doi:10.1016/j.amepre.2004.10.018


I
H
O
B
K

B

M

R

C

I

I
l
l
l

F
l
C
e
A

a
A

P
s
l

A
©

ncreasing Walking
ow Important Is Distance To, Attractiveness, and Size of Public
pen Space?

illie Giles-Corti, PhD, Melissa H. Broomhall, MPH, Matthew Knuiman, PhD, Catherine Collins, MBBS,
ate Douglas, MBBS, Kevin Ng, MBBS, Andrea Lange, BA (Hon), Robert J. Donovan, PhD

ackground: Well-designed public open space (POS) that encourages physical activity is a community
asset that could potentially contribute to the health of local residents.

ethods: In 1995–1996, two studies were conducted—an environmental audit of POS over 2 acres
(n �516) within a 408-km2 area of metropolitan Perth, Western Australia; and personal
interviews with 1803 adults (aged 18 to 59 years) (52.9% response rate). The association
between access to POS and physical activity was examined using three accessibility models
that progressively adjusted for distance to POS, and its attractiveness and size. In 2002, an
observational study examined the influence of attractiveness on the use of POS by
observing users of three pairs of high- and low-quality (based on attractiveness) POS
matched for size and location.

esults: Overall, 28.8% of respondents reported using POS for physical activity. The likelihood of
using POS increased with increasing levels of access, but the effect was greater in the model
that adjusted for distance, attractiveness, and size. After adjustment, those with very good
access to large, attractive POS were 50% more likely to achieve high levels of walking (odds
ratio, 1.50; 95% confidence level, 1.06 –2.13). The observational study showed that after
matching POS for size and location, 70% of POS users observed visited attractive POS.

onclusions: Access to attractive, large POS is associated with higher levels of walking. To increase
walking, thoughtful design (and redesign) of POS is required that creates large, attractive
POS with facilities that encourage active use by multiple users (e.g., walkers, sports
participants, picnickers).
(Am J Prev Med 2005;28(2S2):169 –176) © 2005 American Journal of Preventive Medicine
m
a

r
r
p
i
g
p
t
w
t
e

t
p
P
r
v
m
s

ntroduction

The indescribable innocence and beneficence of
Nature . . . such health, such cheer, they afford!

—Walden, Henry David Thoreau (1817–1842)

n the 19th century, public open space (POS) was
created in the United Kingdom and United States
with a view to improving the health and quality of

ife of the working classes living in squalid and crowded
iving conditions.1–3 Perceived as the “lungs” of pol-
uted cities, POS provided alternative activities for the

rom the School of Population Health, University of Western Austra-
ia (Giles-Corti, Broomhall, Knuiman, Collins, Douglas, Ng, Lange),
rawley, Western Australia, Australia, and Division of Health Sci-
nces, Curtin University of Technology (Donovan), Perth, Western
ustralia, Australia
At the time of the studies discussed here, Robert J. Donovan was

ffiliated with the School of Population Health, University of Western
ustralia, Perth, Western Australia, Australia.
Address correspondence and reprint requests to: Billie Giles-Corti,

hD, Associate Professor, School of Population Health, The Univer-
p
ity of Western Australia, 35 Stirling Highway, Crawley 6009, Austra-
ia. E-mail: billie@cyllene.uwa.edu.au.

m J Prev Med 2005;28(2S2)
2005 American Journal of Preventive Medicine • Published by
asses seen to be slipping into “moral decay,” as well as
place for physical recreation.4,5

Public open space continues to play an important
ole in contemporary society. However, until recent
ecognition of the health benefits of brisk walking,6 its
otential as a community resource for increasing phys-

cal activity has not been the subject of investigation.7 A
rowing body of evidence indicates that a range of
erceived and objectively measured environmental at-
ributes—including access to POS—are associated with
alking.8,9 As yet, however, the characteristics of POS

hat encourage more physical activity have not been
xplored.
Items used to measure usage of POS vary in terms of

ime period, activities, and types of POS studied. This
roduces equally varying estimates of the prevalence of
OS usage. For example, U.S. and Australian parks and
ecreation surveys report that over 70% of those sur-
eyed had visited a park at least once in the previous 12
onths.10,11 However, POS is used for infrequent pas-

ive pursuits (e.g., picnicking) as well as for regular

hysical activity. The prevalence of use for the latter

1690749-3797/05/$–see front matter
Elsevier Inc. doi:10.1016/j.amepre.2004.10.018



p
t
i
1
w
“

e
o
a
o
e
v
g
d
A
s
m
A
f

t
a
o
w
l
a
l
p
i
w
A
f
a
d
f

n
P
c
s
(
s
(
a
a

u
s
i
t
p
l
(
p
a
F

a
p
t
p
i
f
U

t
T
w
a
i
s

M

T
4
E
i
a
p
P

S

T
a
p
s
o
t
(
(
(
o
e
h
a
r
r
s
w
m
s

A

E

A

1

urpose is somewhat lower. In Australia, for example,
he prevalence of adult use of POS for physical activity
n the previous 2 weeks ranges from 13.0%12 to
7.3%.13 This increases to 18% and 23%, respectively,
hen combined with use of undeveloped POS (i.e.,
bushland” or forest).

