Deakin Research Online

Deakin Research Online This is the published version: Leslie, Evie, Cerin, Ester and Kremer, Peter 2010-11, Perceived neighborhood environment and park use as mediators of the effect of area socio-economic status on walking behaviors, Journal of physical activity & health, vol. 7, no. 6, pp. 802-810. Available from Deakin Research Online: http://hdl.handle.net/10536/dro/du:30030468 Reproduced with the kind permission of the copyright owner. Copyright : 2010, Human Kinetics, Inc.

Journal of Physical Activity and Health, 2010, 7, 802-810 2010 Human Kinetics, Inc. Perceived Neighborhood Environment and Park Use as Mediators of the Effect of Area Socio-Economic Status on Walking Behaviors Evie Leslie, Ester Cerin, and Peter Kremer Background: Access to local parks can affect walking levels. Neighborhood environment and park use may influence relationships between neighborhood socioeconomic status (SES) and walking. Methods: Self-report data on perceived park features, neighborhood environment, park use, neighborhood walking and sociodemographics were obtained from a sample of Australian adults, living in high/low SES areas. Surveys were mailed to 250 randomly selected households within 500m of 12 matched parks. Mediating effects of perceived environment attributes and park use on relationships between area-ses and walking were examined. Results: Mean frequency of local park use was higher for high-ses residents (4.36 vs 3.16 times/wk, P <.01), who also reported higher levels of park safety, maintenance, attractiveness, opportunities for socialization, and neighborhood crime safety, aesthetics, and traffic safety. Safety and opportunity for socialization were independently positively related to monthly frequency of visits to a local park which, in turn, was positively associated with walking for recreation and total walking. Residents of higher SES areas reported an average 22% (95% CI: 5%, 37%) more weekly minutes of recreational walking than their low SES counterparts. Conclusion: Residents of high-ses areas live in environments that promote park use, which positively contributes to their weekly amounts of overall and recreational walking. Keywords: adults, social disadvantage, environmental attributes, parks, walking behavior Public health researchers have extensively examined features of the physical environment and their relationship with physical activity and walking. 1 5 Of the many environmental features that have been observed to have significant associations with physical activity levels are the presence of parks and public open space. 6,7 Parks are considered to be important settings in providing opportunities for physical activity. 8 Numerous studies have also found that parks and recreation settings are associated with walking behavior, suggesting that some aspects of parks are of value in enhancing this form of physical activity. 7 In addition, residents proximity to parks has generally been associated with increased physical activity. 6 The closeness of residents to parks has been found to support a direct relationship between proximity and visitation. 9 However these determinants operate within a broader social context that influences level of activity. A number of environmental features related to physical activity have been found to vary according to neighborhood socioeconomic status (SES). 10 In addition, area-level SES has been shown to be associated with physical activity 11,12 and Leslie and Kremer are with the School of Psychology, Deakin University, Geelong, Australia. Cerin is with the Institute of Human Performance, The University of Hong Kong, Hong Kong. differences in walking behavior. 11,13 A number of studies have described disparities in the objective availability of parks and recreational facilities based on levels of disadvantage. These studies have produced mixed results. Some report fewer free for use resources in low SES areas and an unequal distribution of private facilities and public parks. 14 16 Others have shown no differences in the number or total area of public open space available for recreation to residents in areas differing by neighborhood SES 17 but found more amenities and supportive features for physical activity in high SES areas. 18 Perceptions of the neighborhood environment, as well as actual availability of facilities, are also important in determining behavior. It has been suggested that perceptions of local environments may mediate some of the behavioral effects observed and may be associated with an increased or decreased likelihood of physical activity engagement. 19 Research suggests that the use of parks is not only related to park proximity but that certain perceived environmental characteristics may encourage or discourage park use. Previously, Baker and colleagues reported that park use was associated with the presence of parks nearby and other perceived factors such as neighborhood safety, enjoyable scenery, traffic and hills. 20 Wilson and colleagues found that residents of low SES areas reported higher perceptions of crime, unattended dogs, unpleasantness of neighborhoods and less access to public recreational facilities. 21 These and 802

