Professional Documents
Culture Documents
The Relative Influence of Individual, Social and Physical Environment Determinants of Physical Activity
The Relative Influence of Individual, Social and Physical Environment Determinants of Physical Activity
Abstract
Environmental determinants of health are receiving growing attention in the literature, although there is little
empirical research in this area. The Study on Environmental and Individual Determinants of Physical Activity (known
as the SEID project) was a social ecological project that examined the relative influence of individual, social
environmental and physical environmental determinants of recreational physical activity. It involved a community
survey of 1803 healthy workers and home-makers aged 18–59 years living in a 408 km2 area of metropolitan Perth,
Western Australia. Physical environmental determinants were mainly conceptualised as spatial access to popular
recreational facilities. Overall, 59% of respondents exercised as recommended. Recreational facilities located near home
were used by more respondents than facilities located elsewhere. The most frequently used facilities were informal: the
streets (45.6%); public open space (28.8%) and the beach (22.7%). The physical environment’s directs the influence on
exercising as recommended was found to be secondary to individual and social environmental determinants.
Nevertheless, accessible facilities determined whether or not they were used and in this way, support and enhance the
achievement of recommended levels of physical activity behaviour by providing opportunities. The results suggest that
access to a supportive physical environment is necessary, but may be insufficient to increase recommended levels of
physical activity in the community. Complementary strategies are required that aim to influence individual and social
environmental factors. Given the popularity of walking in the community, it is recommended that greater emphasis be
placed on creating streetscapes that enhance walking for recreation and transport. r 2002 Elsevier Science Ltd. All
rights reserved.
Keywords: Physical activity; Spatial access; Environment; Social policy; Health promotion; Western Australia
0277-9536/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 7 7 - 9 5 3 6 ( 0 1 ) 0 0 1 5 0 - 2
1794 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
community physical activity levels (Dishman, 1988; control. This requires that individuals have all the
King et al., 1995; Sallis, Bauman, & Pratt, 1998). The necessary resources, skills and abilities to perform the
physical environment provides cues and opportunities behaviour at will, and its execution requires no co-
for physical activity. Its influence is passive through the operation from anyone else nor is it subject to any
design of the urban environment, domestic appliances environmental barriers (Ajzen, 1985; Sheppard et al.,
and buildings which encourage or discourage incidental 1988; Bagozzi & Warshaw, 1990; Ajzen, 1991).
physical activity (King et al., 1995; Sallis & Owen, 1996; During the 1980s, behavioural scientists questioned
Sallis et al., 1998). Its influence is also active by the assumption that many behaviours are under an
providing physical activity opportunities that are individual’s ‘complete volitional control’ (Ajzen, 1985;
accessible, convenient, safe and appealing (King et al., Bagozzi & Warshaw, 1990) and suggested that, where
1995; Sallis & Owen, 1996). Compared with individual there is incomplete control, individuals form intentions
determinants of physical activity, relatively few studies to perform behavioural goals. Thus, many intentions are
have examined the influence of the physical environment best considered goals, the attainment of which are
on physical activity levels (see Brownell, Stunkard, & subject to some degree of uncertainty (Ajzen & Madden,
Albaum, 1980; Hofstetter, Hovell, & Sallis, 1990; Sallis 1986).
et al., 1990; Linenger, Chesson, & Nice, 1991; McKen- In the early 1990s, the TPB (Ajzen, 1991) and the TT
zie, Sallis, Nader, Broyles, & Nelson, 1992; Macintyre, (Bagozzi & Warshaw, 1990) were developed in response
Maciver, & Sooman, 1993; Hahn & Craythorn, 1994; to these concerns. Ajzen’s TPB extended the original
Blamey, Mutrie, & Aitchison, 1995; Bauman & Smith, theory by adding perceived behavioural control: one’s
1999). belief about how easy or difficult performance of a
The results reported in this paper are based on the behaviour is likely to be. Similar to Bandura’s concept
Study on Environmental and Individual Determinants of self-efficacy (1997), the addition of perceived beha-
of Physical Activity (known as SEID), a social vioural control significantly improved the predictive
ecological project that examined the relative influence ability of the original theory (Ajzen, 1991).
of social environmental, physical environmental and According to Bagozzi and Warshaw (1990), for most
individual factors on recreational physical activity. The behaviours, the key assumptions of the TRA cannot be
overall aims of the research were to measure the met; that is the performance of a behaviour is preceded
individual, social and physical environmental factors by a deliberate process culminating in a conscious
influencing participation in planned recreational physi- decision to act and that, once decided, there are no
cal activity and to examine the relative influence of and barriers (e.g., cost, inconvenience) preventing action.
interaction between, environmental and individual They developed the TT to better understand the steps
factors that influence exercising as recommended. This between intention formation and actual performance of
paper describes the overall results, as well as the the target behaviour. Thus, the individual factors
influence of spatial access on the use of recreational examined in this study were based on Bagozzi and
facilities. Warshaw’s (1990) TT, Ajzen’s (1985) perceived beha-
vioural control and Bandura’s (1997) conceptualisation
Individual factors of behavioural skills.
The individual factors examined in this study were Physical environmental factors
based on the Theory of Planned Behaviour (TPB)
(Ajzen, 1991) and the Theory of Trying (TT) (Bagozzi The limited number of published studies on the
& Warshaw, 1990). These theories are derived from influence of the physical environment on physical
Fishbein and Azjen’s theory of reasoned action (TRA), activity suggest that spatial access to recreational
an ‘expectancy-value’ model (Ajzen & Fishbein, 1980). facilities may influence physical activity participation.
According to expectancy-value models, individuals are Spatial access was a significant predictor of participation
more motivated to perform behaviours they believe will in an executive health program (Teraslinna, Partanen,
result in highly valued outcomes, and, conversely, are Koskela, & Oja, 1969), and had a measurable, albeit
less motivated when they do not believe that valued modest, impact on participation in vigorous exercise in
outcomes will follow when the expected outcomes are community settings (Sallis et al., 1990) and on fitness
not valued (Ajzen & Fishbein, 1980). levels in a naval setting (Linenger et al., 1991). Bauman
While the TRA has had widespread use in physical et al. (1999) found that living in close proximity to the
activity research (for example see Godin, 1993; Blue, coast was associated with exercising as recommended.
