Journal of Physical Activity and Health, 2005, 2, 181-196
© 2005 Human Kinetics Publishers, Inc.
The Relative Contributions of
Psychological, Social, and Environmental
Variables to Explain Participation
in Walking, Moderate-, and VigorousIntensity Leisure-Time Physical Activity
Nicola W. Burton, Gavin Turrell,
Brian Oldenburg, and James F. Sallis
Introduction: This study assessed the relative contributions of psychological,
social, and environmental variables to walking, moderate- and vigorousintensity physical activity. Methods: A questionnaire was mailed to a random
sample (57% response rate). Analyses used a backwards elimination logistic
regression model, removing and replacing individual variables, and adjusting
for age, gender, household composition, and education (N = 1827). Results:
The sociodemographic and correlate variables collectively accounted for 43%
of the variation in total activity, 26% of walking, 22% of moderate-intensity
activity and 45% of vigorous-intensity activity (Nagelkerke R2). Individually,
the correlates accounted for 0.0 to 4.0% of unique variation, with habit, efficacy,
and support having higher values. Physical health, discouragement, competition, and time management contributed more to vigorous-intensity activity.
Anticipated benefits of social interactions and weight management contributed
more to moderate-intensity activity. Neighborhood aesthetics contributed more
to walking. Conclusion: Walking, moderate- and vigorous-intensity activity
might be associated with different correlates.
Key Words: behavioral research, leisure activity, physical exercise, health
promotion
Contemporary ecological models emphasize that behavior is influenced by factors across multiple domains, including psychological, social, and environmental
variables.1 Historically, physical activity research has focused on the psychological domain, assessing variables such as knowledge and attitudes.2 More recently,
researchers have become increasingly interested in the environmental domain, and
examining variables such as neighborhood characteristics and facilities.3 Studies
Burton is with the School of Human Movement Studies, University of Queensland, St. Lucia,
QLD 4072, Australia. Turrell and Oldenburg are with the School of Public Health, Queensland University
of Technology, Kelvin Grove, QLD 4059 Australia. Sallis is with the Dept of Psychology, San Diego
State University, San Diego, CA 92103.
181
182
Burton et al.
that focus on only 1 domain, however, are not optimizing our ability to assess the
maximal amount of variation. If the correlate variables are inter-related, focusing
on only 1 domain could overestimate the amount of unique variance accounted for
by that domain, and might lead to erroneous conclusions of strength of association.
More studies are needed that simultaneously include variables across psychological,
social, and environmental domains.
Studies incorporating variables across multiple domains are also needed to
facilitate the comparison of their relative importance. Given the extensive range of
psychological, social, and environmental correlates of activity,4 minimal work has
been done on empirically contrasting them using the same sample.5,6 Giles-Corti
reported that exercising was more strongly associated with psychological variables,
such as cognition, than either social or environmental variables, such as others’
companionship and activity level and the function and appeal of the environment.7
These multivariable studies could help explain conflicting results on the explanatory ability of correlates, and also guide priorities for practice or policies to target
factors that account for maximum amounts of variation in physical activity.
Many studies have conceptualized physical activity as a homogenous behavior,
making the assumption that correlates are similarly associated with the different types
of activity.8 There is some evidence, however, to suggest that walking and moderateand vigorous-intensity activity might be influenced by qualitatively different factors.9 Males and younger people are more likely to participate in vigorous-intensity
activity, while females and older persons have higher rates of walking.10 Adopting
vigorous-intensity activity has been associated with self-efficacy, while adopting
moderate-intensity activity has been associated with health knowledge.11 Sallis et
al., reported that the psychological and social variables related to vigorous-intensity
activity accounted for much less of the variation in walking activity.12 Perceptions
of neighborhood safety could be relevant for activities such as walking that are
undertaken outdoors or in the immediate neighborhood, but less relevant for other
types of physical activity.3,13 Humpel found that different environmental attributes
were associated with general neighborhood walking, walking for exercise, walking
for pleasure, and walking to get to and from places.14 Accordingly, there is a need
for correlate studies that differentiate among the types of physical activity.5,6,8
This study assessed the relative contributions of self-reported correlates to
variation in leisure time physical activity (LTPA). The study sought to redress the
limitations of previous studies by: (1) incorporating correlate variables from psychological, social, and environmental domains and empirically contrasting their relative contribution to activity using a population-based sample, and (2) differentiating
among walking, moderate- and vigorous-intensity activity, as well as total activity.
Method
The study was approved by the University Human Ethics Research Committee at
the Queensland University of Technology. Survey return was taken as informed
consent.
Participants
A random sample (N = 5000) was drawn from the Australian Commonwealth
electoral roll current as of October 1999, which has compulsory registration for
Physical Activity Correlates
183
all Australian citizens age 18 y and over. The sample was delimited to registered
adults age 18 to 65 and living in Brisbane, the capital city of Queensland.
