Why some walk and others don't: exploring interactions of perceived safety and social neighborhood factors with psychosocial cognitions
Why some walk and others don't: exploring interactions of perceived safety and social neighborhood factors with psychosocial cognitions
Marie¨lle A. Beenackers 0 1
Carlijn B. M. Kamphuis 0 1
Johan P. Mackenbach 0 1
Alex Burdorf 0 1
Frank J. van Lenthe 0 1
0 The Author 2013. Published by Oxford University Press. All rights reserved. Downloaded from https://academic.oup.com/her/article-abstract/28/2/220/595870 by guest on 11
1 Department of Public Health, Erasmus University Medical Center , Rotterdam , The Netherlands
Although physical activity is often believed to be influenced by both environmental and individual factors, little is known about their interaction. This study explores interactions of perceived safety and social neighborhood factors with psychosocial cognitions for leisure-time walking. Cross-sectional data were obtained from residents (age 25-75 years) of 212 neighborhoods in the South-East of the Netherlands, who participated in the Dutch GLOBE study in 2004 (N ¼ 4395, survey response 64.4%). Direct associations of, and interactions between perceived neighborhood safety, social neighborhood factors (social cohesion, social network and feeling at home) and psychosocial cognitions (attitude, self-efficacy, social influence and intention) on two outcomes of leisure-time walking [yes versus no (binary), and among walkers: minutes per week (continuous)] were analyzed in multilevel regression models. The association between attitude and participating in leisure-time walking was stronger in those who felt less at home in their neighborhood. Social influence and attitude were stronger associated with participation in leisure-time walking in those who sometimes felt unsafe in their neighborhood. A positive intention was associated with more minutes walked in those who perceived their neighborhood as unsafe among those who walked. Only limited support was found for interactions between neighborhood perceptions and psychosocial cognitions for leisure-time walking.
Physical inactivity is among the most important and
prevalent risk factors of many major diseases [
Understanding why people are physically inactive is
therefore of key importance in developing strategies
to reduce these major diseases. Walking is a
relatively easy way to be physically active; it is
accessible to most people because it does not require any
financial means and it can be continued into old age.
Known determinants of walking are individual
psychosocial cognitions, such as attitude and
]. In the past decade, many studies
also investigated possible environmental
determinants of walking, such as safety, population density
and access to facilities [
Thus far, many studies have looked at the relation
between psychosocial cognitions and environmental
factors with walking separately or have explored to
what extent psychosocial cognitions mediated the
influence of environmental factors on walking
]. However, a social-ecological perspective
suggests that there is an interplay between the
individual and the environment. According to Emmons
], improving the understanding of health
behaviors in their social context implies that the role of
individual factors for health behaviors may depend
on the environmental context. One of the core
principles of ecological models is that influences
interact across levels [
]. So, although such
interactions are implied in ecological models [
these models do not provide specific hypotheses,
and perhaps as a consequence, empirical studies
into interaction effects are still scarce.
Few studies that did investigate
environment–individual interactions for walking have mainly
focused on built environmental factors including
connectivity of streets, availability of shopping
and sports facilities and neighborhood aesthetics
]. Other factors, such as social environmental
] and safety concerns [
also suggested to be of importance for walking
behavior. Rhodes et al. [
] studied the interactions
between safety and psychosocial cognitions with
respect to walking behavior and found that low levels
of perceived crime resulted in a larger influence of
attitude on the intention to walk compared with
people who perceived high levels of crime. To
date, there are no studies that have looked at
interactions of psychosocial cognitions with social
neighborhood factors such as social cohesion and
social network for walking. Therefore, it is the aim
of this article to explore interactions of safety and
social neighborhood perceptions (neighborhood
social cohesion, neighborhood social network and
feeling at home within your neighborhood) with
psychosocial cognitions (attitude, self-efficacy,
intention and social influence) for leisure-time
In general, two possible interaction mechanisms
can be at play. The first mechanism proposes that the
environment is less important for the decision to
walk for those who have more positive psychosocial
cognitions toward physical activity. When this
interaction exists, people with less positive psychosocial
cognitions would benefit more from a supportive
environment. The other mechanism assumes a
synergy between environmental factors and
psychosocial cognitions; the environment is more important
in the decision to walk for people with more positive
cognitions. This means that the beneficial effects of
having positive psychosocial cognitions and living
in a stimulating environment on walking would
strengthen each other. For example, among those
who report to have a small social network in their
neighborhood, one may expect that having a
positive intention toward physical activity results in less
walking than among those with a large social
network, as having a smaller social network may be a
barrier to putting one’s positive intentions into
action. The aim of this article is to investigate
interactions of perceived safety and social neighborhood
perceptions (neighborhood social cohesion,
neighborhood social network and feeling at home within
your neighborhood) with psychosocial cognitions
(attitude, self-efficacy, intention and social
influence) for two outcomes of leisure-time walking;
any versus no leisure-time walking and among
walkers: minutes per week spent on leisure-time
Materials and methods
Data for this study were collected among a stratified
sample of the adult population of the city of
Eindhoven and its surrounding municipalities in the
Netherlands in 2004, as part of the Dutch GLOBE
study. The baseline sample was stratified by age,
degree of urbanization and socioeconomic status
(SES). More detailed information on the objectives,
study design and data collection of the Dutch
GLOBE study can be found elsewhere [
short, the study started with a baseline survey in
1991. This baseline sample was stratified by age,
degree of urbanization and SES. In 2004, a new
subsample was added to the original cohort to restore
population representativeness of the study sample.