1995), its predictive capacity has been limited by the Convenience appears to be important. Andrew et al.
theory’s underlying assumption that performing a (1981) found that perceptions of the convenience of a
behaviour involves a rational decision-making process rehabilitation program and difficulties encountered
and its performance is under the individual’s volitional with parking influenced participation by post-coronary
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1795
patients. Indeed, Hofstetter et al. (1990) have reported of the park, the size of the water body, the availability of
that positive perceptions about the convenience of boat ramps, opportunities for swimming, and the
facilities and neighbourhood safety increased exercise number of campsites (Kim & Fesenmaier, 1990). In a
self-efficacy, a factor known to influence intention to be study of urban swimming pools, attractiveness included
physically active. the size of the pool, facilities offered and the level of
While it has been suggested that the finding that close crowding anticipated by individuals (Goodchild &
proximity is associated with the use of facilities could ‘be Booth, 1980).
easily understood by common sense’ (Teraslinna et al., Although the recreational studies reviewed were based
1969), conclusive evidence that simply having access to on gravity models, none sought to examine individual
facilities increases regular physical activity remains spatial access. Rather, the objectives were to make
elusive. Sallis et al. (1990) found that enhanced access decisions about the location of future urban recreational
was associated with higher levels of participation in facilities (Goodchild & Booth, 1980); or to evaluate the
vigorous activity regardless of one’s socioeconomic effects of ‘spatial structure’ (i.e., proximity to nearby
status. Consequently, Scottish findings (Macintyre recreational opportunities) as a means of understanding
et al., 1993) demonstrating an inequitable distribution the ‘competitive and agglomerative relationship’ be-
in recreational facilities in favour of high socioeconomic tween tourism facilities (Kim & Fesenmaier, 1990) or to
status suburbs raise concerns that poor access in model recreational trip distribution (Baxter & Ewing,
disadvantaged areas contributes to lower levels of 1981).
participation in physical activity (Macintyre et al., Similarly, there appears to be no published examples
1993; King et al., 1995). of the use of models to assess the individual accessibility
Geographers conceive accessibility as a measure of the of recreational facilities. Generally, researchers have
spatial distribution of facilities adjusted for the desire examined the ease of access of population groups living
and the ability of people to overcome ‘spatial in different areas (Knox, 1978). Examining spatial access
separation’ (i.e., distance or travel time) to access a of population groups assumes that every individual
facility or activity (Hansen, 1959). Accordingly, desire living in a study area has an equal share of the
and ability are influenced by the importance of the trip opportunities, regardless of exactly where they live.
to the user (i.e., the felt need), attractiveness of the The extent to which there are individual variations in
facility (e.g., the facility’s attributes), the facility’s spatial access within a group and the ability to take-up
location and the user’s access to transport (Hansen, available opportunities, is not considered in this
1959). Although it has been demonstrated that use is approach.
inversely related to distance travelled, the extent of the Thus, in this study, individual spatial access to nine
distance of decay factor depends upon the attractiveness facilities commonly used by Australian adults for
of the destination, as well as the nature of the trip and its recreational physical activity (DASETT, 1988; DASET,
importance to the user. 1992) was examined: six formal facilities (gym or health
One approach to measuring accessibility is based on a club; swimming pool; sporting complex or recreation
variation of the gravity model (Hansen, 1959) such that centre; tennis court; golf course or other (e.g., yacht and
X sailing clubs, squash or badminton courts); and three
Ai ¼ mj =dijb ; informal facilities (public open space, river and beach)).
j
It was beyond the scope of this study to examine the
where Ai is a measure of accessibility from origin i; mj is factors that made each of these facilities ‘attractive’,
the attractiveness of destination j; dij is the travel time or although factors that made public open space attractive
distance between origin i and destination j; and b is an were considered (Corti, 1998; Broomhall, 1996). Thus, it
estimated destination-specific distance-decay parameter is assumed that all other facilities are equally attractive.
between i and j:
Distance is generally perceived to be a barrier to use a
facility, with those living closer having increased Methods
opportunities for use and paying less for transport
(Knox, 1978). However, in some instances, the effect of Study design
distance may be moderated by the attractiveness of a
facility or activity and its proximity to other facilities or The results reported are based on a cross-sectional
activities (i.e., its spatial structure) (Fotheringham, 1981; survey of adults aged 18–59 years undertaken between
Kim & Fesenmaier, 1990). Measures of attractiveness August 1995 and March 1996. A probability cluster
appear to be context-specific and aspects of attractive- sample was drawn by the Australian Bureau of
ness vary in surveys depending on the type of recreation. Statistics. One eligible respondent was randomly se-
For example, in a survey examining recreational travel lected from each household (the eligible person whose
to regional parks, attractiveness factors included the size birthday fell closest to the day of the interview was
1796 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
selected). After allowing for three call-backs, a response prompted by providing examples of the type of activity
rate of 52.9% of households was achieved. The final targeted by each question.
sample included 1803 respondents from 277 ABS
‘collectors’ districts: n ¼ 939 (137 CDs) from the 80th Spatial access
percentile and above of ABS social advantage and n ¼
874 (140 CDs) from the 20th percentile or below. Access indices based on Hansen’s accessibility model
The study design controlled for major confounding (Hansen, 1959) were developed for the main formal and
variables for recreational physical activity and the informal facilities used by respondents. The develop-
use of recreational facilities (Miettinen, 1985). Those ment of the access indices involved three tasks: geocod-
excluded from the study were: the unemployed; those ing addresses, road network analysis and estimating the
who were physically active as part of their job; those access indices. The methods are fully described else-
with any medical condition likely to affect participation where (Corti, 1998) and are summarised below. In
in recreational physical activity including walking addition, a summary access index was also developed for
(Hofstetter et al., 1991); those aged over 59 years built (i.e., golf courses, gym/health club/exercise centres;
(because of the likelihood that medical problems sport and recreational centres; swimming pools; tennis
increase with age); and those who had occupied a courts and public open space) and natural (i.e., beaches
household for less than one year. The final sample was and river) facilities.
also restricted to those who were proficient in the The spatial location of destination addresses (i.e., the
English language. recreational facilities) were determined using MapInfo
The study was restricted to a 408 km2 area of Perth, (MapInfo, 1992) (referred to as geocoding). For the
Western Australia (see Fig. 1). With a population of origin addresses (i.e., the respondents’ home addresses),
about 1.2 million, Perth is one of the smaller Australian an automatic look-up file based on the Valuer General’s
capital cities and enjoys a relatively high standard of spatial database was developed. Road network analysis
living. The study area was selected by creating artificial was undertaken to determine the distance between the
boundaries in the north, east and south. The ocean origins and destinations, using the ArcInfo GIS software
created a natural boundary in the west. The study area (ESRI, 1994). Using a street network database of the
was specifically selected to avoid newly established Perth metropolitan area, distances were measured in
suburbs because they may have had insufficiently metres and were based on the shortest possible road
developed community recreational infrastructure. network route between the origins and destinations.
The final task involved calculating the access indices.
Measures Decay of distance parameters was estimated for nine
different facilities. To estimate the decay of distance
A copy of the survey instrument is available upon parameters, the distance between the origins and
request to the principal author. The final instrument had destinations were grouped into 500 m categories.
255 items, including 27 items measuring physical activity The bottom category was 500 m and less and the
behaviour; eight measuring social environmental fac- top category was 20 km and beyond. This variable
tors; 81 measuring cognitive factors and 15 measuring was cross-tabulated by whether or not respondents had
details of the streetscape outside the respondent’s home. used the facility. The cross-tabulated data were used to
The reliability of newly developed items was assessed in create a new data file containing the mid-points of the
an extensive pilot phase. After preliminary analysis of distance categories (e.g., 250 m for the bottom category)
the data (Corti, 1998), a subset of these variables was and the percentage of opportunities to access facilities
used in the analysis that follows. actually used by respondents in each distance category.
