Urban greenspace, physical activity and wellbeing: The moderating role of perceptions of neighbourhood affability and incivility

Land Use Policy 57 (2016) 638–644

Contents lists available at ScienceDirect

Land Use Policy journal homepage: www.elsevier.com/locate/landusepol

Urban greenspace, physical activity and wellbeing: The moderating role of perceptions of neighbourhood affability and incivility Christopher L. Ambrey Cities Research Centre, Gold Coast Campus, Griffith University, Queensland 4222, Australia

a r t i c l e

i n f o

Article history: Received 7 August 2015 Received in revised form 12 June 2016 Accepted 25 June 2016 Keywords: Health Exercise Geographic Information Systems Greenspace Mental health Physical activity Wellbeing

a b s t r a c t The built environment can affect a resident’s health and wellbeing. Land use planning has the potential promote the health and wellbeing of residents. The provision of greener urban environments is one mechanism through which land use planners might achieve this end. The purpose of this study is to examine how greenspace and physical activity may provide synergistic wellbeing benefits and how any such hypothesised synergy might depend on a resident’s perceptions of neighbourhood affability and incivility. Using data from the Household, Income and Labour Dynamics in Australia (HILDA) survey and Geographic Information Systems (GIS), the results suggest that a friendlier and more supportive neighbourhood amplifies (by more than 10 times) the greenspace and physical activity synergy. In contrast, the results suggest that residents who engage in physical activity and have higher levels of greenspace in their local area suffer a wellbeing burden where the resident perceives higher levels of incivility in the neighbourhood. While certainly not without their own limitations, these results extend on existing research efforts directed at disentangling the complexity underpinning the links between greenspace, physical activity and wellbeing. Moreover, the findings reported in this study may prove useful to land use planners and policy makers seeking to reconcile the challenges of maintaining or improving residents’ wellbeing in the face of pervasive neighbourhood perceptions, continuing population growth and declining per capita greenspace. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction The significance of the built environment for health and wellbeing is widely acknowledged (Barton, 2009; World Health Organization, 2010a,b). Land use planning has the potential to create built environments that promote the health and wellbeing of residents. Access to greenspace is a promising channel through which planning can support the health and wellbeing of residents (Barton, 2009; Carmichael et al., 2013). The availability of greenspace in a resident’s neighbourhood is thought to promote physical activity (Coombes et al., 2010), a substantial portion of which occurs within the boundaries of neighbourhoods (Hurvitz et al., 2014). The findings to date though are not unequivocal. For instance, evidence for an area of metropolitan Perth, Western Australia indicates that residents with very good access to large and attractive public open spaces are more likely to engage in higher levels of walking (Giles-Corti et al., 2005). However, in comparison, results from a cross-sectional study of two study areas (St. Louis, Missouri and Savannah, Georgia) in the

E-mail address: c.ambrey@griffith.edu.au http://dx.doi.org/10.1016/j.landusepol.2016.06.034 0264-8377/© 2016 Elsevier Ltd. All rights reserved.

United States report suggest that greenspace within a 5 min walk is not related to meeting physical activity guidelines and that other individual and environmental factors represent the antecedents to engagement in the recommended amount of physical activity (Hoehner et al., 2005). A similar result is observed by Hillsdon et al. (2006) among a cohort of middle-to-older age adults located in the city of Norwich, England and echoed by Maas et al. (2008) for the Netherlands. Coombes et al. (2010) provide more recent evidence for the city of Bristol, England that indicates living close to formal greenspaces (those with an organised layout and structured path network, and generally well maintained) is associated with a greater likelihood of achieving recommended levels of physical activity. Nevertheless, Mytton et al. (2012) and Mitchell (2013) unsettle the very foundations of the hypothesised relationship noting that such a link may not be explained by physical activity undertaken in green spaces. It may be the case that statistically significant associations actually reflect some omitted variable(s). Beyond the well-established physiological and psychological benefits of physical activity (US Department of Health, 1996), there is a subset of the literature which concerns itself with the cobenefits of ‘green exercise’ (cf. Astell-Burt et al., 2013; Bodin and Hartig, 2003; Gladwell et al., 2013; Hug et al., 2009; Mitchell, 2013; Pretty et al., 2007; Sugiyama and Ward Thompson, 2008;