Park usage varies between, and within, countries. For
xample, a North Carolina study found that only 8.6%
f respondents had used a public park for their physical
ctivity in the previous month.14 Unequal distribution
f POS throughout cities and between countries may
xplain apparent cross-cultural and socioeconomic
ariations in POS usage.15 In some Australian states,
overnment policy has been used to ensure equal
istribution of POS across communities. In Western
ustralia, for example, a 1955 metropolitan plan16

tipulated that 10% of land in new housing develop-
ents be allocated to POS. This may explain why in
ustralia, POS is the third single most popular venue

or physical activity, after the streets and home.12,13

Distance from home to POS also seems to influence
he frequency of use and type of usage (for physical
ctivity or for passive recreation). Two studies of users
f a large urban park in Chicago found that compared
ith other ethnic groups, Caucasian users were more

ikely to visit the park on a daily basis, alone or with
nother person. However, they were also more likely to
ive nearby and to walk, rather than drive, to the
ark.17,18 Non-Caucasian users living farther away vis-

ted the park less frequently, were more likely to visit
ith a family group, and stayed longer once there.
ustralian surveys of users of smaller parks19,20 have

ound that, provided there are no physical barriers
ffecting access (e.g., a major road), distance is a major
eterminant of park use, with most users being drawn
rom within a 500-m radius of the park.

A literature review by Broomhall21 concluded that
umerous observable factors may influence the use of
OS. These include the quality and quantity of space;
haracteristics of potential users (e.g., socioeconomic
tatus, age, gender, and ethnicity); psychological factors
e.g., self-efficacy, perceived barriers) influencing per-
onal preferences; access to competing local facilities
e.g., recreational centers); the match between park
ttributes and needs of local users; park maintenance;
nd perceived safety.

Attributes of POS provide cues about how it is to be
sed, and by whom.22,23 Qualitative24 and quantitative
urveys suggest that factors influencing use of POS
nclude perceived proximity17,24 and accessibility (i.e.,
he absence of major roads)24; aesthetic features of the
ark such as the presence of trees, water (e.g., a

ake),10,17,18,24 and birdlife24,26; park maintenance
e.g., irrigated lawns)17,18,24; park size (which, in turn
rovides variety and opportunities to “lose oneself”)24;
nd the availability of amenities such as walking paths.17
actors that influence park usage for passive recre-

70 American Journal of Preventive Medicine, Volume 28, Num
tional outings, such as picnics (e.g., availability of
icnic tables, barbecues, toilets), are different from
hose that encourage physical activity (e.g., walking
aths).17,18 Although not raised as an important issue

n Australian research,24 perceived safety is another key
actor found important to Caucasian users in the
nited States.17

The evidence to date suggests that users and poten-
ial users prefer proximate, attractive, and larger POS.
hus, the aim of this study was to examine the extent to
hich access to POS is associated with using POS and
chieving recommended levels of physical activity, us-
ng three models of accessibility that adjust progres-
ively for distance to, attractiveness, and size of the POS.

ethods

his paper describes three related studies undertaken in a
08-km2 area of metropolitan Perth as part of the Studies of
nvironmental and Individual Determinants of physical activ-

ty. Two of the studies—an environmental scan of 516 POS
nd a survey of adults aged 18 to 59 years (n �1803)—took
lace in 1995–1996, and the third, an observational study of
OS users (n �772), was carried out in 2002.

tudy 1: Environmental Scan

he Ministry for Planning (MP) provided the name and
ddress of all POS in the study area (n �2500). POS included
arks with and without play equipment, recreational grounds,
ports fields, commons, esplanades, and buffer strips. Based
n qualitative research findings,24,25 the study was restricted
o POS �2 acres (n �516). Inaccessible recreational areas
e.g., sports stadia) were also excluded. The POS Tool
known as the POST) was developed by the second author
MB), using information from focus groups24,25 and a review
f literature. Content validity was assessed by a panel of six
xperts (two community architect and planners, one public
ealth academic, one government expert on sport and recre-
tion and two government experts on planning). Inter-rater
eliability was assessed (n �20),21 and unreliable items were
emoved or modified. The reliability of the instrument was
atisfactory with kappa values ranging from 0.6 to 1.0. Data
ere collected in four domains, including activities, environ-
ental quality or aesthetics, amenities, and safety, as de-

cribed below.

ctivities. Two items related to type of usage (active-formal,
active-informal, and passive), and specific activities for
which the space was designed (e.g., tennis, football,
walking).

nvironmental quality. Fifteen items related to the presence
of features including birdlife; the number and placement
of trees; presence and placement of walking paths, and the
amount and quality of shade along the paths; park contours
(i.e., slope); whether lawns were irrigated; whether dogs
were allowed (leashed or unleashed); and the presence of
graffiti.