In this study, questionnaires from the cross-sectional
sample of the fourth wave (October 2004) were used
(N ¼ 4785, response 64.4%). The fourth wave was
chosen because of its particular focus on
neighborhood factors. The use of personal data in the GLOBE
study is in compliance with the Dutch Personal Data
Protection Act and the Municipal Database Act and
has been registered with the Dutch Data Protection
Authority (number 1248943).
Respondents with a missing outcome (n ¼ 182) or
who had more than 25% missing values on the
variables used in the analyses (n ¼ 149) were
omitted from the analyses. Respondents with a
missing neighborhood identifier (n ¼ 59) were also
excluded. Thus, a total of 4395 respondents were
included. Remaining missing values were imputed
(see Statistical Analyses section). The respondents
resided in 212 administrative neighborhoods of
Eindhoven and its surrounding municipalities
(mean number of respondents per neighborhood
n ¼ 21, interquartile range ¼ 6–27).
Leisure-time walking was assessed by the
SQUASH, a validated Dutch questionnaire that
measures different types of physical activity [
Within SQUASH, leisure-time walking (i.e.
walking for recreational purposes, no transportation
walking) was measured by asking the respondent
how many days they walked during leisure-time in
a usual week (frequency) and how much time they
spend on this on those days (duration). Because
many respondents did not walk at all during leisure
time, the first outcome variable we analyzed was
binary, namely any versus no leisure-time walking
(‘yes, does walk during leisure-time’ versus ‘no,
does not walk during leisure-time’). For those who
indicated to do any leisure-time walking, total
minutes of leisure-time walking per week were
calculated using information on frequency and duration.
Psychosocial cognitions were based on the Theory
of Planned Behavior [
] and the Social Cognitive
]. All items were formulated toward
‘sufficient physical activity in line with recommended
]. Attitude was measured with 11 items
(Cronbach’s alpha ¼ 0.79) with a 5-point ordinal
answering scale (1, very important to 5, not
important at all). An example question was whether
respondents found the argument ‘it takes too much
time’ important in their decision to be sufficiently
active. Self-efficacy was measured with two items
(Cronbach’s alpha ¼ 0.77). The first item asked
whether respondents thought it was easy or difficult
to be sufficiently physically active (1, very difficult
to 5, very easy). The second item asked how sure
they could be sufficiently physically active when
they would want to (1, not sure at all to 5, very
sure). Intention was measured with one item
(‘do you plan to be sufficiently physically active?’;
1, no, not sure at all to 5, yes, for sure). Social
influence was measured with three items (Cronbach’s
alpha ¼ 0.73) that addressed whether persons
important to the respondent would (i) think the
respondent should be sufficiently active, (ii) stimulate
the respondent to be physically active and (iii) are
sufficiently active themselves. Answering
categories ranged from 1, ‘not true’ to 3, ‘yes, true’.
For all psychosocial cognitions (except intention),
a mean score was calculated from the relevant items
within each cognition. A higher score on each scale
represented a more positive cognition. All items
used to construct the scales can be found in
Supplementary Table S1.