Perceived Environment, Park Use, and Walking 803 other perceived factors are plausibly related to park use. Other perceived environmental attributes found to be consistently positively associated with neighborhood walking in the general population include aesthetics and convenience of facilities for walking (eg, paths), while safety concerns were found to be negatively associated. 4,22 Furthermore, an extensive review of park settings and physical activity found that 14 of 18 studies reported some association between parks and walking. 6 Interestingly, a number of perceptions of environmental attributes reported to be associated with walking behaviors have been found to vary according to the purpose of walking, such as walking for transport or walking for recreation. 4 These differences, together with area-level differences reported for residents perceptions of environmental features 15,23 may help explain differences in observed park use for low and high SES residents. It is not fully understood how variations in perceptions may be related to the frequency of park use and to common walking behaviors. The main aims of this study were to assess the mediating (indirect) effects of perceived park and neighborhood environments and park use on the cross-sectional relationship between area-level SES and walking for different purposes (total walking, and walking for recreation and walking for transport) within the neighborhood of residence. Design Methods A cross sectional survey was used to obtain responses from adult residents in high and low SES areas in a regional city. Census data were used to identify areas of high and low SES, using the Socio-Economic Index for Areas (SEIFA) Index of Relative Socio-Economic Advantage/Disadvantage 24 in the regional City of Greater Geelong (CGG), Australia. The SES index score was used to identify suburbs located within the top and bottom 25% of disadvantage. A range of parks located within these suburbs were identified. These comprised sports fields, informal parks, a mixture of informal/sports fields, linear linkages and nature reserves. Following an environmental audit of parks using the Public Open Space Tool 25 and using information obtained from planners at the CGG Open Space and Recreation Department, park pairs of similar size in low and high area-level SES areas were matched on type and facilities available. Sample Selection and Participants A total of 12 local parks comprising 6 park pairs were selected for the study. Due to the relationship between park proximity and park use, participants were randomly selected from a 500m buffer around the centroid of each of the selected parks, using Geographic Information Systems data (GIS) from the City of Greater Geelong s Spatial Information archive. Using the City s Property and Asset System database and VicMap Admin Victorian Postcode Boundaries spatial data, a list of all parcel addresses was obtained within each buffer. The addresses were then cleaned to remove any duplicates and nonresidential parcels before randomly selecting 250 household addresses from each buffer. A total of 3000 surveys (250 per selected park buffer) were mailed to randomly selected households in September 2007. The study used the last birthday method to identify 1 participant from each household. To be eligible for the study participants had to be 18 to 65 years of age, permanent residents of the neighborhood for which they had been selected, and able to speak English. The study was approved by the Deakin University Human Research Ethics Committee. Measurements Features of the Neighborhood Environment. Participants were asked to rate perceived environmental features of their local area using a previously reported 10-point scale. 26 This scale assesses environments in 4 dimensions functional, safety, aesthetics and destinations to walk to. The functional dimension is represented by 3 items related to the construction and integrity of footpaths and streets in the neighborhood (local area). The internal consistency (Cronbach s alpha) of this scale was 0.70. The safety dimension includes 4 items associated with neighborhood traffic volume and speed, lighting, and crime. However, since preliminary analyses indicated that these 4 items were insufficiently intercorrelated, lighting and crime were examined as single items. The traffic and crime items were reverse scored to go in the same direction as the other items (renamed traffic safety, crime safety), with a higher score representing a more positive (safer) rating. The internal consistency of the traffic safety scale was 0.62. The aesthetic dimension includes 2 items about neighborhood cleanliness and views of buildings and scenery (Cronbach s alpha: 0.76). The final dimension of destinations comprises 1 item that asks about the availability of places in and around the participant s neighborhood to which they could walk such as shops, parks, work, or school. Features of a Specific Park. Participants were asked to rate the features described for the specific park for which they had been selected (ie, the one at the centroid of the 500m buffer). Features included park safety during the day time, park attractiveness, park maintenance and opportunities for socializing in the park using a scale of 1 to 5, 27 with higher scores representing more positive ratings. Walking Behaviors. Walking in the neighborhood was assessed using the Neighborhood Physical Activity Questionnaire (NPAQ) items on the frequency and duration of walking for both recreation and for transport in the past week. 28 This tool has been found to be reliable for assessing different types of walking (recreation, ICC: frequency =.92, duration =.90; transport, ICC: frequency =.92, duration =.96) and total walking (ICC duration =.91) within local neighborhood areas. These items allow for assessing walking for different purposes