A modified version of the physical activity items used Using a linear regression model, the log of distance was
in other major Australian studies were used to measure regressed on the log of percentage of opportunities for
physical activity (Risk Factor Prevalence Study Man- access to facilities that were used. This produced a
agement Committee, 1990; Bauman et al., 1996a, b). The beta coefficient that represented the decay of distance
frequency and total duration of all types of physical parameter. This procedure was repeated for each
activity undertaken in the previous two weeks were facility.
collected: vigorous activity, light to moderate activity,
walking for recreation and walking for transport. Dependent variables
Consistent with current public health messages at the
time, the walking questions were extended to include Two dichotomous behavioural dependent variables
walking for transport and the question on light to were examined: use of facilities and exercising as
moderate activity specifically mentioned gardening. In recommended. A variable was developed for each of
accordance with Stephen et al.’s (1985) recommenda- the nine formal and informal recreational facilities used
tion, for each type of activity respondents were by respondents for their physical activities in the
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1797
previous two weeks (referred to as ‘use of facility’ independent of body weight was calculated using the
variables). following formula:
‘Exercising as recommended’ was defined as the X
accumulation of the equivalent of 30 min or more of MET min=week ¼ fðMETÞi di fi g;
i
moderate physical activity on most days of the week
(USDHHS, 1996). To enable participation in different where ðMETÞi ; di and fi are, respectively, the MET level,
types of activity to be aggregated, a total activity index duration in minutes and frequency per week of the ith
1798 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
activity (Ainsworth et al., 1993; Wolf et al., 1994). The variables comprising multiple items, internal consistency
‘‘MET’’ is a measure of metabolic equivalents. The was assessed. The results ranged from 0.71 to 0.86
assigned MET levels were based on the compendium of indicating satisfactory internal consistency.
activities developed by Ainsworth et al. (1993). Vigorous A series of unconditional logistic regression analyses
activity was assigned a MET value of 8.0, light to were undertaken. A number of criteria were used to
moderate activity a MET value of 4.0, walking for assess the significance of variables for inclusion in the
recreation a MET value of 3.5, and walking for final model. Variables for inclusion in the final model
transport a MET value of 4.0. The decision to use an were assessed in terms of whether they were statistically,
index independent of body weight was based on two empirically or theoretically important. Statistically sig-
factors: (1) Ainsworth et al.’s (1993) caution against the nificant variables were those with p-values less than 0.05,
use of kcal scores in correlation analyses since coeffi- or 95% confidence intervals that excluded unity.
cients may reflect body weight rather than the energy Empirically significant variables were classified as those
costs of activities; and (2) the fact that public health that did not reach statistical significance, but had point
messages focus on duration, frequency and intensity of estimates at least 20% greater or lower than the
physical activity with no attempt to vary messages for reference category. However, variables with wide con-
people of different body weights. Initial analysis of the fidence intervals were excluded from the model on the
data demonstrated that categorising individuals based basis of the strong possibility that the result had
on weight-dependent energy expenditure had the poten- occurred by chance. Finally, theoretically significant
tial to misclassify individuals who exercise above or variables were those that the literature suggested were
below recommended levels of physical activity, depend- important for inclusion, regardless of the empirical
ing on how much they weighed (Corti, 1998). Thus, results. However, in the interest of parsimony, theore-
‘exercising as recommended’ was classified as those tical importance was assessed using the precision of the
expending 840 MET (minute/week) or more on recrea- estimates, before making a final decision to include a
tional physical activity (i.e., the equivalent of 30 or more variable in the final model. This approach made way for
min week 4 MET). To avoid measurement error due the development of a new ecological model that
to over-reporting, those reporting energy expenditure of combined individual, social environmental and physical
10,000 MET (min/week) or more were excluded (n ¼ 30) environmental variables. The use of the parsimonious
(Booth, Owen, & Bauman, 1995; Bauman et al., model is justified on the grounds of maximum statistical
1996a, b). efficiency, while presenting a high degree of validity of
the point estimates (Hosmer & Lemeshow, 1989).
Independent variables To enable examination of the relative influence of the
individual, social environmental and physical environ-
In the models predicting use of individual recreational mental determinants, in the final stage of the analysis,
facilities, apart from demographic variables, the main the three ‘classes’ of determinants (i.e., individual, social
effect variables were spatial access to each individual environmental and physical environmental) (respectively
facility. Spatial access was recoded into quartiles of referred to as ‘determinant scores’) and the demographic
access where 1=top quartile of access and 4=bottom factors were summarised into multivariate summarisa-
quartile of access. tion scores as suggested by Miettinen (1976). The scores
The choice of variables for inclusion in the model were composite scores constructed using the beta
associating exercising as recommended with individual, coefficients from the final parsimonious models for
social environmental and physical environmental deter- recommended
P levels of exercising overall as
minants was based on the results of previous analyses follows: i bi xi ; where bi is the beta coefficient for the
examining the independent influence of individual, ith binary indicator variable of a particular class (i.e.,
social and physical environmental variables (Corti, individual, social environmental, physical environmental
1998). The overall model included 14 independent main determinants or demographic factors) included in the
effect variables (see Table 1): seven individual cognitive final parsimonious logistic regression model, and xi is
variables; three social environmental variables and four the respondent’s value (0, 1) for the ith binary indicator
physical environmental variables. In addition, six variable to be included in the score.
demographic variables were included to control for The multivariate summarisation score technique was
potential confounding (i.e., age, sex, number of children proposed by Miettinen (1976) to overcome inefficiency
in the household under 18 years, work outside the home, in analyses which required control of a large number of
household income and education). covariates considered to be confounding variables. In
The analysis was based on 1773 respondents who situations where there are a large number of strata,
reported in-scope data and was undertaken using the potential problems arise from small numbers of respon-
Statistical Package for the Social Sciences (SPSS) dents within each stratum. Thus, Miettinen proposed the
software. After creating scales for individual cognitive development of a single multivariate confounding
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1799
Table 1
Variables used in the model
Individual cognitive
Attitude toward trying ‘All things considered, would you say the 7 point: 3 to 3
fact you tried to do a regular exercise
routinea over the next two weeks would
make you feely: foolish/wise, dissatisfied/
satisfied, bad/good’b
Attitude toward the process of trying How ‘unpleasant/pleasant’, ‘difficult/ 7 point: 3 to 3
easy’, ‘bad/good’ the process of trying to
do a regular exercise routinea would be
‘regardless of whether you succeed or
fail’b
Subjective norm ‘How likely or unlikely are each of the 7 point: 3 to 3
following people (your spouse or partner,
a family member, a friend, your doctor) to
think you should try to do a regular
exercise routinea during the next two
weeks?’b
Frequency of trying in the last three During the past three months, how many 5 point: 1=Never, 5=Weekly
months times, if any, did you try to do a regular
physical activity routinea?