menities. Fourteen items related to the presence of chil-
dren’s play equipment, barbecues, picnic tables, parking

facilities, public toilets, public transport within 100 meters,

ber 2S2



S

w
l
a
t
p
(
r
b
p
t
r
t
p
a
r
p
a
s
p

P
P

S

U
w
h
t
4
(
e
s
S
D
i
p
b
c
a
w
a
a
R
i
d
f
2

V
a
a
(
e

D
m
h
w

P
e
m
d
t
i
o
a
o
o

d
s
g
t
a
a
c
l
w

w
o
t
a

w
a
i
e
b
d
d
M
c
(

S
O

T
a
a
P
w
a
a
r
m
(
(
w
a

seating, fencing within park, clubrooms/meeting rooms,
rubbish bins, drinking fountains, a kiosk/cafe, presence
and height of boundary fencing, and availability and
amount of car parking

afety. Four items related to the presence of lighting, visibility
of surrounding houses or roads, type of surrounding roads,
and presence of crossings.

The POST assessed attributes used for active recreation as
ell as passive pursuits (e.g., barbecues). Based on the

iterature review and focus group research,24,25 ten park
ttributes specifically related to participation in physical ac-
ivity were selected for inclusion in a composite score of the
arks in three domains: five environmental quality factors
presence of a water feature, shady trees along walking paths,
eticulated lawns and birdlife, the park being adjacent to the
each or river); three amenity factors (presence of walking
aths, sports facilities, and children’s play equipment); and
wo safety factors (presence of lighting and quiet surrounding
oads). The advice of expert panel members indicated that
hese attributes may not be equally important. Thus, urban
lanners in the 13 local government authorities in the study
rea were approached to form a second expert panel (77%
esponse rate). Based on the importance of each attribute to
articipation in physical activity, the panel was asked to
llocate 100 points across the attributes; and the average
core for each was used as the weight. The weights applied are
ublished elsewhere,27 but also appear in Table 2.
Two observers collected the POST data, visiting 10 to 15

OS per day (n �516). The observers walked through each
OS, checking off each of the items on the POST.

tudy 2: Survey of Residents

sing probability cluster sampling, healthy homemakers and
orkers aged 18 to 59 years were randomly selected from
ouseholds in advantaged and disadvantaged collection dis-

ricts (CDs) (i.e., top and bottom 20th percentile) in a
08-km2 area of metropolitan Perth, Western Australia
n �1803; 52.9% response rate) (referred to later as socio-
conomic status [SES] of area of residence).28 CDs are the
mallest spatial unit defined by the Australian Bureau of
tatistics (ABS) and comprise about 220 households. The
isadvantage Index is derived by the ABS from census

nformation (e.g., income, educational attainment, unem-
loyment, and dwellings without motor vehicles), and was
ased on all households in the CD. To control for potentially
onfounding variables likely to influence engaging in recre-
tional physical activity, ineligible respondents included those
ho were unemployed, aged �59 years, ill or injured, and in
ctive occupations (i.e., three 20-minutes sessions of vigorous
ctivity per week or 1 hour of moderate activity per day).
espondents were interviewed in their homes using a 255-

tem survey that included measures of the frequency and
uration of vigorous and light-to-moderate activity, walking
or recreation, and walking for transportation in the previous

weeks.29

ariables. Four dichotomous dependent variables were ex-
mined: use of POS (defined as use of a POS for physical
ctivity in the previous two weeks); sufficient physical activity
i.e., accumulation of the equivalent of 30 minutes of mod-

rate activity on most days of the week) (see Giles-Corti and s
onovan28 for details); walking as recommended (i.e., five or
ore walking sessions totaling �150 minutes/week); and
igh levels of walking (i.e., six or more sessions of walking/
eek, totaling �180 minutes) (1�Yes, 0�No).
The main independent variable studied was accessibility to

OS. It was based on a gravity model,30 and is described fully
lsewhere.27,28 Geographers conceive of accessibility as a
easure of the spatial distribution of facilities adjusted for the

esire and the ability of people to overcome distance or travel
ime to access a facility or activity.30 Although use of POS is
nversely related to distance, the impact of distance depends
n the attractiveness of the POS (i.e., its attributes), location,
nd the user’s access to transport. The effort required to
vercome distance to use a facility is measured by a distance-
f-decay parameter.
In this study, three models of accessibility were tested: a

istance-only model, which estimated distance from the re-
pondent’s home to all POS in the study area using geo-
raphic information systems software, and which assumed
hat all the POS in the study were equally attractive; a distance
nd attractiveness model that adjusted for distance and the
ttractiveness of the POS, attractiveness being based on a
omposite score derived from the nine weighted items col-
ected using the POST. The attractiveness score for each POS
as estimated as follows:

Att � �
j

Aj * wj

here Att is the attractive score, Aj is a binary indicator (0,1)
f the presence of the jth attribute, and wj is the weight for
he jth attribute. The final model that adjusted for distance,
ttractiveness, and size of the POS was as follows:

Ai � �
j

Attj
�sj

�⁄ d ij
�

here Ai is the accessibility index at origin i, Attj is the
ttractiveness of destination j, sj is the size of destination j, dij
s the distance between origin i and destination j; � is an
stimated destination-specific attractiveness-decay parameter
etween i and j, � is an estimated destination-specific size-
ecay parameter between i and j, and � is an estimated
estination-specific distance-decay parameter between i and j.
ore fully described elsewhere,27,28 destination-specific de-

ay parameters were estimated for distance (�), attractiveness
�), and size (�).

tudy 3: Observations of Public
pen Space Users

he pilot observational study was undertaken by three of the
uthors (KN, KD, CC). The aim was to validate the POST21 to
ssess the impact of the attractiveness of POS, independent of
OS size. Six pairs of POS from the environmental scan study
ere selected, two each from low-, medium-, and high-SES
reas. Each pair was located within the same postal code area,
nd had a POST score differential of 30 points. The study was
estricted to POS �6 hectares in size, and an attempt was
ade to match the size of each pair of low-scoring POS

mean�3.0 ha, range of 1.8 to 4.8 ha) and high-scoring POS
mean�3.3 ha, range of 2.0 to 5.3 ha). An observational tool
as used to record the estimated age and gender of users,
ctivity performed, who the user was with, and total time

pent at the POS. After training observers, the tool was pilot

Am J Prev Med 2005;28(2S2) 171



t
E
0
w
o

S

T
a
S
2
a
t
r
o
r
i

p
2
l
a
e
t
d
s
w
b
q

u

R
D

R
p
a
b
1
p
w
o
i

D

T
t
o
[
(

A
P

R
o
o

s
f
s
w
i
w
a
P
2
t
a

A
a
P

A
n
m
m

A
S
P

A
a
s
l
t
w
t

T

C

A

G

E

S

U
W

W

S

P

1

ested and satisfactory inter-rater reliability was established.
ach pair of POS was monitored on the same Saturday from
730 to1730 hours, with two scheduled breaks. To control for
eather-dependent behavior patterns, observations occurred
nly on days when temperatures ranged from 20°C to 32°C.

tatistical Analysis

he data collected in Study 1 were used to develop the
ccessibility indices variables described for Study 2. Using
PSS, version 11 (SPSS Inc., Chicago), the analysis for Study
was based on 1773 survey respondents. Logistic regression

nalyses were used to examine multivariate associations be-
ween the dependent and independent variables. All models
eported were adjusted for age, gender, education, number
f children aged �18 years at home, and SES of area of
esidence. In one model, use of POS (1�Yes, 0�No) was also
ncluded as an independent variable.

To develop the distance-, attractiveness- and size-decay
arameters used in the accessibility indices described in Study
, a linear regression model was used to separately regress the
og of distance, attractiveness, and size on the log of percent-
ge of opportunities available to access the facilities used. The
xponential coefficients from the linear regressions used as
he decay parameters in subsequent modeling were 1.91 for
istance, 0.52 for attractiveness, and 0.85 for size. The mea-
ures of accessibility developed from the three gravity models
ere re-coded into quartiles with 1�very poor access (i.e.,
ottom quartile of access) and 4�very good access (i.e., top
uartile of access).
Only descriptive analysis of the observational study data was

ndertaken.

esults
escription of Sample

eflecting the sampling method, an almost equal pro-
ortion of respondents were from high and low SES
reas. All age groups were appropriately represented,
ut women were over-represented in the sample (Table
). Overall, 28.8% of respondents had used a POS for
hysical activity in the previous 2 weeks, 23.0% had
alked as recommended, 17.3% reported a high level
f walking, and 59.2% had undertaken sufficient activ-

ty overall.

escription of Public Open Space Attributes

able 2 shows the distribution of POS attributes, and
he weights assigned to each attribute. The average size
f POS in the study area was 6.2 ha (standard deviation
SD]�11.1), and the total average POST score was 47.5
SD�9.3).

ssociation Between Accessibility and Use of
ublic Open Space

egardless of the model used (i.e., a simple distance-
nly model through to the more complex model),

verall use of POS was positively associated with acces- s

72 American Journal of Preventive Medicine, Volume 28, Num
ibility (test for trend p �0.000) (Table 3). Accounting
or attractiveness as well as distance did not produce a
tronger trend with level of access. However, when size
as also taken into account, the odds ratio (OR)

ncreased for those with very good access. Compared
ith those with very poor access, those with very good
ccess to large attractive POS were twice as likely to use
OS (OR�2.05, 95% confidence interval [CI]�1.52–
.75). These results suggest that after distance to POS is
aken into account, size was more important than
ttractiveness in encouraging use.

ssociation Between Use of Public Open Space
nd Achieving Recommended Levels of
hysical Activity

s can be seen in Table 4, those who used POS were
early three times as likely as others to achieve recom-
ended levels of activity, regardless of how it was
easured.