804 Leslie, Cerin, and Kremer but can also be summed to obtain a total walking score in minutes per week. Park Use (Frequency of Visits). The frequency of park visits used 2 items adapted from a previously used survey. 29 Frequency of park visits was assessed separately for any park and for a specific park in the neighborhood of residence using 8 response categories, ranging from 0 to 8 or more times per month. Demographic Variables and Other Personal Information. Sociodemographic variables included age, gender, education, work status, marital status, occupation, housing, home ownership and living arrangement. Additional personal information included dog ownership and self-reported health status. The later variable was measured on a 5-point scale ranging from poor to excellent. 30 Statistical Analyses As 49.2% of the sample had missing data on at least 1 of the variables examined in this study (5% with 1 missing value; 23.7% with 2 missing values), multiple imputation by chained equations 31 were used to create 5 complete datasets. All analyses were performed on all datasets and combined estimates were obtained using a procedure outlined by Rubin 32 and available in Stata 9.0. The parameter estimates using imputations were very similar to those produced by cases with complete data. However, as expected (due to an increase in sample size), data analyses using imputed data yielded more precise estimates. Hence, we report analyses from imputed data. Descriptive statistics were used to summarize information on characteristics of the sample and key walking and park use measures. Mean ratings on the multi-item environmental attributes were computed by summing the scores for each item and then dividing this value by the number of items in each dimension to allow comparison across dimensions. Where appropriate, ordinary, logistic, gamma, multinomial, and interval regression with robust standard errors accounting for clustering effects (participants clustered in 500m buffers) were used to examine (univariate) differences in sociodemographic and other personal factors, perceived environment, park use and neighborhood walking according to area SES. The main aim of this study was to examine possible mechanisms (mediating effects) through which area SES can affect total neighborhood walking and walking for different purposes. It was hypothesized that area SES would affect perceived park and neighborhood environmental attributes, which would determine park use, which in turn would affect the participants amount of neighborhood walking. To test these hypothetical mediating effects we used the joint significance test, 33 a type of mediating variable analysis (MVA). This consisted of (1) testing area SES differences in perceived environmental attributes, adjusting for potential confounders; (2) testing the significance of the effects of perceived environmental attributes on park use, while adjusting for potential confounders and area SES; and (3) testing the significance of the effects of park use on neighborhood walking, while adjusting for potential confounders, area SES and perceived environmental attributes. Support for the hypothesized mediating effects is found if the regression coefficients of interest across all 3 sets of regressions are significant. We did not employ the recommended bootstrap product-of-coefficient test of mediation because its use is not currently supported in conjunction with multiple imputations. In addition, it is not known how one of the regression models used for the MVA (interval regression; see below) behaves in the context of the product-of-coefficient test of mediation. Given that perceived environmental variables were approximately normally distributed, normal regression was used to perform the first step of the MVA. The second step of the MVA was performed using interval regression, appropriate when the dependent variable (park use) has censored values (ie, participants reporting using the park 8 or more times per month were not all equal on park use). The third step of the analyses was performed using generalized linear models with gamma or negative binomial variance (where appropriate) and logarithmic link functions. Gamma and negative binomial functions are appropriate for nonnegative, positively skewed variables (such as minutes of walking). The gamma variance function is appropriate for variables whose standard deviation is approximately equal to their mean, while the negative binomial variance function can accommodate variables with standard deviations larger than the mean. For these models, the antilogarithms of the regression coefficient and relative confidence intervals were computed to facilitate interpretation. These express the proportional increase in the outcome variable followed by a unit increase in the explanatory variable. All analyses were performed in Stata 9 (StataCorp, College Station TX, 2008) using robust standard errors accounting for clustering effects. A probability level of 0.05 was adopted. Results A total of 555 residents