenson, G.M. (1989) A multivariate study of deter-
minants of vigorous exercise in a community sample.
Preventive Medicine 18, 20±34.
Sallis, J.F., Hovell, M.F., Hofstetter, C.R. and
Barrington, E. (1992a) Explanation of vigorous physical
activity during two years using social learning variables.
Social Science and Medicine 34, 25±32.
Sallis, J., Simons-Morton, B., Stone, E., Corbin, C.,
Epstein, L., Faucette, N., Iannotti, R., Killen, J.,
Kesges, R., Petray, C., Rowland, T. and Taylor,
W. (1992b) Determinants of physical activity and inter-
ventions in youth. Medicine and Science in Sports and
Exercise 24, S248±S257.
Slenker, S.E., Price, J.H., Roberts, S.M. and Jurs,
S.G. (1984) Joggers versus nonexercisers: An analysis of
knowledge, attitudes and beliefs about jogging. Research
Quarterly for Exercise and Sport 55, 371±378.
Stephens, T. and Craig, C. L. (1990) The Well-Being of
Canadians: Highlights of the 1988 Campbell's Survey.
Canadian Fitness and Lifestyle Research Institute,
Ottawa.
Tabachnick, B. G. and Fidell, L. S. (1995) Using
Multivariate Statistics (3rd ed.). Harper Collins, New
York.
Wallston, K. and Wallston, B. (1981) Health locus of con-
trol scales. In Lefcourt H. (ed). Research with the Locus
of Control Construct, pp. 189±243. Academic Press,
New York.
Wankel, L. M. and Mummery, W. K. (1992) Factors re-
lated to intention to be physically active: Information
from the Campbell's survey of well-being. Paper pre-
sented at the 1992 International Conference on Physical
Activity, Fitness, and Health, Ottawa.
Warwick, D. P. and Lininger, C. A. (1975) The Sample
Survey: Theory and Practice. McGraw-Hill, New York.
APPENDIX A
Items In Self-ecacy Measure
Participants were asked to ``indicate how con®dent you
are that you can participate in vigorous physical activity''
in the instances listed below. A ®ve-point Likert scale, ran-
ging from not at all con®dent (1) to very con®dent (5) was
used.
1. At least three days per week.
2. If there is a lack of time due to school work.
3. If there is a lack of time due to part-time work.
4. If there is a lack of time due to family responsibilities.
5. If there is a lack of time due to other interests.
6. If you lack energy (too tired).
7. If you lack athletic ability.
8. If there is a lack of programs.
9. If there is a lack of facilities.
10. If you lack a partner.
11. If there is a lack of support from family.
12. If there is a lack of support from friends.
13. If participation costs money.
14. If you lack self-discipline or willpower.
15. If you are self-conscious (feeling uncomfortable).
16. If you have a long-term illness, disability, or injury.
17. If you have a fear of injury.
18. If you feel stressed.
19. If you do not feel in the mood.
20. If you feel discomfort (for example, soreness).
21. If you do not have fun.
Item 1 is derived from similar self-ecacy items used by
Sallis et al. (1988), Reynolds et al. (1990) and McAuley
(1993). Items 2±17 are revisions of similar items assessing
perceived barriers in the Canada Fitness Survey. Item 18
is derived from similar items by Sallis et al. (1988) and
Reynolds et al. (1990). Item 19 is derived from a similar
item used by Marcus et al. (1992). Item 20 is derived from
a similar item used by McAuley and Jacobson (1991).
Item 21 was developed by the authors for this study.