Elements of the neighborhood social environment
were measured using a 13-item scale (Cronbach’s
alpha ¼ 0.87). All items were measured on a 5-point
ordinal scale (1, totally disagree to 5, totally agree).
A principal component analyses with Varimax
rotation and Kaiser Normalization distinguished three
factors. The first factor was labeled ‘social
cohesion’, defined as ‘the extend of connectedness and
solidarity among groups in society’ [
]. An item
that had a high factor loading on this factor was
‘most people in this neighborhood can be trusted’.
The second factor was labeled ‘social network’,
defined as ‘the presence and nature of interpersonal
relationships and interactions; extend to which one
is interconnected and embedded in a community’
]. An item stat had a high factor loading on this
factor was ‘I often visit my neighbors in their home’.
The third factor was labeled ‘feeling at home in this
neighborhood’. An item that had a high factor
loading on this factor was ‘I move out of this
neighborhood if I get the chance (recoded)’. For all three
factors, a standardized factor score (mean ¼ 0,
standard deviation of 1) was constructed using the
factor loadings. The individual social neighborhood
items, their means and standard deviations and the
factor loadings can be found in Supplementary
Perceived safety of the neighborhood was assessed
with four items. The first three items assessed
people’s fear of being home alone or of going out on the
streets in their neighborhood in the daytime or at
night. The items were dichotomized into ‘no,
never feeling afraid’ (0) and ‘neutral/yes, sometimes
feeling afraid’ (1). The fourth item asked the
respondents whether they thought their neighborhood
was unsafe (no ¼ 0, yes ¼ 1). These four
dichotomous items were summed up to form a scale
(Cronbach’s alpha ¼ 0.68). Respondents who did
not agree with any of the items indicating an
unsafe neighborhood were regarded as ‘high’ on
perceived neighborhood safety; they felt safe.
Respondents who agreed once or twice to a measure
indicating an unsafe neighborhood were considered
‘medium’ on perceived neighborhood safety; they
sometimes felt unsafe. Respondents who agreed to
three or four of the items indicative of an unsafe
surrounding were considered ‘low’ on perceived
neighborhood safety; they often felt unsafe.
Potential confounders included were gender, age,
country of origin (the Netherlands and other
country) and educational level ((i) no education or
primary education; (ii) lower professional and
intermediate general education; (iii) intermediate
professional and higher general education; (iv) higher
professional education and university or (v)
missing). Educational level was included as an indicator
for SES and has proven to be a good measure for
SES in the Netherlands [
Overall, missing values of questionnaire items
varied from <1 to 3% per item, with only intention
having 7% missing values. Because complete case
analyses would result in a loss of 25% of the
respondents, missing values for the predictors
were imputed using the Expectation Maximization
] from PASW version 18.0. All the
variables described in the method (psychosocial
cognitions, neighborhood perceptions, demographics and
leisure-time walking) were used in the imputation
Weighted multilevel logistic regression (for
participation in leisure-time walking) and linear
regression (for total minutes walked in a usual week,
within those who walked) models were used to
explore the associations between the predictors and
leisure-time walking of respondents (Level 1)
nested within neighborhoods (Level 2).
Associations among all neighbourhood predictors and
between the neighbourhood predictors and the
psychosocial cognitions were at best modest
(correlation coefficients <0.3). Associations between the
psychosocial cognitions were as expected somewhat
higher (correlation coefficients 0.1; 0.5). Although
multicollinearity is not expected to be a problem
because of these modest correlations, all continuous
variables were mean centered to prevent
multicollinearity in the interaction models and to ease
interpretation. All models were weighted (Level 1
weight) to reflect the source population in terms of
gender, age and educational level. Model 1
contained all neighborhood perceptions. Model 2
contained all psychosocial cognitions. Model 3
combined neighborhood perceptions with
psychosocial cognitions. Subsequently, interactions were
explored whereby each neighborhood–individual
interaction term was added separately to Model 3
(Model 4a–p). Interactions in a logistic regression
model are tested for their departure from
multiplicativity (the combined ‘effect’ of the two factors is
larger or smaller that the product of the individual
‘effects’). Interactions in a linear regression model
are tested for their departure from additivity
(the combined ‘effect’ of the two factors is larger
or smaller that the sum of the individual ‘effects’).