returned the survey (overall response rate = 19.7%) with response rates higher for residents in high-ses areas than for low-ses areas (24.8% vs 14.1% respectively, P <.01). Of these, 502 met eligibility criteria, 36% males, mean age of 47.7 (SD ± 11.3) years, comprising 64.7% residents from high SES areas (see Table 1). Higher proportions of residents from high SES areas were tertiary educated, working in professional/ trade/office occupations and married; lower proportions from high SES areas lived in public housing or were renting, and owned a dog. There were no differences in the proportions living with children or working. High SES residents were older but reported better health and higher frequencies of park visits. No significant unconditional (unadjusted) differences in minutes of neighborhood walking were observed between residents of high and low SES areas. However, the absence of significant univariate associations between area SES and walking does not imply that there are no mediated effects of area SES on walking. 34 This is because the effects of area SES may

Perceived Environment, Park Use, and Walking 805 be underlain by opposing, nearly-balanced mechanisms that cancel out. 35 In addition, the effects of area SES may be too distal to be detected without considering their pathways of influence. 36 Summaries for area SES differences in perceived environmental attributes are shown in Table 2. High-SES area residents reported higher levels of park safety, maintenance, attractiveness and opportunities for socialization. They also reported higher levels of neighborhood crime safety, aesthetics, and traffic safety. There were no significant differences in ratings for destinations, functionality and lighting. Table 1 Sample Characteristics by Area-Level SES Low SES High SES Total P General N (%) 177 (35.5) 325 (64.7) 502 Age, mean years (SD) 45.5 (12.4) 48.9 (10.6) 47.7 (11.4) <.05 Gender, % men 35 36 36 NS Education, % tertiary educated 23 49 40 <.001 Work status, % working 57 66 63 NS Occupation, % professional trade/sales/office 47 62 57 <.001 Housing, % public 15 1 6 <.001 Home ownership, % renting 35 13 21 <.001 Marital status, % couple 48 70 62 <.001 Living arrangement, % living with children 37 33 34 NS Dog ownership, % own dog 58 44 49 <.01 Health status (scale 1 5), mean (SD) 3.32 (0.97) 3.91 (0.91) 3.70 (0.97) <.01 Neighborhood walking and park use Recreational walking (mean mins/wk, SD) 162 (165) 177 (154) 172 (158) NS Transport walking (mean mins/wk, SD) 140 (218) 107 (222) 119 (221) NS Total walking (mean mins/wk, SD) 294 (287) 272 (237) 280 (255) NS Any park visit frequency, mean (SD) 3.73 (3.11) 5.19 (2.99) 4.68 (3.11) <.001 Local park visit frequency, mean (SD) 3.16 (2.91) 4.36 (3.10) 3.94 (3.09) <.01 Table 2 First Step of Mediating Variable Analyses: Area SES Differences in Perceived Environmental Factors Environmental attribute Park a Mean (SD) Regression coefficient (difference between low and high SES areas) 95% CI P Safety 4.20 (0.97) 0.40 0.61, 0.19 <.001 Maintenance 3.64 (1.08) 0.43 0.70, 0.16.002 Attractiveness 3.65 (1.07) 0.65 0.97, 0.32 <.001 Opportunities for socializing 2.99 (1.22) 0.70 0.91, 0.48 <.001 Neighborhood b Functionality 5.68 (1.66) 0.11 0.47, 0.24.523 Lighting 5.07 (2.10) 0.07 0.55, 0.70.817 Crime safety 6.32 (2.20) 1.58 2.35, 0.81 <.001 Traffic safety 4.67 (1.89) 0.96 1.45, 0.48 <.001 Aesthetics 6.32 (1.69) 0.78 1.44, 0.13.019 Destinations 7.43 (2.18) 0.18 1.19, 0.83.725 Note. Models adjusted for age, children in household, gender, working status, self-reported health status, educational attainment, and dog ownership. Results of normal regression models. a Park attributes; higher scores represent more positive ratings; b Neighborhood attributes; higher scores represent more positive ratings.

806 Leslie, Cerin, and Kremer Park safety, opportunities for socializing in the park, neighborhood functionality, crime and traffic safety, and access to destinations in the neighborhood were all independently positively related, while lighting was negatively related, to monthly frequency of visits to a specific local park (Table 3). After accounting for perceived environmental attributes, no significant associations were observed between area SES and monthly frequency of visits to a specific park. These results, in combination with those presented in Table 2, indicate that, controlling for sociodemographic and other personal variables, residents of high SES areas may tend to visit their local park more frequently than do residents of low SES areas because their parks and neighborhoods are safer, and they provide more opportunities for socializing. After adjusting for area SES and perceived environmental attributes, monthly frequency of visits to a specific local park was positively associated with neighborhood walking for recreation and total walking, but not with walking for transport (Table 4). Specifically, a unit difference in monthly park visits was associated with a 9% and 6% difference in mean weekly minutes of walking for recreation and total walking, respectively. Overall, in conjunction with the results reported in Tables 2 and 3, this suggests that residents of high SES areas live in environments that promote park use, which in turn positively contributes to their weekly amounts of overall and recreational walking. Apart from crime safety, perceived