ssociation Between Access to Public Open
pace and Achieving Recommended Levels of
hysical Activity

s shown in Table 5, regardless of which model of
ccessibility was used, the accessibility of POS was not
ignificantly associated with achieving overall sufficient
evels of activity or walking as recommended. However,
hose with very good access to attractive and large POS
ere 50% more likely (OR�1.50, 95% CI�1.06 –2.13)

o achieve high levels of walking, that is, six walking

able 1. Description of sample

haracteristic % (n � 1773)

ge group (years)
18–29 26.2
30–39 28.4
40–49 27.1
50–59 17.2
ender
Male 32.1
Female 67.9

ducation
Subsecondary 21.5
Secondary 23.5
Trade school 5.4
Certificate 22.5
Tertiary 27.0

ES of area of residence
Disadvantaged 48.5
Advantaged 51.5
sed POS for physical activity 28.8
alking five sessions/week totaling >150

minutes 23.0
alked six sessions/week totaling >180

minutes 17.3
ufficiently active 59.2

OS, public open space; SES, socioeconomic status.
essions/week, totaling �180 minutes.

ber 2S2



O

T
P
d
O
S
j
o
u
a
h
s
u
i
s
m
i

D

A
m
l

u
C

l
l
f

m
c
w
t
W
o
d
(
i
e
s
r
f
r
n
t
g
i
e
p
r
e
r
a
s
f

T

A

S

L
W
S

A
W
Q

L

B
T

A
a

b

c

P

T
a

T

D

D

D

a

h
b

bservational Study Results

he observational study was designed to validate the
OST by examining whether parks of equal size but
ifferential POST scores attracted more or less users.
verall, 772 people were observed using the POS.

ixty-four percent of those observed were walking or
ogging, 12% were cycling, and 5% were engaging in
rganized sports. However, 70% of those observed were
sing high-scoring POS. Furthermore, 70% of walkers
nd joggers and 75% of cyclists observed were using
igh-scoring POS. All of those engaged in organized
ports were in low-scoring POS. A total of 18.4% of POS
sers were engaged in passive pursuits such as picnick-

ng, and 82.3% of passive users were visiting high-
coring POS. This suggests that high-scoring POS were
ore likely to attract walkers, joggers, and those seek-

ng passive pursuits.

iscussion

ccess to proximate and large POS with attributes that
ake them attractive appears to encourage higher

evels of walking.
Having a proximate POS is important because POS

se is sensitive to distance.28 Tinsley et al.17 found that

able 2. Description of POS attributes

ttributes (n � 516)
Weight
assigneda

hade along paths (%)
Very good 1.9 16.90
Good 3.1 13.52
Medium 7.0 10.14
Poor 11.0 6.76
Very poor 11.0 3.38
No paths 65.9 0.00

awns irrigated (%)b 63.2 15.30
alking paths present (%)b 34.1 13.90

porting facilities present
(%)b 46.7 13.30

djacent ocean or river (%)b 9.6 13.10
ater feature present (%)b 13.0 8.30
uiet surrounding roads

(i.e., cul de sac or minor
road only)b 54.5 8.00

ighting present (%)
Along paths 4.8 6.80
In some areas 23.6 5.10
In barbecue/play

equipment areas only
3.3 3.40

No lighting 68.2 0.00
irdlife present (%) 10.9 3.80
otal average score for

parks/100 47.5 (SD � 9.3)
verage size of POS (ha)c 6.2 (SD � 11.1)

Weights assigned based on the presence of each attribute.
If attribute not present, weight � 0.
Excludes two outliers.
OS, public open space; SD, standard deviation.
aucasian users of a large, attractive urban park lived
q
C

ocally and walked daily, while non-Caucasian users who
ived farther away visited the park infrequently as a
amily and for passive recreational pursuits.

However, these results suggest that although proxi-
ate parks encourage use generally, having good ac-

ess to larger POS is associated with higher levels of
alking. Larger parks tend to have more attributes21

hat provide more satisfying experiences for the user.
hen asked about factors that they liked about POS17

r that influenced use for physical activity,24 respon-
ents described trees, water features, bird life, and size,
which provided opportunities to “lose oneself”). This
s consistent with Kaplan and Kaplan’s31 hypothesis that
xposure to nature— even in local parks— can be “re-
torative.”31–33 Natural environments are said to be
estorative when they give users a sense of being away
rom their usual setting, and a sense of fascination
esulting from exposure to (for example) birdlife or
atural beauty.31 Exposure to restorative environments

hat provide satisfying experiences may encourage
reater use and help maintain regular walking behav-
or. A small experimental study of runners and walk-
rs34 randomized to either using the streets or urban
arks for their physical activity, found that those who
an or walked through urban parks perceived the
xperience as more restorative. The respondents also
eported higher ratings of happiness, lower anger/
ggression scores34 or anxiety/depression/anger
cores,35 and had lower levels of postexercise mental
atigue.34

able 3. Logistic regression associating use of POS to
ccess to POS

ype of model
Adjusted
odds ratiosa 95% CI

istance-only model
Very poor accessb 1.00
Poor access 1.28 0.94–1.76
Good access 1.87 1.38–2.53
Very good access 1.87 1.37–2.54
Test for trend p � 0.000
istance and attractiveness

model
Very poor access 1.00
Poor access 1.03 0.76–1.41
Good access 1.67 1.23–2.25
Very good access 1.62 1.20–2.19
Test for trend p � 0.000
istance, attractiveness, and

size model
Very poor access 1.00
Poor access 0.90 0.65–1.23
Good access 1.20 0.88–1.64
Very good access 2.05 1.52–2.75
Test for trend p � 0.000

Adjusted for age, gender, education, children aged �18 years at
ome and socioeconomic status of area of residence.
Very poor access � bottom quartile of access; very good access � top

uartile of access.
I, confidence interval; POS, public open space.