Because additive interactions are considered more
intuitive and more relevant to public health [
to increase comparability of the results for the two
outcomes, the relative risk due to interaction
(RERI), a measure to quantify interaction on an
additive scale, was also calculated for all
interactions departing from multiplicativity [
The RERI is a measure of interaction between two
parameters with a value further away from zero
indicating a stronger interaction. The tool created
by Knol and coworkers [
] was used to
calculate the RERI and the accompanying 95%
confidence interval (CI).
All multivariable models were adjusted for age,
gender, educational level and country of origin.
Significance was interpreted by using the 95% CI.
All regression analyses were carried out in STATA
12 using GLLAMM [
] for the logistic regression
analysis to study participation in leisure-time
walking and using XTMIXED to study the amount of
leisure-time walking within those who walked.
Significant interactions have been visualized by
simple slope analyses.
Table I shows the characteristics of the sample.
Approximately one-third (32.7%) of the respondents
reported no leisure-time walking at all. Those who
did walk spent on average 212 min per week on
leisure-time walking. Crude analyses as presented
in Table II show that females, higher educated and
older respondents were more likely to participate in
leisure-time walking. Among the walkers, minutes
spent per week on leisure-time walking increased
with age, but decreased with educational level.
Crude analyses also showed that a positive
attitude [odds ratio (OR) 1.67, 95% CI 1.42–1.95], a
strong self-efficacy (OR 1.20, 95% CI 1.11–1.29), a
positive social influence (OR 1.39, 95% CI 1.22–
1.57) and a strong intention toward physical activity
(OR 1.37, 95% CI 1.27–1.47) were positively
associated with participating in leisure-time walking
(Table II). Those with a larger social network in
the neighborhood (OR 1.16, 95% CI 1.08–1.24)
were also more likely to walk in leisure time. A
positive attitude (b 33.77, 95% CI 14.19–53.34),
strong self-efficacy (b 39.12, 95% CI 29.05–49.20)
and a positive intention toward physical activity
(b 16.06, 95% CI 7.40–24.73) were also associated
with more walking in those who walked during
leisure time (Table II). None of the neighbourhood
perceptions were significantly associated with minutes
Adjusted for potential demographic confounders
and the other neighborhood perceptions, individuals
with a larger social network (OR 1.14, 95% CI
1.07–1.22) were more likely to engage in walking
in leisure time (Model 1, Table III). The association
remained significant after additional adjustment for
the psychosocial cognitions (Model 3, Table III). Of
the psychosocial cognitions, all but self-efficacy
remained a significant predictor of leisure-time
walking after adjusting for the potential demographic
confounders and the other psychosocial cognitions
(Model 2, Table III). After additional adjustment for
the neighborhood perceptions, the associations
between social influence and leisure-time walking
were no longer significant [although there was
only little change in the point estimate (Model 3,
Table III)], whereas attitude and intention remained
In those who walked during leisure time, a strong
self-efficacy was associated with longer total
duration of walking during leisure time, also in the
fully adjusted model (b 38.31, 95% CI 27.37–
49.25) (Model 3, Table IV). In Model 3, there was
also a significant inverse association between
perceived social cohesion in the neighbourhood and
minutes walked (b 11.69, 95%CI 21.00 to
2.38]) (Model 3, Table IV).
Additional inclusion of the interaction terms
resulted in three significant interactions for
participation in leisure-time walking in the regression
models. The calculated RERIs basically followed
the results of the multiplicative interactions. Safety
interacted significantly with both attitude and social
influence. The association between attitude and
participation in leisure-time walking in people who
sometimes felt unsafe was 1.59 times as high
compared with those who never felt unsafe (95% CI
1.10–2.31) (as visualized in Fig. 1). This pattern
was not observed for those who often felt unsafe
aThe numbers and percentages presented are unweighted and
are therefore a representation of the actual numbers in the
dataset. bPercentages are presented, unless otherwise stated.
cSocial neighbourhood factors (‘social cohesion’, ‘social
network’ and ‘feeling at home’) were not included in this table
because they were standardized factor scores (mean ¼ 0,
standard deviation of 1). The mean and standard deviations for the
individual items that were used to construct the factor scores
can be found in Supplementary Table S2.
in their neighborhood (OR 1.04, 95% CI 0.59–
1.83]). The interaction between social influence
and safety was similar in such a way that the
association between social influence and participation in
leisure-time walking was 1.36 as high in those who
felt sometimes unsafe compared with those who
never felt unsafe in their neighborhood (as
visualized in Fig. 2). This pattern was not observed for
those who often felt unsafe (OR 0.85, 95% CI 0.53–
1.35). The third interaction was between feeling at
home in your neighborhood and attitude (OR 0.87,
95% CI 0.75–1.00); among those feeling more at
home, attitude had a weaker association with
participation in leisure-time walking, than among those
feeling less at home in their neighborhood
(visualized in Fig. 3).