environmental factors did not have a significant direct effect on walking; that is their effects operated through frequency of park use. Higher levels of crime safety (ie, more safe from crime) were associated with lower levels of walking for transport and total walking but not walking for recreation (Table 4). After adjusting for perceived environmental attributes and park visits, area SES did not significantly contribute to the explanation of neighborhood walking for transport and total walking but it contributed to the explanation of neighborhood walking for recreation, whereby residents of lower SES areas reported an average of 22% (95% CI: 5%, 37%) fewer weekly minutes of walking for recreation than their higher SES counterparts. Discussion This study examined the mediating effects of perceived neighborhood and park attributes on the relationship of area SES with park use and with neighborhood walking for different purposes. It was postulated that area SES differences in perceived environmental attributes would explain differences in park use and walking. Our study found evidence indicating that the cross-sectional association between area-level SES and walking behaviors cannot be understood in terms of a simple direct relationship. Perceived environmental and park attributes and the use of local parks are likely mediators of the associations between area SES and neighborhood walking for recreation. To understand the multiple mediated effect of SES on walking behavior it is useful to consider each stage of this suggested causal chain. In the first step of our analyses we found that perceptions of attributes of local parks and the neighborhood varied for SES (Table 2). Generally, those from low SES areas rated attributes of Table 3 Second Step of Mediating Variable Analyses: Associations of Perceived Environmental Attributes and Area SES With Monthly Frequency of Visits to a Specific Park Monthly frequency of visits to a specific park Factor Regression coefficient (95% CI) P Park a Safety 0.71 (0.53, 0.90) <.001 Maintenance 0.18 ( 0.37, 0.01).069 Attractiveness 0.35 ( 0.03, 0.73).072 Opportunities for socializing 0.34 (0.28, 0.41) <.001 Neighborhood b Functionality 0.11 (0.09, 0.13) <.001 Lighting 0.14 ( 0.17, 0.12) <.001 Crime safety 0.05 (0.01, 0.09).010 Traffic safety 0.04 (0.01, 0.08).017 Aesthetics 0.01 ( 0.03, 0.02).888 Destinations 0.04 (0.02, 0.06) <.001 Area SES (ref: high SES) 0.54 ( 1.72, 0.65).376 Note. Models adjusted for age, children in household, gender, working status, self-reported health status, educational attainment, dog ownership, and type of park. Results of interval regression models. ref = reference category. a Park attributes; higher scores represent more positive ratings; b Neighborhood attributes; higher scores represent more positive ratings.

Perceived Environment, Park Use, and Walking 807 Table 4 Third Step of Mediating Variable Analyses: Associations of Monthly Frequency of Visits to a Specific Park, Perceived Environmental Attributes, Area SES With Neighborhood Walking Weekly minutes of walking for transport Weekly minutes of walking for recreation Total weekly minutes of walking Factor Exp(b) (95% CI) P Exp(b) (95% CI) P Exp(b) (95% CI) P Monthly frequency of visits to a 1.01 (0.94, 1.09).718 1.09 (1.05, 1.13) <.001 1.06 (1.02, 1.09).001 specific park Park a Safety 0.97 (0.81, 1.15).693 0.94 (0.85, 1.04).206 0.95 (0.86, 1.05).309 Maintenance 1.10 (0.88, 1.37).418 0.99 (0.91, 1.09).877 1.03 (0.91, 1.16).651 Attractiveness 0.89 (0.68, 1.17).402 0.97 (0.88, 1.08).573 0.97 (0.82, 1.14).690 Opportunities for socializing 0.99 (0.86, 1.15).930 0.98 (0.91, 1.05).577 0.99 (0.94, 1.04).728 Neighborhood b Functionality 1.02 (0.92, 1.13).706 1.00 (0.94, 1.06).928 1.02 (0.98, 1.08).426 Lighting 1.00 (0.92, 1.10).920 0.99 (0.95, 1.03).573 0.99 (0.94, 1.04).623 Crime safety 0.92 (0.86, 0.99).030 0.96 (0.92, 1.00).063 0.95 (0.91, 0.99).016 Traffic safety 0.99 (0.91, 1.08).888 0.99 (0.95, 1.03).636 0.99 (0.94, 1.05).763 Aesthetics 1.05 (0.92, 1.20).474 0.96 (0.91, 1.01).150 1.00 (0.94, 1.07).963 Destinations 1.00 (0.91, 1.11).967 1.01 (0.97, 1.06).610 1.01 (0.95, 1.07).749 Area SES (ref: high SES) 1.10 (0.69, 1.76).682 0.78 (0.63, 0.95).014 0.93 (0.71, 1.21).575 Note. Models adjusted for age, children in household, gender, working status, self-reported health status, educational attainment, dog ownership, and type of park. Results of generalized linear models. ref = reference category. a Park attributes; higher scores represent more positive ratings; b Neighborhood attributes; higher scores represent more positive ratings. their local park and neighborhood environments lower than those from high SES areas. Park attributes including safety, maintenance, attractiveness, and opportunities for socializing were all rated lower in low-ses areas. Neighborhood attributes including crime safety, traffic safety and aesthetics were also rated lower in low-ses areas. These findings suggest that parks in high-ses areas are perceived by residents to be more attractive, safer and better maintained than those in low-ses areas. However there were no SES-based differences for neighborhood functionality, lighting and destinations. It should be noted that the differences described are based on perceptions and not objective differences. Since we did not include objective measures it is not possible to