Data Collection Instrument
The study used a structured self-administered questionnaire that is available from
the first author on request. The questionnaire was 10 pages in booklet form, included
189 items, and took approximately 45 min to complete. Items were grouped into
sections: activity-related beliefs and attitudes, previous activity experiences, social
support, self-efficacy, perceived benefits of physical activity, barriers to participation, availability of recreation facilities, perceptions of neighborhood characteristics,
physical activity participation, intentions to be physically active, self-assessed health
and well-being, and sociodemographic characteristics.
Procedure
Data were collected using a mail survey method adapted from Dillman. 15 The
questionnaires were delivered in September 2000, and accompanied by a personalized cover letter, a pre-addressed pre-paid reply envelope, and an instant lottery
ticket. One week later a postcard was mailed to the entire sample, to thank those
individuals who had returned their survey, and remind those who had not to do so.
Seven weeks after the initial mailing, a personalized reminder letter and replacement questionnaire were sent to all nonrespondents.
Measures
Sociodemographic Variables. Sociodemographic covariates included age,
gender, household composition, and education level, and were assessed using
items consistent with the national census.16
Correlates. The psychological, social, and environmental correlates (see Table 1)
were measured using a battery of scales that were previously developed by the
authors using both qualitative and quantitative research, and assessed constructs
consistent with social cognitive theory and suggested by previous research to be
associated with physical activity levels.4 Of the 28 scales that were factorially
derived, 25 have acceptable levels of internal consistency with Cronbach’s alpha
values ranging from 0.69 to 0.89. An additional index was used to assess neighborhood activity-related facilities such as a gym or swimming pool. The development
and initial psychometric evaluation of these scales has been reported elsewhere.17
With the exception of the index to assess neighborhood activity facilities that used
a (1) “yes,” (2) “no,” (3) “don’t know” response format, items utilized a 5-point
Likert scale response format. Items assessing activity experiences, activity-related
cognitions, anticipated benefits, perceived barriers, and neighborhood characteristics
had response options ranging from (1) “strongly disagree,” to (5) “strongly agree.”
Items assessing social support asked how often significant others had performed
specific actions and used a response format ranging from (1) “never” to (5) “very
often.” Efficacy items assessing confidence to be active in specific situations used
a response format ranging from (1) “I know I could not” to (5) “I know I could.”
Scale scores were obtained by summing across the relevant items, and then the
scales were collated into eight thematic domains as presented in Table 1.
184
Table 1
Burton et al.
Correlate Variables (N = 29) and Thematic Domains (N = 8)
Used in the Study
Correlate domains
and scales
Description
Activity history
Mastery
Habit
Exposure
Previous experiences of activity participation
Perceptions of achievement with physical activity
Frequency of past participation in activity
Previous opportunities to be involved in physical activity
Perceived health
Physical health
Psychological health
Subjective judgment of health
General health and physical limitations
History of stress and depression, overall life satisfaction
Cognition
Activity schemata
Self-schemata
Need
Demand
Knowledge
Thoughts about physical activity
Beliefs on who should be physically active, e.g., obese, athletes
Image of self in relation to physical activity
Judgment of personal need for physical activity
Perception of physical effort and commitment to be active
Understanding recommendations for participation, e.g., amount
Efficacy
Confidence to include physical activity in life
Anticipated benefits
Psychological
Health
Competition
Improved appearance
Social
Weight management
Positive anticipated outcomes of regular activity participation
Improved cognitive functioning and mood
Illness avoidance, health maintenance, longevity
Enjoyment of competition against self/others
Enhanced looks and body image
Enhanced opportunities to meet and spend time with others
Weight loss and maintenance
Perceived barriers
Poor access
Low skill
Low personal
functioning
Time management
Disinterest
Family obligations
Factors perceived as limiting participation in physical activity
Difficulties with utilization of resources, e.g., transport, cost
Inability to match others, inexperience, self consciousness
Poor health, injury, disability, overweight, mood disturbance
Social support
Encouragement
Discouragement
Factors relating to interpersonal relations with others
Verbal and functional assistance from others to be active
Criticism and lack of assistance to be active
Neighborhood
environment
Physical features
Aesthetic features
Traffic
Facilities
Perceived characteristics of the local neighborhood
Difficulty with routinization, irregular time commitments
Boredom with and dislike of physical activity
Child care needs, competing family activities
Footpaths, public transport, services, streetlights
Perceived safety, ambience, cleanliness, friendliness
Busyness of streets and extent of traffic flow
Facilities for activity participation, e.g., gyms, pools, paths
Physical Activity Correlates
185
Physical Activity. Physical activity items were consistent with those used in
Australian national surveys,18 and have been recommended for use in population-based monitoring of activity.19 Items assessed the frequency and total time
of participation in walking (for sport, recreation, or to get to and from places),
moderate-intensity activity (causing a moderate increase in heart rate or breathing,
such as slow swimming, golf, slow cycling) and vigorous-intensity activity (resulting
in a large increase in heart rate or breathing, such as athletics, aerobics, football)
during the previous week. To reduce possible measurement error resulting from overreporting time, data were truncated via methods used in other Australian research.18
The maximum “allowable” time doing any one of the three types of activity was
14 h/wk; any greater time was recoded to 14 h. The maximum “allowable” time
across the 3 activities was 28 h/wk, any greater time was recoded to 28 h.