Perceived behavioural control Assuming that you tried to do a regular 7 point: 1=Very unlikely, 7=Very likely
exercise routinea over the next two weeks,
how likely or unlikely is it that you would
actually stick to your routine?’
Behavioural skills used in the last month How frequently in the last month did you: 5 point: 1=Never, 5=Weekly
set a goal for how much physical activity
you would like to do; plan particular days
on which you would do physical activity;
and arrange to meet someone to do
physical activity with?a
Intention to try in the next two weeks Before this interview, how likely or 7 point: 1=Very unlikely, 7=Very likely
unlikely is it that in the next two weeks
you would try to do a regular exercise
routinea?
Social environmental
Club membership Are you a member of a sport, exercise or 1=Yes, 0=No;
outdoor recreational group or club?
Frequency of participation in physical How often over the last month, did the 5 point: 1=Never, 5=More than once a
activity by five significant others following peoplec do physical activity weekd,e
including walking?
Frequency of a significant other doing Over the past three months, which of these 4 point: 0=Never, 4=weeklyd,f
physical activity with respondent people,c if any, did physical activity with
you including walking?
Physical environmental
Functional environment The interviewer noted whether the street 4 categories: no footpath or shop; no
that had a footpath (on both or one side footpath, but a shop; a footpath, but no
of street) or shop visible in the street shop; or both a footpath and a shop
Appeal of environment The interviewer noted the type of street 4 categories: no trees and major traffic;
(cul de sac; minor local road; major local major traffic and some trees; minor traffic
road; highway or major thoroughfare) and no trees or minor traffic and some
and whether the street was tree lined; trees
had good coverage (more than one tree to
a house block); medium coverage (one tree
of a house block); poor coverage (less than
one tree to a house block); mixed tree
coverage; no trees
1800 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
Table 1 (continued)
Overall spatial access to built facilities Total spatial access to a range of purpose- 4 categories: 1=Top quartile of access,
built recreational facilities including golf 4=Bottom quartile of access
courses, gym/health club/exercise centres;
sport and recreational centres; swimming
pools; tennis courts and public open
space.
Overall spatial access to natural facilities Total spatial access to naturally occurring 4 categories: 1=top quartile of access,
recoded into quartiles of access recreational facilities such as the river and 4=bottom quartile of access
beach)
a
Respondents were given a choice of preferred exercise, i.e., either three 20 min sessions of vigorous exercise each week, or 30 min to
one hour of light to moderate exercise each day.
b
Recoded as the mean of the sum of these items.
c
Spouse or partner; close family members; people at work; close friends and people in your neighbourhood. (Based on items
developed by Sallis et al. (1990).
d
After an initial examination of the results, these variables were recoded to ‘more than once a week’=1 and all other responses=0.
A single item was developed called the ‘Number of significant others who did physical activity’.
e
More than once a week in the last month.
f
Weekly with the respondent in the last three months.
summarisation score which, when grouped into cate- not complete secondary school (51.8%). Similarly, the
gories, could be used for cross-classification purposes. prevalence of exercising as recommended increased with
The multivariate summarisation scores or determi- household income.
nant scores are comparable to ‘risk scores’, which have
had various applications (Shaper, Pocock, Phillips, &
Walker, 1986; Lemeshow, Teres, Avrunin, & Pastides, Distance of decay parameters
1988). In this study, the individual, socio-environmental
and physical environmental determinant scores were Decay of distance parameters (b) were estimated for
grouped into tertiles representing low, medium and high nine formal and informal recreational facilities (see
‘risk’ scores for each determinant. The demographic Table 3). A decay of distance parameter >1 indicates
score, used to adjust for confounding, was grouped into that as distance doubles, facility use reduces by more
deciles (Miettinen, 1976). than one-half. The results indicated that the public’s use
of public open space was more sensitive to distance
(b ¼ 1:91) than sporting and recreation centres
(b ¼ 1:16) and golf courses (b ¼ 1:06). The sensitivity
Results to distance for use of public open space is consistent
with the literature (Boyle, 1983; Just, 1989), which
Physical activity behaviour suggests that most users of public open space live within
close proximity (i.e., around 500 m). The decay of
In the two weeks before the survey, approximately distance parameters was used to estimate individual
one-half of all respondents reported participating in access indices for each respondent to each facility.
vigorous physical activity (47.8%), a similar proportion
that participated in light to moderate (52.5%) physical
activity, while 82.5% walked for either or both transport Use of recreational facilities
(72.1%) and recreation (68.5%). Overall, 59.2% of
respondents exercised as recommended (see Table 2). In the two weeks before the survey, 76.3% of
The prevalence in women was 0.90 (95% CI 0.83–0.98) respondents used one or more community facilities
of that in men; and in households with one or more (including the streets) for their recreational physical
children it was 0.88 (95% CI 0.81–0.95) of that in activity. The most popular facilities were informal
households with no children. Conversely, education facilities, including the streets (45.6%), public open
level and household income were positively associated space (28.8%) and the beach (22.3%). Some 8.9% of
with physical activity. Exercising as recommended was respondents used the river. A number of formal facilities
17–26% higher in those with certificates (60.4%) or were used by a sizeable proportion of the sample: 10.8%
tertiary education (65.6%) compared with those who did used a gym, health club or exercise centre; 8.9% a
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1801
Table 2
Relationship of exercising as recommended to demographic determinants
n % PRa 95% CI
Marital status
Married/defacto 1074 57.8 1.00
Separated/widowed/divorced 212 54.2 0.94 0.82–1.07
Never married 483 64.2 1.11 1.02–1.21
swimming pool; 8.7% a sport or recreation centre; and physical activity in the previous two weeks reported
7.1% a tennis court. using one or more formal or informal facility ‘near
Regardless of the intensity of the physical activity home’. Facilities located ‘at or near work’ were used by
undertaken (i.e., vigorous, light to moderate or walk- substantially fewer respondents (o7.3% and o5.6%,
ing), the vast majority of respondents who undertook respectively). Similarly, only a minority of respondents
1802 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
Table 4
Odds ratios from logistic regression associating use of each individual facility to level of access
Access to beach
Top quartile of access 1.00 1.00 1.00
2nd top quartile 0.56 0.52 0.51 0.42–0.93 0.000
3rd quartile 0.28 0.25 0.26 0.19–0.37 0.000
10th–25th percentile 0.16 0.14 0.14 0.09–0.23 0.000
Bottom 10th percentile 0.09 0.08 0.10 0.05–0.20 0.000
Access to river
Top quartile of access 1.00 1.00 1.00
2nd top quartile 0.32 0.32 0.33 0.23–0.47 0.000
3rd quartile 0.20 0.20 0.21 0.14–0.32 0.000
10th–25th percentile 0.04 0.04 0.05 0.02–0.13 0.000
Bottom 10th percentile 0.04 0.04 0.04 0.01–0.14 0.000
Table 4 (continued)
Compared with those in the lowest determinant score cient to achieve recommended levels of recreational
categories, those in the highest individual determinant activity in the community.