conclude whether these differences reflect true area differences or some other perceptual process. Nevertheless, perceptual differences are likely to be important regardless of actual variations in the environment. The pattern of findings is consistent with previous studies utilizing objective measures 18 that have reported better amenities and supportive features for physical activity in high SES areas, suggesting that perceptual judgments of residents are accurate representations of observable SES-based differences on key attributes of their environment. The second step of our analysis examined associations between park and neighborhood attributes and area-ses with the frequency of park visits (Table 3). Two park attributes, safety and opportunities for socializing were positively associated with visits, and neighborhood attributes including functionality, lighting (negative), crime and traffic safety, and destinations were all associated with visitation, but area-ses was not. As we have already identified (step 1) many of these attributes vary according to area-ses, and thus people from high-ses areas may visit their local park more frequently as they consider their local neighborhoods are safer and provide better opportunities for socializing. In step 3, we were able to show that frequency of park visits was associated with neighborhood walking for recreation and total walking but not walking for transport (after adjusting for park and neighborhood attributes, and area-ses). In combination with the findings for steps 1 and 2, this suggests that residents of high-ses areas perceive their environments differently (ie, safer, more attractive, better maintained and likely to create opportunities for socialization) which encourages greater park use and that this higher frequency of park use contributes to their levels of neighborhood recreational walking. Having a park nearby has been previously associated with more walking for recreation, 37 but our study design ensured that all parks were a similar distance to residents homes. Thus, our results suggest that, for more frequent visits to occur, the park also needs to be perceived positively. In contrast, perceived environment and park attributes and subsequent park use did not appear to influence neighborhood walking for transport. One explanation for this

808 Leslie, Cerin, and Kremer is that walking for transport is fundamentally different in nature and purpose to walking for recreation, and is associated with different built environment attributes that we did not capture. 38,39 Interestingly, after adjusting for environmental attributes and park visits, area-ses was associated with neighborhood walking for recreation (Table 4), while no significant univariate associations between area SES and walking for recreation were observed (Table 1). This is likely to be due to the covariates in the multivariate model explaining a substantial portion of the residual variance of walking for recreation from the unadjusted model, thus, increasing the precision of the estimate of area SES effect. Further, after adjusting for area-ses and park visits, park and neighborhood attributes were generally not associated with any of the 3 walking measures. The only exception to this was for crime safety which was significantly negatively associated with neighborhood walking for transport and total walking, but only marginally with recreational walking. One possible explanation for this observed finding is that regular walkers may be more aware of signs of crime in their neighborhood. Overall our findings suggest that after adjusting for sociodemographic, neighborhood attribute, and park visitation variables, residents from high-ses areas report more time walking for recreation than do their low-ses counterparts. However, the observed significant findings across the sequence of steps indicate that, in addition to this direct SES effect, the association between SES and neighborhood walking behavior is mediated by residents perceptions of their neighborhood and park environments and the frequency of visits to their local park. Thus, it appears that 2 interconnected mechanisms are involved. First, people in low- and high-ses areas perceive their environments differently. Second, these perceptions act to influence park usage which in turn contributes to the amount of time devoted to neighborhood walking for recreation. This study includes a number of limitations. The use of self-reported measures for assessing attributes of the environment may not accurately represent actual (objective) aspects of the park and neighborhood environments. However, both objective and perceived features of the neighborhood environment may be important influences on behavior. 40 Previous studies 41 have highlighted differences between subjectively reported and objectively quantified attributes of local environments, although we have previously reported good agreement between objective and perceived measures of neighborhood environment attributes. 42 Furthermore, the nature or type of park may affect interpretation of the findings. 