For each type of activity, the total time (in minutes) was multiplied by an
intensity value of 3.0 METs for walking, 4.0 METs for moderate-intensity activity, and 7.5 METs for vigorous-intensity activity, where a MET is a measure of
metabolic equivalents. These intensity values were derived from published MET
values for actual activities reported by men and women in response to questions
about moderate- and vigorous-intensity activities and have been used in previous
Australian research.20 Walking, moderate- and vigorous-intensity energy expenditure were then categorized into dichotomous variables using zero as the cut
off. Respondents were categorized as participating in “none” or “some” each of
walking, moderate-intensity, and vigorous-intensity activity during the preceding
week. To measure total activity participation, the time and MET product scores for
walking, moderate- and vigorous-intensity were summed to provide a total energy
expenditure score for the preceding week. As the distribution of this variable was
not normal, and unable to be improved by mathematical transformation, it was
converted to a dichotomous variable based on the criterion of ≥ 600 METs/min as
adequate for health, which reflects 30 min of moderate activity (4 METs) on 5 d
of the week, and is consistent with Australian guidelines.21
Analyses
After excluding nonusable returns (e.g., return-to-sender, left address etc.) a final
response rate of 56.7% (2532/4465) was achieved, with 2537 surveys returned
with data, 525 marked as returned mail, 104 returned blank, and 26 classified as
ineligible (e.g., persons with cognitive disabilities, non-English speaking persons).
Five cases were excluded from the analyses as accompanying letters indicated that
the questionnaires had been completed by someone other than the intended recipient. Among the returned surveys, 15% of respondents had incomplete correlate
data and 28% had incomplete physical activity data. Detailed analyses of the sociodemographic pattering of item nonresponse for the physical activity items has been
reported elsewhere.22 Less than 4% of physical activity data required truncation,
with 2.8% truncated on one activity item, 0.8% truncated for two activity items,
0.1% truncated across all three activity items, and 0.5% subject data truncated for
the total activity time. Data were complete for variables of age and gender, and
1.6% of respondents had incomplete data on household composition and 2.5%
provided incomplete education data.
186
Burton et al.
Analyses were conducted using SPSS version 10 (SPSS, Inc., Chicago, IL).
Descriptive statistics were obtained to profile the sociodemographic characteristics of respondents, and where possible, compared with data from the national
census.16 The interrelationships among the 29 correlate variables were examined
using Pearson correlation coefficients. The independent contribution of the variables to unique variation in LTPA was assessed in a backwards elimination logistic
regression model, adjusting for age, gender, household composition, and education. Regression analyses were conducted separately for walking, moderate- and
vigorous-intensity, and total activity. Each of the eight correlate domains was
removed from the full regression model, and the change in the variation accounted
for was examined using the Nagelkerke R2, which is analogous to the R2 in ordinary least squares regression in that it provides an approximation of the explained
variance for a logistic model.23 Values of the Nagelkerke R2 range from 0 to 1,
where 0 indicates that the independent variables have no usefulness in predicting
variation in physical activity. The domain variable was then placed back into the
full regression model and another domain removed to examine the change in the
variation accounted for. Analyses were repeated to individually remove and replace
each of the 29 correlate variables for walking, moderate- and vigorous-intensity,
and total activity.
Results
Sample Profile
Table 2 presents the sociodemographic profile of the survey respondents, the analytic
sample, and available comparative data from the Australian census.16 The analytic
sample comprised 1827 persons between 18 and 65 years of age, who were registered as living in Brisbane, Australia in 2000, and who provided data on all the
relevant physical activity, covariate, and correlate variables. The analytic sample
was very similar to the respondent sample, with the exception of having a lower
proportion of people engaged in walking, moderate- and vigorous intensity activity,
although the proportion of people sufficiently active for health was comparable
with the respondent sample.