score category were 8.14 times more likely to exercise as
recommended (95% CI 6.00–11.05; test for trend
p ¼ 0:000). This was twice as high as the influence of Study limitations
the highest social environmental determinant score (OR
3.72; 95% CI 2.76–4.98; test for trend p ¼ 0:000) and Before discussing the results in detail, the study
five times as high as the highest physical environmental limitations are considered. By national and international
determinant score (OR 1.43; 95% CI 1.09–1.88; test for standards the city of Perth enjoys a comparatively high
trend p ¼ 0:017). There was no evidence of multi- standard of living and is a relatively homogeneous
plicative interaction, indicating that the results were society. Moreover, the study was confined to a defined
consistent with no effect modification on a multiplicative area of Perth rather than including the entire metropo-
scale. This suggests that the relative effect of each litan area. The decision to limit the study area was based
determinant was consistent across the categories of each on the resources available to undertake the environ-
of the other determinants. mental scan component of the study which involved a
Overall, the results suggest that the likelihood of survey of all formal recreational facilities in the study
exercising as recommended was greatly enhanced in area (Corti, 1998) and an observational study of all
those with positive individual factors and a positive public open space (Broomhall, 1996). The creation of an
social environment conducive to exercising. A suppor- artificial boundary is likely to have resulted in non-
tive physical environment as measured by this study had differential measurement error, particularly for those
a significant, but a more moderate, influence. living close to the coast and in the north- and south-west
corners of the study area.
The study conceptualised spatial access using a model
taken from the geography literature (Hansen, 1959).
Discussion However, the application of the model was somewhat
different because we examined the influence of spatial
This study found that the influence of individual and access on achieving a ‘gold standard’ of behaviour (i.e.,
social environmental determinants outweighed the role achieving recommended levels of activity). It is possible
played by physical environmental determinants of that an alternative approach may have been more
exercising as recommended. Nevertheless, access to appropriate for assessing ‘gold standard’ behaviours.
recreational facilities determined whether or not they In addition, distance-only access models were used for
were used, and in this way, support and enhance the assessing spatial access to built and natural facilities
achievement of recommended levels of physical activity without adjustment for ‘attractiveness’. This assumes
behaviour by providing opportunities. Individual and that all facilities were equally attractive, and represents a
social environmental determinants on the other hand, somewhat limited approach to adopting the accessibility
appeared more important in directly determining model. Initially, we adjusted for the attractiveness of
whether or not recommended levels of activity were public open space (Corti, 1998). However, in the absence
achieved. In this sense, it appears that having good of additional research to guide decision-making, it
access to recreational facilities is necessary but insuffi- became clear that examination of the attractiveness of
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1805
Table 5
Odds ratios from logistic regression associating exercising as recommended with individual, social environmental and physical
environmental determinants
Subjective norm
Negative/neutral 1.00 1.00
Positive 0.69 0.007 0.68 0.51–0.89 0.005
Very positive 0.75 0.109 0.72 0.51–1.01 0.059
Number of significant others who exercised with respondent weekly over the three months
None 1.00 1.00
One 1.75 0.000 1.76 1.34–2.29 0.000
Two 2.18 0.000 2.11 1.46–3.04 0.000
1806 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
Table 5 (continued)
Functional environment
No footpath, no shop 1.00 N/S
No footpath, shop 3.50 0.192
Footpath, no shop 1.12 0.437
Footpath, shop 0.82 0.468
Appeal of environment
No trees, major traffic 1.00 N/S
Major traffic, some trees 0.96 0.795
Minor traffic, no trees 1.18 0.565
Minor traffic, some trees 1.03 0.884
many different types of recreational facilities was beyond distance, as well as individual and social environmental
the scope of the current study. This could be the subject factors are likely to influence the uptake of opportunities
of future research because adjustment for attractiveness available in the physical environment.
of public open space was found to influence the
likelihood of use (Corti, 1998).
Finally, as social ecological approaches are new to Physical activity behaviour
public health research, it was decided from the outset to
include not only statistically significant variables in final In this study, less than 60% of respondents achieved
models, but also variables that were empirically and recommended levels of activity, findings consistent with
theoretically important. The approach taken is consis- those elsewhere (Armstrong, Bauman, & Davies, 2000;
tent with an epidemiologist’s view of data analysis Bull, Milligan, Rosenberg, & MacGowan, 2001). Given
(Kleinbaum, Kupper, & Morgenstern, 1982; Rothman, the popularity of walking (Siegel, Brackbill, & Heath,
1986) which considers the size and precision of estimates 1995; Armstrong et al., 2000; Bull et al., 2001) and the
rather than simply focussing on statistical significance. potential public health benefits, there is an opportunity
However, chance cannot be ruled out as an explanation for public health messages to promote brisk walking for
for some of these findings and should be the subject of recreation and transport as a means of increasing
future research. community physical activity levels. However, to increase
Despite these limitations, this initial attempt at community physical activity levels, it appears that there
creating an individual index of a physical environmental is need for more guidance about how to gain maximum
factor associated with a preventive behaviour, paves the health benefits from all activities undertaken throughout
way for a much wider application of this technique to the day (e.g., getting to work; while at work; doing
the study of physical activity and other health beha- housework; working in the garden) (US Department of
viours. This approach is relevant in situations where Health and Human Services, 1996).
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1807
Holman, 1996) suggests that the mere provision of confidence in achieving behavioural outcomes (Bandura,
proximate space is insufficient to encourage its use for 1997).
physical activity. Park usage can be predicted by its It is noteworthy that the final social ecological model
proximity to the user, as well as its size and its design did not include attitudes. Previous reviews have found
characteristics (Boyle, 1983; Just, 1989). Moreover, the that, in general, attitudes are weakly associated with
design of public open space influences perceived and behaviour (Sallis & Owen, 1999), despite their role in
actual safety; as well as providing environmental cues predicting behavioural intention. However, the model
about whether to engage in active or passive behaviour examined in this study included attitudes as well as more
(Rutledge, 1971; Gold, 1977; Whyte, 1980). objective proximate determinants of behaviour (e.g.,
recent use of behavioural skills, club membership,
number of regular exercise partners, exercise habit and
Determinants of achieving recommended levels of physical spatial access to natural or built facilities). In this
activity circumstance, attitudinal variables became insignificant
and were dropped from the final ecological model.