6 We did not adjust for park type and it is possible that different types of parks may have different attributes (and hence rated perceptions). However, in selecting parks to the study, we matched for park type (as well as size and available facilities) so that park pairs in high- and low-ses areas had similar attributes. As the study used a cross-sectional design, the findings indicating that both perception of neighborhood and park attributes and frequency of park use mediate the relationship between area SES and neighborhood walking behavior can only be inferred since causality cannot be determined. The response rate in this study was relatively low and respondents were sampled from a narrow geographical location. We did not ask about any physical limitations that may have affected walking ability. Hence, the findings from this study may lack generalizability. Nevertheless, they are theoretically plausible and in line with extant research, which speaks in favor of their validity. Strengths of this study include all analyses having been controlled for key sociodemographic variables and the inclusion of multiple mediators into analysis of the relationship between SES and neighborhood walking behavior. This study showed that recreational walking behavior varied for residents from different SES areas. However it also highlights that perceptions of neighborhood and park environments differ according to SES, and the frequency of visits to local parks can be partly explained by these differences, and further that variation in park visits is also associated with recreational walking behavior. Thus, to understand the exact nature of the association between area-ses and neighborhood walking behavior it is important to not only consider individual differences that may exist across different SES demographics, but also the different type of environments associated with these areas and how these may impact on the perception and use of local parks and ultimately walking behavior. Efforts to encourage those in low-ses areas to increase physical activity should not only focus on addressing individual differences (eg, attitudes to walking and physical activity, perceived barriers for physical activity) but also on changing perceptions of local neighborhood and park environments that have been shown to independently influence walking behavior. Our study used a measure of neighborhood walking only. It would be helpful if future research on parks and walking behaviors included details on where the walking reported was taking place (for example, walking to/from the park or within the park), as has been the call for improved specificity for ecological models of physical activity behavior. 43 Health professionals and local government administrators should consider how policy and planning decisions around local parks may impact on park use and walking behavior. The extent to which any improvements to local environments and park quality may overcome apparent individual differences and perceptions of these is unclear. Acknowledgments The authors gratefully acknowledge the assistance of the City of Greater Geelong s Open Space and Recreation Department and for providing access to property and parks data through their Spatial Information Archive. The Public Open Space Tool (POST) used to conduct environmental audits of parks was developed by Melissa Broomhall and Billie Giles-Corti, Department of Public Health, The University of Western Australia. The study was supported by the Deakin University Promoting Equity and Social Justice Research Cluster through the Faculty of Health, Medicine, Nursing and Behavioural Sciences. Dr Leslie is supported by an NHMRC Public Health Fellowship #301261.

Perceived Environment, Park Use, and Walking 809 References 1. Berrigan D, Troiano RP. The association between urban form and physical activity in U.S. adults. Am J Prev Med. 2002;23(2S):74 79. 2. Humpel N, Owen N, Leslie E. Environmental factors associated with adults participation in physical activity. A review. Am J Prev Med. 2002;22(3):188 199. 3. Humpel N, Owen N, Leslie E, Marshall AL, Bauman A, Sallis JF. Associations of location and perceived environmental attributes with walking in neighborhoods. Am J Health Promot. 2004;18(3):239 242. 4. Owen N, Humpel N, Leslie E, Bauman A, Sallis JF. Understanding environmental influences on walking: review and research agenda. Am J Prev Med. 2004;27(1):67 76. 5. Owen N, Cerin E, Leslie E, et al. Neighborhood walkability and the walking behavior of Australian adults. Am J Prev Med. 2007;33(5):387 395. 6. Kaczynski AT, Henderson KA. Environmental correlates of physical activity: a review of evidence about parks and recreation. Leis Sci. 2007;29:315 354. 7. Kaczynski AT, Henderson KA. Parks and recreation settings and active living: a review of associations with physical activity function and intensity. J Phys Act Health. 2008;5:619 632. 8. Bedimo-Rung AL, Mowen AJ, & Cohen DA. The significance of parks to physical activity and public health. A conceptual model. Am J Prev Med. 2005;28(2S2):159 168. 9. Mowen A, Orsega-Smith E, Payne L, Ainsworth B, Godbey G. The role of park proximity and social support in shaping park visitation, physical activity, and perceived health among older adults. J Phys Act Health. 2007;4:167 179. 10. Macintyre S, Ellaway A. Social and local variations in the use of urban neighbourhoods: a case study in Glasgow. Health Place. 1998;4:91 94. 11. Cerin E, Leslie E. How socio-economic status contributes to participation in leisure-time physical activity. Soc Sci Med. 2008;66:2596 2609. 