Among those who walked during leisure time,
one significant interaction was observed for total
minutes walked per week. In those who felt
unsafe, a positive intention was associated with
over 30 min more walking during leisure time
compared with those who did not feel unsafe and had a
positive intention toward physical activity. This
interaction has been visualized in Fig. 4.
This study is among the first to evaluate interactions
between elements of the social environmental and
safety in neighborhoods and psychosocial
cognitions toward leisure-time walking. Several
interactions were found but no clear pattern could be
Our finding of an association among attitudes,
self-efficacy, social influence, intention and
leisure-time walking is in line with both theory
and previous empirical research [
a positive social influence was associated with
participating in leisure-time walking but not with
minutes walked. Also, our finding that a large social
network was positively related to participating in
leisure-time walking has been found in previous
]. The negative association between
neighborhood social cohesion and minutes walked
among the walkers was unexpectedly and without a
This study extends on previous research by
exploring environment–individual interactions. Three
interactions were found with perceived
neighborhood safety. For participation in leisure-time
aBold figures indicate statistical significance (P < 0.05), *P < 0.05, **P < 0.01, ***P < 0.001.
walking, perceived neighborhood safety interacted
with attitude and social influence: in those who
sometimes felt unsafe, a positive attitude and a
positive social influence were significantly stronger
associated with any leisure-time walking. This
pattern was not observed for those who often felt
unsafe. This finding was different from findings by
Rhodes et al. [
] who found that low levels of
perceived crime resulted in a larger influence of
attitude on the intention to walk compared with people
who perceived high levels of crime. For our second
outcome, minutes walked among those persons who
engaged in leisure-time walking, also an interaction
with safety was found: those who perceived feelings
of unsafety but had a positive intention to walk in
leisure time walked 30 min per week more than
persons who felt safe in their neighborhood and
persons who lacked intention to walk. Although
these unexpected interactions with safety are hard
to interpret, a possible explanation may be found in
the association between safety and walking itself.
Although we were primarily interested in the
influence of neighborhood safety on leisure-time
walking, the cross-sectional nature of this study cannot
preclude the direction of association. Therefore, it is
possible that those who walk in their neighborhood
are more likely to report feelings of unsafety
because they are more exposed to their neighborhood.
This inverse association between neighborhood
safety and physical activity has been observed
The final interaction observed was between
feeling at home and attitude, whereby feeling at home in
your neighborhood was stronger associated with
1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 fi o n
,1 ly e e
.0 n o n
0 o h e
< re n f
* s t
5 an re n
0 c f n
< se d io
y n .
P* teh l
b g s
a o r
,) ,r to c to
5 e n t c
.0 v ah fa
0 e e t
w b e
. w R ic
n M io to s
ig f c e a
s ch in
e a n v b
a o l
d ith it v u
n n o
in w g i
re tl e ti
y c n s
u h h i m
g t n r
g i g e
fi ls ro o t
ld ry s e io
o a R th t
B v c
O r a
. an e f t
in c th s in
ig ) , R
r s ) e
c d O h
o li – e t
f a a
o it (4 th fo
y n y l s
r i t e r
t ( e d te
u 3 a o e
o s m m
c l a
e d e r
nd od oo th a
r R n h
fo O it
1 4 3 2 5 5 5
.4 .3 .2 .6 6 .0 .2 11 .48
9 2 1 .7 7 9 .