A main aim of this study was to explore a social It seems self-evident that people who regularly
ecological model of recreational physical activity that exercised in the past are more likely to do so in the
considered individual, social and physical environmental future. Although previous research (DASET, 1992) has
determinants of behaviour. After adjustment for other shown that even those who exercise as recommended
social and physical environmental factors, achieving complain about having insufficient time to exercise or to
recommended levels of planned recreational physical exercise more, many have developed an exercise habit
activity was more likely in those with a high level of which they are able to maintain or to re-establish even
perceived behavioural control; those with high levels of after a periodic relapse (Aarts, Paulussen, & Schaalma,
behavioural intention; those for whom exercise had 1997). It is tempting to ignore this group because of their
become a habit and those with whom they had someone very positive attitude towards the target behaviour.
to exercise. However, it has been argued that gains to public health
Social support is important for the maintenance of are likely to accrue from preventing the already active
physical activity (Sallis & Owen, 1999), and in this study, from slipping into inactivity rather than focussing solely
those with exercise partners or who were members of on the completely sedentary (Bull, 1997). Thus, in
sporting clubs were more likely to achieve recommended accordance with advice from Sheth and Frazier (Sheth &
levels of activity. Subjective norms (i.e., perceptions of Frazier, 1982), a communication reinforcement strategy
what others think one should do) were positively targeting those who already exercise regularly is required
associated with forming an intention to exercise (Blue, to ensure that positive exercise habits are maintained
1995), but paradoxically, negatively associated with (Donovan & Owen, 1994).
achieving recommended levels of activity. Previous In this study, after adjusting for individual and social
reviews have found inconsistent evidence for the role environmental factors, the physical environment ap-
played by subjective norms in influencing behaviour (US peared to have only a weak direct influence on
Department of Health and Human Services, 1996). It is behaviour. Sallis et al. (1990) examined the influence
possible that the conceptualisation of subjective norms of the environment on vigorous activity. They found a
as a form of ‘social pressure’ (Courneya & McAuley, weak positive association between the density of ‘pay’
1995) as adopted in this study, may have contributed to facilities and vigorous activity, but no such association
the finding. The results suggest that those who know for ‘free’ facilities. This was in contrast to findings by
that others think that they should try to be regularly Bauman et al. (1999) who found that after adjustment
active may form an intention to do so, but do not for other demographic factors, urban and rural ‘coastal
necessarily translate their ‘good intentions’ into action. dwellers’ were more physically active than urban and
This finding supports the notion of an intention- rural inland dwellers due to the so-called ‘coastal-effect’
behaviour gap (Norman & Conner, 1996). (Bauman et al., 1999).
Consistent with numerous other studies (Godin,
Valois, & Lepage, 1993; Wankel, Mummery, Stephens, Relative influence and interaction between determinants
& Craig, 1994; Kimiecik, 1992), the importance of
perceived behaviour control was confirmed in this social Despite the lack of empirical evidence, contemporary
ecological study. Perceived behavioural control and the health promotion places considerable emphasis on
use of behavioural skills were both determinants of creating supportive environments (World Health Orga-
physical activity behaviour. Although not explored, it is nization, 1986). Consistent with this trend, there have
possible that the effect of behavioural skills was been calls for greater consideration of the physical
mediated through perceived behavioural control, as it environment in physical activity research and practice
is likely that the use of behavioural skills increases (Sallis & Owen, 1999). This seems self-evident: a
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1809
supportive environment is likely to provide opportu- consideration needs to be given to streetscape design
nities which, in Nancy Milio’s words, make ‘health to maximise physical activity levels (Sallis et al. (1998);
choices, easy choices’ (Milio, 1986). HEA, 1999).
Thus, perhaps the most striking finding of this study Finally, the results of this study suggest that public
was that the influence of the physical environment on education campaigns aimed at increasing recreational
physical activity levels, generally was relatively weak and physical activity need to address five direct determinants
secondary to individual and social environmental of physical activity: perceived behavioural control;
determinants. The significance of the results is not that behavioural skills; behavioural intention; encouraging
those with high determinant scores were most likely to physical activity habit; and creating social support by
achieve recommended levels of activity, but rather that encouraging regular physical activity with exercise
those with high individual and social determinant scores partners. The results suggest that public education
were substantially more likely to do so than those with a campaigns could focus on providing more concrete
high physical environmental determinant score. cognitive training focussing on behavioural skills that
translate good intentions into action.
Conclusion
Acknowledgements
Based on the findings of this study, it is tempting to
conclude that the physical environment may be of The Western Australian Health Promotion Founda-
limited importance to achieve overall recommended tion (Healthway) provided funding for this project. A
levels of physical activity in the community. Such a number of colleagues from the Department of Public
conclusion would be premature given the limitations of Health provided advice in relation to the spatial access
the approach taken in this study and the early stage of component of this study including Ms Jilda Hyndman
development of methods to study environmental deter- (road network analysis) Dr. John Bass, Dr. Richard
minants of behaviour. Moreover, in accordance with Parsons, Ms. Diana Rosman, Mr. Mark Divitini and
health promotion rhetoric, this study found that Mr. Michael Winter. Mr. David Blackmore also
accessible recreational facilities determined use of those provided advice. Dr. Ray James provided editorial
facilities and thus, good access is necessary to create a advice for this manuscript.
supportive environment. Nevertheless, in line with the
work of others (Brownell et al., 1980; Sallis et al., 1990;
Blamey et al., 1995), it appears that access to a References
supportive physical environment alone may be insuffi-
cient to increase community recreational activity levels. Aarts, H., Paulussen, T., & Schaalma, H. (1997). Physical
Rather, it would seem that environmental modification exercise habit: On the conceptualization and formation of
must be complemented by strategies that aim to habitual health behaviours. Health Education Research,
influence individual and social environmental factors 12(3), 363–374.
to support participation. Ainsworth, B. E., Haskell, W. L., Leon, A. S., Jacobs, D. R.,
This study also showed that only a small proportion Montoye, H. J., Sallis, J. F., & Paffenbarger, R. S. (1993).
of healthy Western Australian adults use many of the Compendium of physical activities: Classification of energy
recreational facilities available in the community, costs of human physical activities. Medicine and Science in
Sports and Exercise, 25(1), 71–80.
particularly formal facilities. In cities where there is
Ajzen, I. (1985). From intentions to action: A theory of planned
relatively easy access to recreational facilities, designing behavior. In J. Kuhl, & J. Beckman (Eds.), Action-control:
campaigns that promote the use of existing facilities from cognition to behavior (pp. 11–39). Heidelberg: Springer.
(rather than solely focussing on promoting physical Ajzen, I. (1991). The theory of planned behaviour. Organisa-
activity per se) may be helpful. These campaigns might tional Behaviour and Human Decision Processes, 50,
provide some useful information concerning whether 179–211.
providing cues to use existing facilities can influence Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and
planned recreational activity in the same way that visual predicting social behaviour. Englewoods Cliffs, NJ: Prentice-
cues have been shown to increase use of stairs (Brownell Hall, Inc..
et al., 1980; Blamey et al., 1995). Ajzen, I., & Madden, T. J. (1986). Prediction of goal-directed
behaviour: Attitudes, intentions and perceived behavioral
Given the popularity of walking for recreation and
control. Journal of Experimental Social Psychology, 22,
transport, there needs to be greater emphasis on the 453–474.
streetscape as an important component of the physical Andrew, G. M., Oldridge, N. B., Parker, J. O., Cunningham,
activity facility infrastructure. While more research is D. A., & Rechnitzer, P. A. (1981). Reasons for dropout
required to fully understand the role played by from exercise programs in post-coronary patients. Medical
streetscapes in influencing physical activity, greater and Science in Sports and Exercise, 13(3), 164–168.