Inter-Relationships Among Correlates
With the large sample size, the majority of correlations were significant at P =
0.001. Eleven pairs of variables had correlations greater than 0.50 (absolute value)
indicating at least 25% shared variance. The strongest correlation was between selfschemata and habit (0.73). Other correlations above 0.5 were between disinterest
and self-schemata (0.60), efficacy and habit (0.59), barriers of low personal functioning and low skill (0.58), the barrier of low personal functioning and physical
health (–0.56), disinterest and habit (–0.55), efficacy and self-schemata (0.54),
competition and self-schemata (0.52), poor time management and efficacy (0.52),
and benefits of improved appearance and efficacy (0.52). A table illustrating the
correlations among the 29 correlate variables is available from the first author on
request.
Physical Activity Correlates
187
Table 2 Sociodemographic Profile of Survey Respondents, Analytic Sample, and
Selective Comparison Data from the National Censusa
Respondent
sample
(n = 2532)
%
Analytic
sample
(n = 1827)b
%
2001 Censusa
(n =
11,685,192)
%
Gender
Female
Male
55
45
54
46
50
50
Age category
18-24 y
25-34 y
35-44 y
45-54 y
55-64 y
11
21
27
29
12
11
20
27
29
13
15
23
25
22
15
Household composition
Couple with children
Couple with no children
Single parent with children
Single, living with others
Single, living alone
45
27
4
16
8
45
27
3
17
8
–
–
–
–
–
Educational qualifications
School only
Trade
Diploma
University
38
24
13
24
38
26
13
23
56
20
8
17
Annual personal income before tax ($ Aus)
< $20,799c
$20,800 – $36,399c
$36,400 – $51,999d
> $52,000d
13
17
17
53
13
17
18
52
Self-assessed health
Excellent
Very good
Good
Fair
Poor
9
29
42
17
3
10
29
42
16
3
–
–
–
–
–
Smoking status
Never smoked
Ex-smoker
Current smoker
55
27
18
55
27
18
–
–
(continued)
188
Table 2
Burton et al.
(continued)
Respondent
sample
(n = 2532)
%
Analytic
sample
(n = 1827)b
%
2001 Censusa
(n =
11,685,192)
%
Body-mass index (BMI)
Normal range (20 ≤ BMI ≤ 25)
Underweight (BMI < 20)
Overweight (25 < BMI ≤ 30)
Obese (BMI > 30)
44
11
31
14
44
11
31
13
–
–
–
–
Physical activity participation (%)
Some walking activity
Some moderate-intensity activity
Some vigorous-intensity activity
Total activity sufficient for health
77
33
46
55
73
29
40
55
–
–
–
57e
Note. Because of rounding, proportions might not add to 100%.
a
Australian Bureau of Statistics (2001).
b
Derived by deleting cases with missing data on physical activity and correlate variables
included in the multivariate analyses.
c
Income markedly below the Australian average in 2000 (Australian Bureau of Statistics,
2001).
d
Income above the Australian average in 2000 (Australian Bureau of Statistics, 2001)
e
Armstrong et al., (2000).
Independent Contribution of Correlate Domains
Table 3 presents the amount of unique variation accounted for by each of the correlate domains. The full model with covariates and the correlate domains accounted
for 45% of the variation in vigorous-intensity activity, 43% of total activity, 26%
of walking, and 22% of moderate-intensity activity. Without the covariates, the
correlate domains collectively accounted for 40% of the variation in total activity,
35% of vigorous-intensity activity, 24% of walking, and 16% of the variation in
moderate-intensity activity.
Individually, the correlate domains accounted for 0.1 to 4.1% of unique variation in each of the different types of activity, with activity history, efficacy, and
anticipated benefits having relatively higher values across the activity types. The
contributions of health, perceived barriers, and social support were relatively higher
for vigorous-intensity activity than the other activity types. Anticipated benefits
contributed more to vigorous-intensity and total activity. Cognitions contributed
more to walking and total activity. Environmental variables contributed least to
vigorous intensity activity.
Relative Contribution of Correlate Variables
Table 4 presents the proportion of unique variation accounted for by each of the
correlate variables, for walking, moderate- and vigorous-intensity, and total activity.
Independently, the correlate variables accounted for 0.0 to 4.0% of unique variation
Physical Activity Correlates
189
Table 3 Proportion of Unique Variation Accounted for in Walking,
Moderate-Intensity, Vigorous-Intensity Activity, and Total Physical Activity
by Each of the Correlate Domains (Nagelkerke R2)
Correlate group
Walking
activity
(%)
Moderateintensity
activity
(%)
Vigorousintensity
activity
(%)
Total activity
sufficient for
health
(%)
Full modela
All correlate scalesb
Covariatesc
Activity history
Health
Cognitions
Efficacy
Benefits
Barriers
Social support
Environmental
26.1
23.6
1.6
1.9
0.3
1.1
2.9
0.4
0.2
1.0
0.6
22.0
16.0
3.5
1.2
0.1
0.3
1.6
1.3
0.6
1.2
1.1
44.6
35.2
2.9
3.7
0.7
0.5
1.5
2.0
1.1
1.5
0.4
43.4
40.1
0.5
4.1
0.2
0.8
2.7
2.0
0.6
1.1
1.2
Model includes all correlates and covariates of age, gender, education, and household
composition.