12. Kavanagh AM, Goller JL, King T, Jolley D, Crawford D, Turrell G. Urban area disadvantage and physical activity: a multilevel study in Melbourne, Australia. J Epidemiol Community Health. 2005;59:934 940. 13. Ball K, Timperio A, Salmon J, Giles-Corti B, Roberts R, Crawford D. Personal, social and environmental determinants of educational inequalities in walking: a mulitlevel study. J Epidemiol Community Health. 2007;61:108 114. 14. Abercrombie LC, Sallis JF, Conway TL, Frank LD, Saelens BE, Chapman JE. Income and racial disparities in access to public parks and private recreation facilities. Am J Prev Med. 2008;34(1):9 15. 15. Estabrooks PA, Lee RE, Gyurcsik NC. Resources for physical activity participation: does availability and accessibility differ by neighborhood socioeconomic status? Ann Behav Med. 2003;25(2):100 104. 16. Moore LV, Diez Roux AV, Evenson KR, McGinn AP, Brines SJ. Availability of recreational resources in minority and low socioeconomic status areas. Am J Prev Med. 2008;34(1):16 22. 17. Timperio A, Ball K, Salmon J, Roberts R, Crawford D. Is availability of public open space equitable across areas? Health Place. 2007;13(2):335 340. 18. Crawford D, Timperio A, Giles-Corti B, et al. Do features of public open spaces vary according to neighbourhood socio-economic status? Health Place. 2008;14(4):887 891. 19. Nasar JL. Assessing perceptions of environments for active living. Am J Prev Med. 2008;34(4):357 363. 20. Baker EA, Schootman M, Kelly C, Barnidge E. Do recreational resources contribute to physical activity? J Phys Act Health. 2008;5:252 261. 21. Wilson DK, Kirtland KA, Ainsworth BE, Addy CL. Socioeconomic status and perceptions of access and safety for physical activity. Ann Behav Med. 2004;28(1):20 28. 22. McCormack G, Giles-Corti B, Lange A, Smith T, Martin K, Pikora TJ. 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(1S):81 92. 23. Giles-Corti B, Donovan RJ. Socioeconomic status differences in recreational physical activity levels and real and perceived access to a supportive physical environment. Prev Med. 2002;35:601 611. 24. Australian Bureau of Statistics. Socio-economic indexes for areas, 2001. Canberra: Australian Bureau of Statistics; 2003. 25. Giles-Corti B, Broomhall MH, Knuiman M, Collins C, Douglas K, Ng K, et al. Increasing walking. How important is distance to, attractiveness, and size of public open space? Am J Prev Med. 2005;28(2S2):169 176. 26. Suminski RR, Poston WSC, Petosa RL, Stevens E, Katzenmoyer LM. Features of the neighbourhood environment and walking by U.S. adults. Am J Prev Med. 2005;28(2):149 155. 27. Phongsavan P, Merom D, Bowles H, van der Ploeg H. Quasi-experimental evaluation of an urban park regeneration program: Example of a naturalistic study. 10 th International Congress of Behavioural Medicine, Tokyo, Japan, August 27-30. Conference proceedings, P196-RT, 124; 2008 [Abstract]. 28. Giles-Corti B, Timperio A, Cutt H, et al. Development of a reliable measure of walking within and outside the local neighborhood: RESIDE s Neighbourhood Physical Activity Questionnaire. Prev Med. 2006;42:455 459. 29. Parks Victoria. State of the parks 2007 report. Melbourne: Parks Victoria; 2007. ([ISBN 978 07311 8370 8]) 30. Armstrong T, Bauman A, Davies J. Physical activity patterns of Australian adults. Canberra: Australian Institute of Health and Welfare (CVD-10); 2000. 31. van Buuren S, Boshuizen HC, Knook DL. Multiple imputation of missing blood pressure covariates in survival analysis. Stat Med. 1999;18:681 694. 32. Rubin D. Multiple imputation for nonresponse in surveys. New York: Wiley; 1987. 33. MacKinnon DP. Introduction to statistical mediation analysis. New York: Lawrence Erlbaum Associates; 2008. 34. Cerin E, MacKinnon DP. A commentary on current practice in mediating variable analyses in behavioural nutrition and physical activity. Public Health Nutr. 2009;12:1182 1188. 35. Cerin E, Leslie E, Owen N. Explaining socio-economic status differences in walking for transport: an ecological analysis of individual, social and environmental factors. Soc Sci Med. 2009;68(6):1013 1020. 36. Shrout PE, Bolger N. Mediation in experimental and nonexperimental studies: new procedures and recommendations. Psychol Methods. 2002;7:422 445. 37. Ball K, Bauman A, Leslie E, Owen N. Perceived environmental aesthetics and convenience and company are associated with walking for exercise in Australian adults. Prev Med. 2001;33:434 440. 38. 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(2):80 91.

810 Leslie, Cerin, and Kremer 39. Cerin E, Leslie E, du Toit L, Owen N, Frank LD. Destinations that matter: associations with walking for transport. Health Place. 2007;13(3):713 724. 40. Gebel K, Bauman A, Owen N. Correlates of non-concordance of perceived with objectively-derived measures of neighborhood walkability. Ann Behav Med. In press. 41. Boehmer TK, Hoehner CM, Wyrwich KW, Brennan Ramirez LK, Brownson RC. Correspondence between perceived and observed measures of neighborhood environmental supports for physical activity. J Phys Act Health. 2006;3:22 36. 42. Leslie E, Saelens B, Frank L, et al. Residents perceptions of walkability attributes in objectively different neighbourhoods: a pilot study. Health Place. 2005;11(3):227 236. 43. Giles-Corti B, Timperio A, Bull F, Pikora T. Understanding physical activity environmental correlates: increased specificity for ecological models. Exerc Sport Sci Rev. 2005;33:175 181.