Am J Prev Med 2005;28(2S2) 173



n
e
m
t
o
s
a
s
t
e
b
c
i
a

I

W
c
n
b

w
u
s
v
a
m
u
u

A
w
i
a
t
u
fl
e
fi
o
o
s
a
r
u
s
t
u
o

f
t
a
b
v
g

T
a

T

O
F

S

a

h
C

T

T
o

O

F

S

a

b

C

1

This study found that the impact of POS attractive-
ess on park use and higher levels of walking was
quivocal without the inclusion of park size in the
odel. However, larger POS generally have more at-

ributes that make them attractive.21 In addition, the
bservational study (Study 3), which controlled for POS
ize and compared high- and low-quality POS, found
dditional support for the hypothesis that even in
maller POS of equivalent size, POS with more at-
ributes attract more users. Thus, in the main study, the
quivocal results related to attractiveness may have
een due to the selection of attributes used in the
omposite score, the assignment of weights, or the
nclusion of the attractiveness-decay parameter in the
ccessibility model.

mplications for Research and Practice

ell-designed public open spaces are an important
omponent of the recreational mix providing opportu-
ities for physical activity and social interaction. It may
e possible to attract more users to POS by creating

able 4. Logistic regression associating use of POS to
chieving recommended levels of physical activity

ype of model
Adjusted
odds ratiosa 95% CI

verall levels of sufficient activity 2.66 2.10–3.37
ive or more walking sessions/
week totaling �150 minutes 2.78 2.19–3.54

ix or more walking
sessions/week totaling �180
minutes 2.82 2.17–3.67

Adjusted for age, gender, education, children aged �18 years at
ome and socioeconomic status of area of residence.
I, confidence interval; POS, public open space.

able 5. Logistic regressions associating different levels of ph

ype of behavior and level
f accessb

Distance-only
model
OR (95% CI)

verall sufficient physical activity
Very poor access to POS 1.00
Poor access to POS 0.69 (0.52–0.92)
Good access to POS 0.89 (0.67–1.17)
Very good access to POS 0.87 (0.66–1.15)

ive or more walking sessions/week totaling >150 minutes
Very poor access to POS 1.00
Poor access to POS 1.01 (0.73–1.41)
Good access to POS 1.04 (0.75–1.44)
Very good access to POS 1.20 (0.87–1.65)

ix or more walking sessions/week totaling >180 minutes
Very poor access to POS 1.00
Poor access to POS 1.02 (0.70–1.48)
Good access to POS 1.19 (0.83–1.71)
Very good access to POS 1.14 (0.79–1.65)

Adjusted for age, gender, education, children aged �18 years at ho

Very poor access � bottom quartile of access; very good access � top qua
I, confidence interval; OR, odds ratios; POS, public open space.

74 American Journal of Preventive Medicine, Volume 28, Num
alking trails that link smaller local parks through the
se of signage, developing shaded walking paths land-
caped with trees and shrubs selected to maximize
isibility,37 creating interest by developing undulating
reas around the perimeter of flat POS, and better
aintenance and care of the POS. Encouraging more

se will have a synergistic effect by attracting even more
sers and thereby making the POS safer.37

Redesigning existing space is also important. An
ustralian study found that despite the popularity of
alking, a disproportionate amount of community POS

s zoned for organized sports (i.e., playing fields known
s “ovals” in Australia) rather than for informal activi-
ies such as walking or jogging.36 Playing fields are
sually characterized by being well irrigated, green, and
at, and thus, insufficiently interesting to attract walk-
rs. When not being used for organized sports, playing
elds are usually under-utilized and mainly used for
ccasional informal ball sports by children or by dog
wners exercising their dogs.24 The small observational
tudy confirmed that fewer people use POS with fewer
ttributes. With thoughtful design, it is possible to
edesign playing fields with public access for multiple
sers— organized sports participants, walkers, and pas-
ive recreational users—thereby making better use of
his important community resource.36 Similarly, greater
se could be made of school playing fields, which are
ften not used during out-of-school hours.
Despite a number of limitations and the need for

urther development, gravity models may be useful
ools for physical activity research. In attempting to
djust for attractiveness and size, this study tried to go
eyond simply thinking about distance as the only
ariable that encourages use of a destination. As sug-
ested by Handy and Neimeier,38 it is also important to

l activity to accessibility of POSa

Distance and
attractiveness
model
OR (95% CI)

Distance,
attractiveness,
and size model
OR (95% CI)