b
Model excludes covariates of age, gender, education, and household composition.
c
Covariates of age, gender, education, and household composition
a
within the different types of activity. Habit, self-efficacy, and social encouragement contributed relatively greater amounts of unique variation within each type
of activity. Discouragement contributed to more unique variation in both total and
vigorous-intensity activity than to other activity types, while anticipated social and
weight management benefits contributed more to moderate-intensity activity. Physical health, anticipated competitiveness, and time management barriers contributed
to more unique variation in vigorous-intensity activity than to the other types of
activity. Neighborhood aesthetics contributed more to walking, and the barrier of
family obligations contributed more to total and moderate-intensity activity.
Discussion
The combination of sociodemographic covariates and psychological, social, and
environmental correlates accounted for 43% of variation in total activity, which is
favorably consistent with other studies.8 The combined correlate variables accounted
for different amounts of variation in walking, moderate- and vigorous-intensity
activity, and total activity. A greater amount of variation was accounted for in
vigorous-intensity (45%) and total activity (43%), compared with walking (26%)
and moderate-intensity activity (22%). This is consistent with previous research
reporting correlate variables accounting for much less of the variation in walking
compared with vigorous-intensity activity.12 Given the similarity in the variation
accounted for, studies using total activity as an outcome might therefore be reflecting vigorous-intensity activity, rather than walking and moderate-intensity activity.
190
Burton et al.
Table 4 Proportion of Unique Variation Accounted for in Walking,
Moderate-Intensity, Vigorous-Intensity Activity, and Total Physical Activity
by Each of the Correlate Variables (Nagelkerke R2)
Walking
activity (%)
Moderate–
intensity
activity (%)
Vigorous–
intensity
activity (%)
Total activity
sufficient for
health (%)
26.1
22.0
44.6
43.4
All correlate scales
23.6
16.0
35.2
40.1
Activity history
Exposure
Mastery
Habit
0.1
–0.1
1.9
0.1
0.1
1.0
0.0
0.3
3.5
0.0
0.2
4.0
Perceived health
Physical health
Psychological health
–0.1
0.3
0.0
0.1
0.8
0.0
0.3
0.1
Cognition
Self–schemata
Activity schemata
Demand
Need
Knowledge
0.0
0.3
0.2
0.1
–0.1
0.0
–0.1
0.4
–0.1
0.1
–0.1
0.3
0.1
0.4
0.0
0.2
0.0
0.4
0.2
0.0
2.9
1.6
1.5
2.7
Anticipated benefits
Psychological
Positive health
Challenge
Improved appearance
Social
Weight management
–0.1
–0.1
0.1
0.2
0.1
–0.1
–0.1
–0.1
0.3
0.6
0.9
0.8
0.1
0.1
0.7
0.5
0.1
0.2
0.3
–0.1
0.5
0.4
0.0
0.1
Perceived barriers
Poor access
Low skill
Low personal functioning
Poor time management
Disinterest
Family obligations
–0.2
0.4
–0.2
0.1
0.0
0.0
0.0
0.1
0.3
0.2
0.3
0.5
0.1
0.1
–0.1
0.7
0.0
–0.1
0.4
0.3
–0.1
0.4
0.0
0.4
0.8
0.2
0.9
0.4
1.0
0.9
1.0
0.7
Correlate scale
Full modela
b
Efficacy
Social support
Encouragement
Discouragement
(continued)
Physical Activity Correlates
Table 4
191
(continued)
Correlate scale
Neighborhood environment
Physical features
Aesthetic features
Traffic
Facilities
Walking
activity (%)
Moderate–
intensity
activity (%)
Vigorous–
intensity
activity (%)
Total activity
sufficient for
health (%)
0.4
0.4
0.0
–0.3
0.3
0.0
0.2
0.1
0.4
0.1
0.0
0.3
0.1
0.1
0.1
0.1
Model includes all correlate scales and covariates of age, gender, household composition, and
education.
b
Model excludes covariates of age, gender, household composition, and education.
a
Individually, the correlates accounted for very small amounts of unique variation
(0 to 4%). One explanation for this is that the correlates were strongly associated
with each other and contributed shared variation. Analyses of the inter-relationships
however, indicated most had less than 25% shared variance.