1 7 1 1 2 4 0 1
3 7 1 2 8 4 8 5 4
2 1 3 1
0 5 9 5
2 2 2
2 6 4 5
9 0 8 2 4
3 7 1 2 8 4 8 5 4
2 1 3 1
tirrsceod irfaebhohogdo irfsceeeaevdP ifseadumM fsaenU iliscaceoonohS iltracenookSw liteeeanghoFm iilsscccaohoyogncittt)(eud5–1A lf-f(eeccay–1fiS ilicaeeconnuSflctitI()enon–51n l:ea4dop–odM .tttfsaaea4y+ ifseaedumM fsaenU .tfssaeeb4y+ ifseaedumM fsaenU .tfssace4yo+ ifseaedumM fsaenU .tifsaed4yn+ ifseaedumM fsaenU .isceeo4onh+ .ifsceo4onh+ .iscegoh4on+ .iscehoh4on+ .itren4ko+w .jtren4ko+w .trekn4ko+w .ltren4ko+w .lfee4ohmm+ .lfeen4ho+m .lfeeo4ho+m .lfeep4ho+m alllseodAwm <P..**0100* .illiaynnomm trrsseeeepohn trfreeeceehn
P N P M
engaging in leisure-time walking in those with a
below average attitude. This interaction could
indicate the existence of the first mechanism proposed
in the Introduction: those who have negative
psychosocial cognitions benefit more from a
positive neighborhood environment than those with
more positive psychosocial cognitions toward
physical activity. Or, stated the other way around, not
feeling at home in your neighborhood may not
be a barrier for walking among those with a
positive attitude toward physical activity, as this positive
attitude makes them more likely to be active
Overall, we found limited empirical support for
interactions, and neither of the proposed
mechanisms was clearly favored in our results although
the interaction between feeling at home and attitude
hints at the first mechanism in which those with
negative psychosocial cognitions benefit most
from a positive neighborhood environment. The
recent study by Carlson et al. also found a limited
number of interactions [
]. In their article, they
studied the interactions among walkability, parks
and recreation facilities, aesthetics and walking
facilities within the neighborhood with social
support, self-efficacy and perceived barriers on
leisuretime walking. They found one significant interaction
between walking facilities and self-efficacy in
which self-efficacy was only associated with
leisure-time walking in neighborhoods with few
walking facilities. This interaction also supports the first
proposed mechanism in which positive psychosocial
cognitions can help to overcome neighborhood
barriers. Although methodological reasons, including
lack of statistical power and measurement error in
environmental and (to a lesser extent) individual
factors, may have contributed to this finding, it is
also possible that walking behavior mainly is a result
of a combination of environmental and individual
factors, in which only few interactions are involved,
which have little implications for public health
practice. However, the strong theoretical support for
environment–individual interactions in ecological
models prompts for more research that indentifies
and quantifies these interactions.
Study limitations and strengths
Several limitations need to be considered in the
interpretation of the findings of this study. First, the
cross-sectional design restricts interpretation on
causality and direction of the associations. This is
particularly relevant because of the increasing
recognition of a dynamic interrelation in which
individuals change places and places change
]. Second, our psychosocial cognitions
were measured with regard to ‘sufficient physical
activity in line with recommended levels’ where it
would have better preferred to ask this specifically
for leisure-time walking. This may have resulted in
an underestimation of associations with leisure-time
walking. Third, self-reported physical activity data
are known for overestimations. In addition, the
SQUASH questionnaire was validated for total
physical activity but not for the underlying specific
activities such as leisure-time walking. Because this
study used a robust dichotomous measure, it is
expected to be of little influence although we can not
exclude some bias in the associations. Finally, the
results of this study should be interpreted in the
context of a medium-sized city in the Netherlands. The
situation in Dutch urban areas may not be
representative for other urban areas in the world.
This study explored interactions between
neighborhood factors and psychosocial cognitions for
explaining leisure-time walking in adults and found
limited evidence for these interactions. The
relationship between neighborhood and individual
determinants of walking and environment–individual
interactions remains complex, and more studies
are needed that incorporate these interactions to
strengthen these results.
Supplementary data are available at HEALED
The GLOBE study is carried out by the Department
of Public Health of the Erasmus University Medical
Centre in Rotterdam, in collaboration with the
Public Health Services of the city of Eindhoven
and region South-East Brabant.
The Netherlands organization for health research
and development (ZonMw; 122000003).
Conflict of interest statement
1. Centers for Disease Control and Prevention . 1996 . Physical Activity and Health: A Report of the Surgeon General . Washington DC: U.S. Department of Health and Human services, 1996 .
2. Centers for Disease Control, National Center for Health Statistics, Division of Health Promotion Statistics. 2008. Data 2010 . . . The Healthy People 2010 Database - Focus area: 22 - Physical Activity and Fitness. Available at: http:// www.healthypeople.gov/Data/data2010.htm. Accessed: 29 April 2011 .