1.00 1.00
0.71 (0.54–0.94) 0.82 (0.62–1.09)
0.90 (0.68–1.19) 0.73 (0.55–0.96)
0.87 (0.66–1.16) 0.91 (0.68–1.20)

1.00 1.00
0.98 (0.70–1.36) 0.68 (0.48–0.95)
1.19 (0.86–1.65) 0.96 (0.69–1.32)
1.23 (0.89–1.69) 1.24 (0.91–1.70)

1.00 1.00
1.05 (0.72–1.53) 0.73 (0.50–1.08)
1.27 (0.88–1.82) 1.11 (0.77–1.59)
1.24 (0.86–1.79) 1.50 (1.06–2.13)

d socioeconomic status of area of residence.
ysica

me an

rtile of access.

ber 2S2



c
a
a
i
c
“
r
o
t
a
o
c
b
f
i
u
a
a
m
t

L

W
A
h
s
w
c
r
w
r
e
F
m
a

C

T
c
l
S
t
t
r
m
p

T
P
d
a
b
2

o

R

1

1

1
1

1

1

1

1

1

1
2

2

2

2

2

2

2

2

onsider the quality of destinations and how this might
ffect use. The approach taken in this study could be
pplied to other destinations (e.g., access to shops)
mportant for walking. Applying a gravity model over-
omes the problem of having to define a specific
neighborhood.”39,40 For example, in this study, all
espondents had access to all destinations (regardless
f where they were located). However, by incorporating
he distance-of-decay parameter, destinations farther
way had little impact on access. This approach also
vercomes concerns about the “ecologic fallacy,” be-
ause the exposure variables were linked to individual
ehavioral outcomes,41 while allowing for adjustment
or confounding factors. Future research could exam-
ne the specific attributes that make parks attractive to
sers, more sophisticated methods of weighting park
ttributes before deriving an overall score, different
pproaches to using attractiveness factors in gravity
odels, and the interaction between factors such dis-

ance, size, and attractiveness.

imitations

ith a population of about 1.2 million, Perth is one of
ustralia’s smaller capital cities, and enjoys a relatively
igh standard of living by national and international
tandards.28 Due to limited resources, a study area
ithin Perth was selected, and to control for potential
onfounding variables, those who might have some
eason not to engage in recreational physical activity
ere excluded. In addition, the sample was limited to
esidents of socially advantaged and disadvantaged ar-
as. These factors may limit the study’s generalizability.
inally, the approach to weighting the attributes that
ake POS attractive may have resulted in the results on

ttractiveness being equivocal.

onclusions

his study confirmed that POS is an important
ommunity resource. Good access to attractive and
arge POS is associated with higher levels of walking.
imply providing proximate POS appears insufficient
o increase walking: Consideration needs to be given
o its size and attributes that make it attractive. More
esearch is required to understand the attributes that
ake POS attractive and which encourage more

hysical activity.

his research was funded by the Western Australian Health
romotion Foundation (Healthway). Kathryn Boyd and An-
reana Kursar, who assisted with auditing, are gratefully
cknowledged. The first author (BG-C) is currently supported
y a NHMRC/NHF Career Development Award (grant

54688). 2
No financial conflict of interest was reported by the authors
f this paper.

eferences
1. MacMaster N. The battle for Mousehold Heath 1857–1884: popular politics

and the Victorian public park. Past and Present, 1990, pp. 117–54.
2. Arnold HF. Trees in urban design. 2nd ed. New York: Van Norstrand

Reinhold, 1993.
3. Percifull E, Thomas S, Kendle T. Multi-cultural parks. Landscape Design,

1993, pp. 9 –12.
4. Solecki W, Welch J. Urban parks: green spaces or green walls? Landscape

Urban Planning 1995;32:93–106.
5. Maller C, et al. Healthy parks, healthy people: the health benefits of contact

with nature in a park context. Melbourne: Deakin University, 2002.
6. U.S. Department of Health and Human Services. Physical activity and

health: a report of the Surgeon General. Atlanta GA: U.S. Department of
Health and Human Services, Centers for Disease Control and Prevention,
National Center for Chronic Disease Prevention and Health Promotion,
1996.

7. Sallis J, Bauman A, Pratt M. Environmental and policy interventions to
promote physical activity. Am J Prev Med 1998;15:379 –97.

8. Owen N, et al. Understanding environmental influences on walking. Am J
Prev Med 2004;27:67–76.

9. McCormack G, et al. An update of recent evidence of the relationship
between objective and self-report measures of the physical environment
and physical activity behaviours. J Sci Med Sport 2004;7:81–92.

0. Veal T. Using Sydney’s parks. Parks and Leisure Australia, September
21–23, 2001.

1. Payne L, Mowen A, Orsega-Smith E. An examination of park preferences
and behaviors among urban residents: the role of residential location, race
and age. Leisure Sci 2002;24:81–98.

2. DASETT. Physical activity levels of Australians. Canberra: AGPS, 1988.
3. McCormack G, et al. Physical activity levels of Western Australian adults

2002: results from the adult physical activity survey. Perth, Western Austra-
lia: Government Printing Office, 2003.