1810 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
Armstrong, T., Bauman, A., & Davies, J. (2000). Physical Corti, W. (1998). The relative influence of, and interaction
activity patterns of Australian adults. Results of the 1999 between, environmental and individual determinants of
national physical activity survey. Canberra: Australian recreational physical activity in sedentary workers and home
Institute of Health and Welfare. makers. Ph.D. thesis, Department of Public Health, The
Bagozzi, R. P., & Warshaw, P. R. (1990). Trying to consume. University of Western Australia, Perth.
Journal of Consumer Research, 17, 127–140. Corti, B., Donovan, R., & Holman, C. J. (1996). Factors
Bandura, A. (1997). Self-efficacy. The exercise of control. New influencing the use of physical activity facilities: Results
York: WH Freeman and Company. from qualitative research. Health Promotion Journal of
Bauman, A., Bellew, B., Booth, M., Hahn, A., Stoker, L., & Australia, 6(1), 16–21.
Thomas, M. (1996a). NSW Health Promotion Survey 1994. Courneya, K. S., & McAuley, E. (1995). Cognitive mediators of
Towards best practice for the promotion of physical activity the social influence-exercise adherence relationship: A test of
in the Areas of NSW. NSW Health Department, Centre for the Theory of Planned Behavior. Journal of Behavioral
Disease Prevention & Health, NSW. Medicine, 18(5), 499–515.
Bauman, A., Wallner, F., Miners, A., & Westley-Wise, V. DASET (Department of the Arts Sport the Environment and
(1996b). No ifs. No buts. Illawara physical activity project. Territories). (1992). Pilot survey of the fitness of Australians.
Baseline research report. Illawarra: Illawarra Area Health Canberra: Australian Government Publishing Service.
Service. DASETT (Department of the Arts Sport the Environment,
Bauman, A., & Smith, B. (1999). Geographical influences upon Tourism and Territories). (1988). Physical activity levels of
physical activity participation: Evidence of a ‘coastal effect’. Australians. Canberra: Australian Government Publishing
Australian and New Zealand Journal of Public Health, 23(3), Service.
322–324. Dishman, R. K. (1988). Exercise adherence research: Future
Baxter, M., & Ewing, G. (1981). Models of recreational trip directions. American Journal of Health Promotion, 3(1), 52–
distribution. Regional Studies, 15(5), 327–344. 56.
Blamey, A., Mutrie, N., & Aitchison, T. (1995). Health Donovan, R. J., & Owen, N. (1994). Social marketing and
promotion by encouraged use of stairs. British Medical population interventions. In R. Dishman (Ed.), Advances in
Journal, 311, 289–290. Exercise Adherence. Champaign, IL: Human Kinetics.
Blue, C. L. (1995). The predictive capacity of the Theory of Environmental Systems Research Institute, Inc. (1994). ArcInfo
Reasoned Action and the Theory of Planned Behaviour in Version 7. Redlands, California, CA: Environmental Sys-
exercise research: An integrated literature review. Research tems Research Institute, Inc.
in Nursing and Health, 18, 105–121. Fotheringham, A. S. (1981). Spatial structure and distance-
Booth, M., Owen, N., & Bauman, A. (1995). Active & inactive decay parameters. Annals of the Association of American
Australians. Assessing and understanding levels of physical Geographers, 71(3), 425–436.
activity. Canberra: Department of the Environment, Sport Francis, M. (1987). Urban open spaces. In E. H. Zube, & G. T.
and Territories. Moore (Eds.), Advances in environment, behavior and design,
Boyle, R. (1983). Survey of the use of small parks. Australian vol. 1 (pp. 71–106). New York: Plenum Press.
Parks and Recreation, 41–43. Godin, G. (1993). The theory of reasoned action and planned
Broomhall, M.H. (1996). Study of the availability and environ- behaviour: Overview of findings, emerging research pro-
mental quality of urban open space used for physical activity. blems and usefulness for exercise promotion. Journal of
Master of public health dissertation, Department of Public Applied Sport Psychology, 5, 141–157.
Health, The University of Western Australia, Perth. Godin, G., Valois, P., & Lepage, L. (1993). The pattern
Brownell, K. D., Stunkard, A. J., & Albaum, J. M. (1980). of influence of perceived behavioral control upon exercis-
Evaluation and modification of exercise patterns in the ing behavior: An application of Ajzens Theory of
natural environment. American Journal of Psychiatry, Planned Behaviour. Journal of Behavioural Medicine,
137(12), 1540–1545. 16(1), 81–102.
Bull, F. (1997). The role of general practitioners in the promotion Gold, S. M. (1977). Neighbourhood parks. The non-use
of physical activity. Doctor of Philosophy thesis, Depart- phenomenon. Evaluation Quarterly, 1(2), 319–328.
ment of Human Movement, The University of Western Goodchild, M., & Booth, P. (1980). Location and allocation of
Australia; Burgess, J., Harrison, C.M., & Limb, M. (1988). recreation facilities: Public swimming pools in London,
People, parks and the urban green: A study of popular Ontario. Ontario Geography, 15, 35–51.
meanings and values for open spaces in the city. Urban Hahn, A., & Craythorn, E. (1994). Inactivity and the physical
Studies, 25, 455–473. environment in two regional centres. Health Promotion
Bull, F., Milligan, R., Rosenberg, M., & MacGowan, H. (2001). Journal of Australia, 4(2), 43–45.
Physical activity levels of Western Australian Adults 1999. Hansen, W. G. (1959). How accessibility shapes land use.
Perth: Health Department of Western Australia and Journal of the American Institute of Planners, 15, 73–76.
Sport and Recreation Way2Go, Western Australian Health Canada, Health Promotion and Programs Branch.
Government. (1997). Framework for action. 1998–99–2001. Health
Carroll, J. (1995). Local government sport & recreation Canada.
provision in Australia. A report examining local govern- Health Education Authority. (1999). Making T.H.E. links.
ment involvement in sport and recreation provision and Integrating sustainable transport, health and environmental
associated issues. Confederation of Australian Sport, policies: A guide for local authorities and health authorities.
Canberra. London: Health Education Authority.
B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812 1811
Hofstetter, C. R., Hovell, M. F., & Sallis, J. F. (1990). Social Norman, P., & Conner, M. (1996). The role of social cognition
learning correlates of exercise self-efficacy: Early experiences models in predicting health behaviours: Future directions.
with physical activity. Social Science & Medicine, 31(10), In M. Conner, & P. Norman (Eds.), Predicting health
1169–1176. behaviour. Research and practice with social cognition
Hofstetter, C. R., Hovell, M. F., Macera, C., Sallis, J. F., Spry, models. Buckingham: Open University Press.
V., Barrington, E., Callender, L., Hackley, M., & Rauh, M. RFPS (Risk Factor Prevalence Study) Management Commit-
(1991). Illness, injury, and correlates of aerobic exercise and tee. (1990). Risk factor prevalence study survey no 3 1989.
walking: A community study. Research Quarterly for Canberra: National Heart Foundation of Australia and
Exercise and Sport, 62(1), 1–9. Australian Institute of Health.