3. Warburton DE , Nicol CW , Bredin SS . Health benefits of physical activity: the evidence . Can Med Assoc J 2006 ; 174 : 801 - 9 .
4. Bauman AE . Updating the evidence that physical activity is good for health: an epidemiological review 2000-2003 . J Sci Med Sport 2004 ; 7 : 6 - 19 .
5. Trost SG , Owen N , Bauman AE et al. Correlates of adults' participation in physical activity: review and update . Med Sci Sports Exerc 2002 ; 34 : 1996 - 2001 .
6. Sallis JF , Owen N. Physical Activity and Behavioral Medicine . Thousand Oaks , CA: Sage Publications, 1999 .
7. Owen N , Humpel N , Leslie E et al. Understanding environmental influences on walking: review and research agenda . Am J Prev Med 2004 ; 27 : 67 - 76 .
8. Saelens BE , Sallis JF , Frank LD . Environmental correlates of walking and cycling: findings from the transportation, urban design, and planning literatures . Ann Behav Med 2003 ; 25 : 80 - 91 .
9. Saelens BE , Handy SL . Built environment correlates of walking: a review . Med Sci Sports Exerc 2008 ; 40 : S550 - 66 .
10. McCormack G , Giles-Corti B , Lange A 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 .
11. Van Holle V , Deforche B , Van Cauwenberg J et al. Relationship between the physical environment and different domains of physical activity in European adults: a systematic review . BMC Public Health 2012 ; 12 : 807 .
12. Ball K , Timperio A , Salmon J et al. Personal, social and environmental determinants of educational inequalities in walking: a multilevel study . J Epidemiol Community Health 2007 ; 61 : 108 - 14 .
13. De Bourdeaudhuij I , Teixeira PJ , Cardon G et al. Environmental and psychosocial correlates of physical activity in Portuguese and Belgian adults . Public Health Nutr 2005 ; 8 : 886 - 95 .
14. Giles-Corti B , Donovan RJ . Relative influences of individual, social environmental, and physical environmental correlates of walking . Am J Public Health 2003 ; 93 : 1583 - 9 .
15. Granner ML , Sharpe PA , Hutto B et al. Perceived individual, social, and environmental factors for physical activity and walking . J Phys Act Health 2007 ; 4 : 278 - 93 .
16. Kamphuis CB , van Lenthe FJ , Giskes K et al. Socioeconomic differences in lack of recreational walking among older adults: the role of neighbourhood and individual factors . Int J Behav Nutr Phys Act 2009 ; 6 : 1 .
17. Emmons KM . Health behaviors in a social context . In: Berkman LF , Kawachi I (eds). Social Epidemiology . New York: Oxford University Press, 2000 .
18. Sallis JF , Owen N , Fisher EB . Ecological models of health behavior . In: Glanz K , Rimer BK , Viswanath K (eds). Health Behavior and Health Educaton Theory, Research, and Practice . San Fransisco, CA: Jossey-Bass , 2008 , 465 - 86 .
19. Kremers S , de Bruijn G-J , Visscher T et al. Environmental influences on energy balance-related behaviors: a dual-process view . Int J Behav Nutr Phys Act 2006 ; 3 : 9 .
20. Bauman AE , Sallis JF , Dzewaltowski DA et al. Toward a better understanding of the influences on physical activity: the role of determinants, correlates, causal variables, mediators, moderators, and confounders . Am J Prev Med 2002 ; 23 : 5 - 14 .
21. Carlson JA , Sallis JF , Conway TL et al. Interactions between psychosocial and built environment factors in explaining older adults' physical activity . Prev Med 2012 ; 54 : 68 - 73 .
22. Rhodes RE , Brown SG , McIntyre CA . Integrating the perceived neighborhood environment and the theory of planned behavior when predicting walking in a Canadian adult sample . Am J Health Promot 2006 ; 21 : 110 - 8 .
23. Cerin E , Vandelanotte C , Leslie E et al. Recreational facilities and leisure-time physical activity: an analysis of moderators and self-efficacy as a mediator . Health Psychol 2008 ; 27 : 26 - 35 .
24. Addy CL , Wilson DK , Kirtland KA et al. Associations of perceived social and physical environmental supports with physical activity and walking behavior . Am J Public Health 2004 ; 94 : 440 - 3 .