4. Huston S, et al. Neighbourhood environment, access to places for activity,
and leisure-time physical activity in a diverse North Carolina population.
Am J Health Promotion 2003;18:58 – 69.

5. King AC, et al. Environmental and policy approaches to cardiovascular
disease prevention through physical activity: issues and opportunities.
Health Educ Q 1995;22:499 –511.

6. Stephenson G, Hepburn J. Plan for the metropolitan region Perth and
Fremantle. Perth, Western Australia: Government Printing Office, 1955.

7. Tinsley H, Tinsley D, Croskeys C. Park usage, social milieu and psychosocial
benefits of park use reported by older urban park users from four ethnic
groups. Leisure Sci 2002;24:199 –218.

8. Gobster P. Managing urban parks for a racially and ethnically diverse
clientele. Leisure Sci 2002;24:143–59.

9. Boyle R. Survey of the use of small parks. Aust Parks Recreation 1983;:41–3.
0. Just D. Appropriate amounts and design of open spaces. Aust Parks

Recreation 1989;25:32–9.
1. Broomhall MH. Study of the availability and environmental quality of

urban open space used for physical activity. Master’s thesis, Department of
Public Health, University of Western Australia, Perth, 1996.

2. Rutledge AJ. Anatomy of a park: the essentials of recreation area planning
and design. New York: McGraw-Hill Book Company, 1971.

3. Whyte WH. The social life of small urban spaces. Washington DC:
Conservation Foundation, 1980.

4. Corti B, Donovan RJ, Holman CDJ. Factors influencing the use of physical
activity facilities: results from qualitative research. Health Promotion J Aust
1996;6:16 –21.

5. Corti B, Donovan RJ, Holman CDJ, Shilton TR. Encouraging the sedentary
to be active every day: qualitative formative research. Health Promotion J
Aust 1995;5:10 –7.

6. Raymore LD, Scott D. The characteristics and activities of older adult
visitors to a metropolitan park district. J Park Recreation Admin
1998;16:1–21.

7. Giles-Corti B, Donovan R. Increasing walking: the relative influence of
individual, social environmental and physical environmental factors. Am J
Public Health 2003;93:1583–9.
8. Giles-Corti B, Donovan R. The relative influence of individual, social and

Am J Prev Med 2005;28(2S2) 175



2

3

3

3

3

3

3

3

3

3

3

4

1

physical environmental determinants of physical activity. Social Sci Med
2002;54:1793– 812.

9. Risk Factor Prevalence Study Management Committee. Risk Factor Preva-
lence Study Survey No 3 1989. Canberra: National Heart Foundation of
Australia and Australian Institute of Health, 1990.

0. Hansen WG. How accessibility shapes land use. J Am Inst Planners
1959;15:73– 6.

1. Kaplan R, Kaplan S. The experience of nature: a psychological perspective.
New York: Cambridge University Press, 1989.

2. Kaplan S. The restorative benefits of nature: toward an integrative frame-
work. J Environ Psych 1995;15:169 – 82.

3. Herzog T, et al. Reflection and attentional recovery as distinctive benefits
of restorative environments. J Environ Psych 1997;17:176 – 80.
4. Hartig T, Mang M, Evans G. Restorative effects of natural environment
experiences. Environ Behav 1991;23:3–26.

4

76 American Journal of Preventive Medicine, Volume 28, Num
5. Bodin M, Hartig T. Does the outdoor environment matter for pyschological
restoration gained through running? Psychol Sport Exerc 2003;4:141–53.

6. Hahn A, Craythorn E. Inactivity and the physical environment in two
regional centres. Health Promotion J Aust 1994;4:43–5.

7. Kuo F, Sullivan. Environment and crime in the inner city: does vegetation
reduce crime? Environ Behav 2001;33:343– 67.

8. Handy S, Neimeier D. Measuring accessibility: an exploration of issues and
alternatives. Environ Planning 1997;29:1175–94.

9. Pikora T, et al. Developing a reliable audit instrument to measure the
physical environment for physical activity. Am J Prev Med 2002;23:187–94.

0. Pikora T, et al. Developing a framework for assessment of the environ-
mental determinants of walking and cycling. Social Sci Med 2003;56:
1693–793.
1. Hennekens C, Buring J. Epidemiology in medicine. Boston: Little, Brown
and Company, 1987.

ber 2S2


	Increasing Walking
	Introduction
	Methods
	Study 1: Environmental Scan
	Activities.
	Environmental quality.
	Amenities.
	Safety.

	Study 2: Survey of Residents
	Variables.

	Study 3: Observations of Public Open Space Users
	Statistical Analysis


	Results
	Description of Sample
	Description of Public Open Space Attributes
	Association Between Accessibility and Use of Public Open Space
	Association Between Use of Public Open Space and Achieving Recommended Levels of Physical Activity
	Association Between Access to Public Open Space and Achieving Recommended Levels of Physical Activity
	Observational Study Results

	Discussion
	Implications for Research and Practice
	Limitations

	Conclusions
	Acknowledgment
	References