Hosmer, D. W., & Lemeshow, S. (1989). Applied logistic Rothman, K. J. (1986). Modern epidemiology. Little Brown:
regression. New York: John Wiley. Boston.
Just, D. (1989). Appropriate amounts and design of open Rutledge, A. J. (1971). Anatomy of a park. The essentials of
spaces. Australian Parks and Recreation, 25(2), 32–39. recreation area planning and design. New York: McGraw-
Kim, S.-I., & Fesenmaier, D. R. (1990). Evaluating spatial Hill Book Company.
structure effects in recreation travel. Leisure Sciences, 12, Sallis, J. F., Bauman, A., & Pratt, M. (1998). Environmental
367–381. and policy interventions to promote physical activity.
Kimiecik, J. (1992). Predicting vigorous physical activity of Preventive Medicine, 15(4), 379–397.
corporate employees: Comparing the Theories of Reasoned Sallis, J. F., Hovell, M. F., Hofstetter, C. R., Elder, J. P.,
Action and Planned Behavior. Journal of Sport and Exercise Hackley, M., Caspersen, C. J., & Powell, K. E. (1990).
Psychology, 14, 192–206. Distance between home and exercise facilities related to
King, A. C., Jeffery, R. W., Fridinger, F., Dusenbury, L., frequency of exercise among San Diego residents. Public
Provence, S., Hedlund, S. A., & Spangler, K. (1995). Health Reports, 105(2), 179–185.
Environmental and policy approaches to cardiovascular Sallis, J. F., & Owen, N. (1996). Ecological models. In K.
disease prevention through physical activity: Issues and Glanz, F. M. Lewis, & B. K. Rimer (Eds.), Health behavior
opportunities. Health Education Quarterly, 22(4), 499–511. and health education: Theory, research, and practice (2nd ed.)
Kleinbaum, D. G., Kupper, L. L., & Morgenstern, H. (1982). (pp. 403–424). San Francisco: Jossey-Bass.
Epidemiologic research. Principles and quantitative methods. Sallis, J. F., & Owen, N. (1999). Physical activity and behavioral
London: Lifetime Learning Publications. medicine. Thousand Oaks, CA: Sage publications.
Knox, P. (1978). The intraurban ecology of primary medical Secretary of State for Health. (1998). Our healthier nation.
care: Patterns of accessibility and their policy implications. A contract for health. London: Her Majestys Stationery
Environment and Planning, 10, 415–435. Office.
Lemeshow, S., Teres, D., Avrunin, J., & Pastides, H. (1988). Shaper, A., Pocock, S., Phillips, A., & Walker, M. (1986).
Predicting the outcome of intensive care unit patients. Identifying men at high risk of heart attacks: Strategy for
Journal of American Statistical Association, 83, 348–356. use in general practice. British Medical Journal, 293(23),
Linenger, J. M., Chesson, C. V., & Nice, D. S. (1991). Physical 474–479.
fitness gains following simple environmental change. Amer- Sheth, J. N., & Frazier, G. L. (1982). A model of strategy mix
ican Journal of Preventive Medicine, 7(5), 298–310. choice for planned social change. Journal of Marketing, 46,
Macintyre, S., Maciver, S., & Sooman, A. (1993). Area, class 15–26.
and health: Should we be focusing on places or people? Siegel, P. Z., Brackbill, R. M., & Heath, G. W. (1995). The
Journal of Social Policy, 22(2), 213–234. epidemiology of walking for exercise: Implications for
Mapping Information Systems Corporation. (1992). MapInfo promoting activity among sedentary groups. American
(Version 2.1.1 for Windows). Troy, New York: Mapping Journal of Public Health, 85(5), 706–710.
Information Systems Corporation. Sheppard, B. H., Hartwick, J., & Warshaw, P. R. (1988). The
McKenzie, T. L., Sallis, J. F., Nader, P. R., Broyles, S. L., & theory of reasoned action: A meta-analysis of past research
Nelson, J. A. (1992). Anglo– and Mexican–American with recommendations for modifications and future re-
preschoolers at home and at recess: Activity patterns and search. Journal of Consumer Research, 15, 325–343.
environmental influences. Developmental and Behavioral Stephens, T., Jacobs, D., & White, C. (1985). A descriptive
Pediatrics, 13(3), 173–180. epidemiology of leisure-time physical activity. Public Health
McLeroy, K. R., Bibeau, D., Steckler, A., & Glanz, K. (1988). Reports, 100(2), 147–158.
An ecological perspective on health promotion programs. Stevenson, H. M., & Burke, M. (1992). Bureaucratic logic in
Health Education Quarterly, 15(4), 351–377. new social movement clothing: The limits of health
Miettinen, O. S. (1976). Stratification by a multivariate promotion research. Canadian Journal of Public Health
confounder score. American Journal of Epidemiology, Supplement, 1, s47–s53.
104(6), 609–620. Stokols, D. (1992). Establishing and maintaining healthy
Miettinen, O. (1985). Theoretical epidemiology: Principles of environments. American Psychologist, 47(1), 6–22.
occurrence research in medicine. New York: Wiley. Stokols, D. (1996). Translating social ecological theory into
Milio, N. (1986). Promoting health through public policy. guidelines for community health promotion. American
Ottawa: Canadian Public Health Association. Journal of Health Promotion, 10(4), 282–298.
Minkler, M. (1989). Health education, health promotion and Teraslinna, P., Partanen, T., Koskela, A., & Oja, P.
the open society: An historical perspective. Health Educa- (1969). Characteristics affecting willing of executives to
tion Quarterly, 16, 17–30. participate in an activity program aimed at coronary heart
1812 B. Giles-Corti, R.J. Donovan / Social Science & Medicine 54 (2002) 1793–1812
disease prevention. Journal of Sports Medicine, 9(4), psychological variables: Results from the Campbells survey
224–229. of well-being. Journal of Sport and Exercise Psychology, 16,
Tesh, S. N. (1988). Hidden arguments: Political ideology and 56–69.
disease prevention policy. New Brunswick: Rutgers Whyte, W. H. (1980). The social life of small urban spaces.
University Press. Washington, DC: The Conservation Foundation.
US Department of Health, Human Services. (1996). Physical Wolf, A. M., Hunter, D. J., Colditz, G. A., Manson, J. E.,
activity and health. A report of the Surgeon General. US Stampfer, M. J., Corsano, K. A., Rosner, B., Kriska, A., &
Department of Health and Human Services, Centres for Willett, W. C. (1994). Reproducibility and validity of a self-
Disease Control and Prevention, National Center for administered physical activity questionnaire. International
Chronic Disease Prevention and Health Promotion, Atlan- Journal of Epidemiology, 23(5), 991–999.
ta, Georgia. World Health Organization. (1986). The Ottawa Charter for
Wankel, L. M., Mummery, W. K., Stephens, T., & Craig, C. L. health promotion. Health Promotion International, 1, 3–5.
(1994). Prediction of physical activity intention from social