The results of this study contribute to other evidence suggesting that walking
and moderate- and vigorous-intensity activity could be influenced by qualitatively
different factors.9 Although activity history, efficacy, and social support contributed relatively higher amounts of unique variation, the relative contribution of
these and the other correlate variables differed across the types of activity. For
vigorous-intensity activity, physical health, discouragement, anticipated competitiveness, and time management barriers also contributed relatively high amounts
of unique variation. For moderate-intensity activity, anticipated social interaction
and weight management benefits contributed relatively high amounts of unique
variation. Neighborhood aesthetics contributed to more unique variation in walking
than to other types of activity. Consistent with Giles-Corti, environmental variables
accounted for relatively less variation in activity compared with psychological and
social variables.7
These results could be interpreted in several ways. First, it might be that
walking, moderate- and vigorous-intensity activity are in fact influenced by different factors. It seems plausible that as the different types of physical activity
have varying characteristics, demands, and contexts, they are therefore subject to
different influences. Research focusing on walking as a means of active commuting for example, has highlighted the role of neighborhood variables previously
unrecognized in the vigorous-intensity physical activity literature, such as urban
design, street connectivity, population density, and mixed development.24
An alternate explanation could be associated with how the respondents conceptualized physical activity when responding to correlate assessment items. The
questionnaire instructions provided a broad definition of physical activity as “including things like exercise, competitive or friendly training, swimming, walking, and
cycling.” Respondents, however, might have adopted a more narrow and traditional
conceptualization of physical activity that approximated vigorous-intensity exercise.
The similarity between the amount of variation accounted for in vigorous-intensity
activity and total activity would support this interpretation.
192
Burton et al.
Dishman proposed that differences in the amount of variation accounted for
between vigorous-intensity and walking activity could be explained by measurement error.9 Because of its more incidental nature, self-reported assessment of
walking and moderate-intensity activity has lower levels of reliability and validity
than vigorous-intensity activity.25,26 Common types of moderate-intensity activities include walking at work and for exercise, household and garden chores, and
conditioning activities.25 In the present study, however, physical activity associated
with household and garden chores and occupational settings was not included in
calculations of energy expenditure as there is limited research on the self-reported
validity of these activities.18 By excluding these types of physical activity, our
measurement of moderate-intensity activity might have been compromised and
contributed to the comparatively lower amount of variation accounted for.
These interpretations could also explain the variation in physical activity not
accounted for in the present study; in particular for walking and moderate-intensity
physical activity. Both respondent interpretations of “physical activity,” as well
as measurement error for the correlate and physical activity variables might have
contributed to unexplained variation. In addition, the range of correlate variables
assessed in this study was constrained to those assessed by a battery of scales
developed by the authors.17 These scales were designed to be relevant to physical
activity other than vigorous-intensity exercise but could not have been exhaustive
in terms of the range of correlate variables measured.
Methodological Limitations
When interpreting the findings of this study, consideration needs to be given to a
number of methodological issues. Data were based on self-report, which is prone
to bias and measurement error27 that can result in an underestimation in the level of
physical activity both among sedentary populations and for walking.28 Self-report
physical activity data are, however, pragmatic for large samples and considered
appropriate for population monitoring,28 and are reliably associated with morbidity
and mortality outcomes.29 While there are objective data collection methods for
some correlates such as environmental variables, we are reliant on self-report data
for psychological correlates, such as self-efficacy. The generalizability of these
results might be limited by errors associated with the sampling frame, mail survey,
nonresponse, and exclusion of cases with missing data on the relevant variables.
The sample was drawn from the electoral roll, which can underrepresent transient,
migrant, younger, and socially disadvantaged individuals.30 Mail methodologies
could yield less complete data than interview studies,31 with both survey and item
nonresponse more prevalent among individuals of low socioeconomic position.32 A
comparison of our sample profile with census data16 suggested an overrepresentation
of females and university-educated individuals, and an underrepresentation of people
18 to 34 years of age. Results might therefore be less applicable to young adults
and individuals of low socioeconomic position, and a detailed examination of these
groups could yield different relative contributions of the correlate variables.
Implications for Research, Practice, and Policy
Understanding the interplay of psychological, social, and environmental influences
on health and health-related behaviors such as physical activity, represents a key
Physical Activity Correlates
193
research need to direct more policy attention to health promotion.33 By understanding the relative contribution of the different correlates, interventions can be
tailored to focus on factors that account for maximum amounts of variation, thereby
optimizing their impact and cost-effectiveness. Interventions targeting a particular
correlate might be more relevant to one specific type of activity. The results of
this study suggest that correlates associated with a total level of physical activity
sufficient for health are habit, self-efficacy, and social encouragement. Change
programs to promote physical activity generally could therefore involve strategies
that aim to regularize participation, enhance confidence (which might be related
to other variables such as skill or time management), and enhance social support
from significant others.