25. Lindstro¨m M , Hanson BS , O¨ stergren PO. Socioeconomic differences in leisure-time physical activity: the role of social participation and social capital in shaping health related behaviour . Soc Sci Med 2001 ; 52 : 441 - 51 .
26. Cleland VJ , Timperio A , Crawford D. Are perceptions of the physical and social environment associated with mothers' walking for leisure and for transport? A longitudinal study . Prev Med 2008 ; 47 : 188 - 93 .
27. McNeill LH , Kreuter MW , Subramanian SV . Social environment and physical activity: a review of concepts and evidence . Soc Sci Med 2006 ; 63 : 1011 - 22 .
28. Pikora T , Giles-Corti B , Bull F et al. Developing a framework for assessment of the environmental determinants of walking and cycling . Soc Sci Med 2003 ; 56 : 1693 - 703 .
29. Loukaitou-Sideris A , Eck JE . Crime prevention and active living . Am J Health Promot 2007 ; 21 : 380 - 9 , iii.
30. van Lenthe FJ , Schrijvers CTM , Droomers M et al. Investigating explanations of socio-economic inequalities in health: the Dutch GLOBE study . Eur J Public Health 2004 ; 14 : 63 - 70 .
31. Mackenbach JP , van de Mheen H , Stronks K. A prospective cohort study investigating the explanation of socio-economic inequalities in health in the Netherlands . Soc Sci Med 1994 ; 38 : 299 - 308 .
32. Wendel-Vos GCW , Schuit AJ , Saris WHM et al. Reproducibility and relative validity of the short questionnaire to assess health-enhancing physical activity . J Clin Epidemiol 2003 ; 56 : 1163 - 9 .
33. Ajzen I. The theory of planned behavior . Organ Behav Hum Decis Process 1991 ; 50 : 179 - 211 .
34. Bandura A. Social Foundations of Thought and Action: A Social Cognitive Theory . Englewood Cliffs, NJ: Prentice Hall, 1986 .
35. Kemper HCG , Ooijendijk WTM , Stiggelbout M. Consensus about the Dutch recommendation for physical activity to promote health [Consensus over de Nederlandse Norm voor Gezond Bewegen] . Tijdschrift voor Gezondheidswetenschappen 2000 ; 78 : 180 - 3 .
36. Kawachi I , Berkman LF . Social cohesion, social capital, and health . In: Berkman LF , Kawachi I (eds). Social Epidemiology . New York: Oxford University Press, 2000 .
37. Van Berkel-Van Schaik AB , Tax B . Towards a Standard Operationalisation of Socioeconomic Status for Epidemiological and Socio-medical Research [Naar een standaardoperationalisatie van sociaal-economische status voor epidemiologisch en sociaal-medisch onderzoek] . Rijswijk: Ministerie van Welzijn, Volksgezondheid en Cultuur , 1990 , 129 .
38. Dempster AP , Laird NM , Rubin DB . Maximum likelihood from incomplete data via the EM algorithm . J Roy Stat Soc B Met 1977 ; 39 : 1 - 38 .
39. Rothman KJ , Greenland S , Lash TL. Modern Epidemiology . Philadelphia. PA: Lippincott Williams & Wilkins, 2008 .
40. Hosmer DW , Lemeshow S. Confidence interval estimation of interaction . Epidemiology 1992 ; 3 : 452 - 6 .
41. Knol MJ , van der Tweel I , Grobbee DE et al. Estimating interaction on an additive scale between continuous determinants in a logistic regression model . Int J Epidemiol 2007 ; 36 : 1111 - 8 .
42. Rabe-Hesketh S , Skrondal A , Pickles A . Maximum likelihood estimation of limited and discrete dependent variable models with nested random effects . J Econom 2005 ; 128 : 301 - 23 .
43. Gauvin L , Riva M , Barnett T et al. Association between neighborhood active living potential and walking . Am J Epidemiol 2008 ; 167 : 944 - 53 .
44. Humpel N , Owen N , Iverson D et al. Perceived environment attributes, residential location, and walking for particular purposes . Am J Prev Med 2004 ; 26 : 119 - 25 .
45. Diez Roux AV . Complex systems thinking and current impasses in health disparities research . Am J Public Health 2011 ; 101 : 1627 - 34 .