Promotion of vigorous-intensity activity should acknowledge additional correlates of physical health, social discouragement, anticipated benefits of competition, and time management barriers. Change programs could, therefore, be more
likely to be adopted by people who report good health, have few social barriers
to participation, enjoy a level of competitive physical activity, and have flexible
time organization. Health constraints, low social support, a dislike of competitive
activity, and competing time demands might be barriers specifically relevant to
participation in vigorous-intensity activity. Strategies for moderate-intensity activity and walking could promote anticipated benefits of weight management and
social opportunities, and include urban planning policy that promotes “aesthetic”
neighborhoods characterized by pleasant walks, personal safety, well-maintained
paths and lighting, and opportunities for social interactions. The relatively low
contribution of environmental variables compared with psychological and social
variables also suggests that intra-personal correlates should not be overlooked by
“upstream” change programs targeting correlates more distal to individuals such
as the physical environment. Instead, the different types of psychological, social,
and environmental correlates should be seen as a complex interplay of multilevel
factors that needs to be better understood to promote physical activity.
To improve conceptual models, future research should identify the correlates of specific types of activity rather than global activity levels. Multilevel
studies that take into account area-level variables and environmental constraints
and facilitators, or use objective measures of environmental variables could also
enhance predictive ability. More in-depth qualitative studies designed to specifically
identify the barriers and facilitators of physical activity might be useful to assess
those individuals typically excluded in quantitative research, such as individuals
of low socioeconomic position. Traditionally, much physical activity research has
focused on vigorous-intensity activity, and the results of this study suggest that
this evidence base could not be directly generalizable to moderate-intensity and
walking activity. Accordingly, more research is needed to understand the correlates
specifically associated with walking and moderate-intensity activity. It might be
worthwhile in some contexts to conceptualize walking and moderate-intensity
activity as similar to each other and dissimilar from vigorous-intensity activity.
Data for walking and moderate-intensity activity could therefore be combined to
consider physical activity other than vigorous-intensity exercise.
More studies are also needed to examine the reliability and validity of selfreported walking and moderate-intensity activity so that the contribution of measurement error to differences in the amount of variation accounted for by correlate
194
Burton et al.
models is minimized. In the current study, participants were asked to respond to
a series of correlate items about “physical activity,” which was broadly defined to
include walking, moderate-, and vigorous-intensity activity. An alternate approach
would be to obtain responses to correlate items for each type of activity separately,
and then compare the relative importance of the correlates. This method, however,
would markedly increase subject burden.
Conclusions
Comprehensive models that integrate psychological, social, and environmental
correlate variables are more likely to reflect the many factors associated with physical activity and the varied opportunities for intervention. To better understand the
explanatory power and relative importance of these factors, there is a need to focus
separately on the specific types of physical activity, in particular differentiating
between vigorous-intensity and moderate-intensity or walking activity. Policies
and interventions aimed at increasing physical activity might need to address different variables to acknowledge the relative contributions made to vigorous- and
moderate-intensity activity, and walking.
Acknowledgments
This research was supported by funding from the Queensland University of Technology and the National Heart Foundation of Australia (G98B0087), and conducted at the
Queensland University of Technology. We gratefully acknowledge the responses from study
participants, and biostatistical consultation from Dr. Diana Battistutta. Dr. Turrell is a Senior
Research Fellow supported by a National Health and Medical Research Council/National
Heart Foundation Career Development Award (CR 01B 0502).
References
1. Sallis J, Owen N. Ecological models of health behavior. In: Glanz K, Rimer BK, Lewis
FM, eds. Health Behavior and Education. Theory, Research, and Practice. 3rd ed.
San Francisco: Jossey-Bass; 2002.
2. King AC, Stokols D, Talen E, Brassington GS, Killingsworth R. Theoretical approaches
to the promotion of physical activity: forging a transdisciplinary paradigm. Am J Prev
Med. 2002;23(2S):15-25.
3. Humpel N, Owen N, Leslie E. Environmental factors associated with adults’ participation in physical activity: a review. Am J Prev Med. 2002;22:188-199.
4. Trost SG, Owen N, Bauman AE, Sallis JF, Brown WJ. Correlates of adult’s participation
in physical activity: review and update. Med Sci Sport Exerc. 2002;34:1996-2001.
5. Dishman RK, ed. Advances in exercise adherence. Champaign IL: Human Kinetics
Publishers; 1994.
6. Bouchard C, Shephard RJ, Stephens T. Physical activity, fitness, and health: status and
determinants. In: Bouchard C, Shephard RJ, Stephens T, eds. Physical activity, fitness
and health. Consensus statement. Champaign, IL: Human Kinetics Publishers; 1993:
33-39.
7. Giles-Corti B, Donovan RJ. The relative influence of individual, social, and physical
environment determinants of physical activity. Soc Sci Med. 2002;54:1793-1812.
Physical Activity Correlates
195
8. Baranowski T, Anderson C, Carmack C. Mediating variable framework in physical
activity interventions. How are we doing? How might we do better? Am J Prev Med.
1998;15:266-297.
9. Dishman RK. Exercise adherence. In: Singer RN, Murphy M, Tennant LK, eds.
Handbook of research on sport psychology. New York: MacMillan Publishing; 1993:
779-798.
10. Pratt M, Macera C, Blanton C. Levels of physical activity and inactivity in children
and adults in the United States: current evidence and research issues. Med Sci Sports
Exerc. 1999;31(11 suppl S):S526-S533.
11. Sallis J, Haskell WL, Fortmann SP, Vranizan KM, Taylor CB, Solomon DS. Predictors
of adoption and maintenance of physical activity in a community sample. Prev Med.
1986;15:331-341.
12. Sallis J, Hovell MF, Hofstetter CR, Faucher P, Elder JP, Blanchard J, et al. A multivariate study of determinants of vigorous exercise in a community sample. Prev Med.
1989;18:20-34.
13. Humpel N, Owen N, Iverson D, Leslie E, Bauman A. Perceived environmental
attributes, residential location and walking for particular purposes. Am J Prev Med.
2004;26:119-125.
14. Ross CE. Walking, exercising, and smoking: does neighborhood matter? Soc Sci Med.
2000;51:265-274.
15. Dillman DA. Mail and internet surveys: the tailored design method. Chichester,
England: Wiley; 2000.
16. Australian Bureau of Statistics. Census 2001. Canberra: Australian Government
Publishing Service; 2001.
17. Burton NW, Oldenburg B, Turrell G, Sallis JF. The SCAPES questionnaire: psychometric evaluation of scales to measure correlates of leisure-time physical activity. Int
J Behav Med. 2004;11(suppl):140.
18. Armstrong T, Bauman A, Davies J. Physical activity patterns of Australian adults.
Results of the 1999 National Physical Activity Survey. Canberra: Australian Institute
of Health and Welfare, 2000.
19. Brown WJ, Bauman A, Chey T, Trost SG, Mummery WK. Comparison of surveys
used to measure physical activity. Aust NZ J Pub Health. 2004;28:128-34.
20. Brown WJ, Bauman A. Comparison of estimates of population levels of physical
activity using two measures. Aust NZ J Pub Health. 2000;24:520-525.
21. Commonwealth Dept of Health and Aged Care. National physical activity guidelines
for Australians. Canberra: Department of Health and Aged Care; 1999.
22. Burton NW, Turrell G, Oldenburg B. Item non-response in a population-based study
of physical activity. J Phys Act Health. 2004;1:344-362.
23. Nagelkerke NJD. A note on a general definition of the coefficient of determination.
Biometrika, 1991;78:691-692.
24. Saelens BE, Sallis J, Frank LD. Environmental correlates of walking and cycling:
findings from the transportation, urban design, and planning literatures. Ann Behav
Med. 2003;25:80-91.
25. Ainsworth BE, Bassett DR, Strath SJ, Swartz AM, O’Brien WL, Thompson RW, et al.
Comparison of three methods for measuring the time spent in physical activity. Med
Sci Sports Exerc. 2000;32:S457-464.
26. Richardson MT, Leon AS, Jacobs DR, Ainsworth BE, Serfass R. Comprehensive
evaluation of the Minnesota Leisure Time Physical Activity Questionnaire. J Clin
Epidemiol. 1994;47:271-281.
196
Burton et al.
27. Newell SA, Girgis A, Sanson-Fisher RW, Savolainen NJ. The accuracy of self-reported
health behaviors relating to cancer and cardiovascular disease in the general population. A critical review. Am J Prev Med. 1999;17:211-229.
28. Tudor-Locke CE, Myers AM. Challenges and opportunities for measuring physical
activity in sedentary adults. Soc Sci Med. 2001;31:91-100.
29. Bauman AE. Updating the evidence that physical activity is good for health: an epidemiological review 2000-2003. J Sci Med Sport. 2004;7:6-19.
30. Turrell G, Najman JM. Collecting food-related data from low socioeconomic groups:
how adequate are our current research designs? Aust J Public Health. 1995;19:410416.
31. Van Campen C, Sixma M, Kerssens JJ, Peters L. Comparisons of the costs and quality
of patient data collection by mail versus telephone versus in-person interviews. Eur
J Public Health 1998;8:66-70.
32. Van Loon AJM, Tijhuis M, Picavet HSJ, Surtees PG, Ormel J. Survey non-response
in the Netherlands: effects on prevalence estimates and associations. Ann Epidemiol.
2003;13:105-110.
33. McGinnis JM, Williams-Russo G, Knickman JR. The case for more active policy
attention to health promotion. Health Affairs. 2002;21:78-93.