C.L. Ambrey / Land Use Policy 57 (2016) 638–644

Thompson Coon et al., 2011; Ward Thompson et al., 2012, inter alia). These co-benefits relate to psychological and even physiological benefits of natural environments in terms of reducing stress (Ulrich et al., 1991) and rejuvenating directed attentional capacity (Kaplan, 1995). Further synergistic benefits report include but are not limited to: improvements in social connectedness; greater appreciation of nature; enhanced mood, increased self-esteem, heightened vigour; and reduced feelings of anger, depression, tension and fatigue. It is argued that wellbeing (mental health) is improved by the enjoyment derived from spending time with other people, getting out into the fresh air, the scenery, the wildlife and participating in exercise itself (Barton et al., 2009; Peacock et al., 2007; Pretty et al., 2007). The hypothesised synergy between physical activity and greenspace is thought to improve the efficacy and longevity of physical activity interventions and concomitantly aid in informing residents on the importance of protecting the natural environment and embracing sustainable development (Thompson Coon et al., 2011). However, despite the best efforts of earlier investigators a number of caveats remain (Mitchell, 2013). Thompson Coon et al. (2011) lament the research design problems they identified which hamper the interpretation and extrapolation of findings. For instance: small sample sizes; unrepresentative samples (often comprised of university students or already highly motivated participants); little geographic variation (most studies reviewed were conducted within university campuses in the United States); little information on participants (factors such as age and physical health which may confound results); a lack of independent assessment of participant outcomes and heterogeneous outcome measures which are not easily transferable. Further and more carefully designed research is needed (Gladwell et al., 2013; Thompson Coon et al., 2011). The approach of this study involves matches data from the Household, Income and Labour Dynamics in Australia (HILDA) survey, a large spatially referenced national probability sample of Australians, to Geographic Information Systems (GIS) data. Quite unique in this literature this study employs a cluster-specific fixed effects estimator. An estimator which features more prominently in the applied microeconomics literature (cf. Shields et al., 2009), this estimator uses deviations of the dependent and independent variables from their broader local area (the Local Government Area (LGA) more specifically) means to identify the (moderated) association between greenspace and wellbeing (the measures for which importantly vary within the broader local area). In doing so, this technique allows for potentially confounding influences unique to the local area to be differentiated out using multiple observations (multiple residents) per local area. These potential confounders would otherwise lead to biased and inconsistent estimates (Cameron and Trivedi, 2009, 2010). Despite the abovementioned criticisms of this increasingly active area of scholarly inquiry there are also studies which inspire optimism. In a particularly novel exploratory study (n = 25) Ward Thompson et al. (2012) use ecologically valid objective measures of stress (diurnal patterns of cortisol) to identify the links between exposure to greenspace and stress. The results point to greener environments offering better opportunities for moderating or coping with stress through pathways other than those related to physical activity. In contrast, in a large study of 260,061 middleto-older age Australians Astell-Burt et al. (2013) investigate an interaction (explicitly operationalised) between greenspace and physical activity levels and observe that more greenspace in one’s neighbourhood is associated with a lower likelihood of experiencing psychological distress for more physically active residents (although not for more sedentary residents). While such research efforts admirable further research is required to reach a definitive conclusion regarding the greenspace and physical activity synergy.

639

In particular, it may be the case that the triadic relationship between greenspace, physical activity and wellbeing is in fact be more complicated than has tended to be hypothesised. Different types of environment may elicit different psychological responses (Mitchell, 2013). Thus, the realisation of hypothesised synergistic benefits may be contingent on particular individual, social environmental and physical environmental conditions such as fear of crime in the neighbourhood (cf. Sreetheran and van den Bosch, 2014) or different dimensions of a resident’s social environment (cf. McNeill et al., 2006). It may be that a resident’s perceptions of neighbourhood affability1 or incivility2 motivate a resident’s actions, which have their own implicit and explicit costs and benefits (Becker, 1968) and hence, implications for wellbeing. This is a point which accords with earlier related research efforts into the link between greenspace and physical activity which identifies “. . .the potential moderating effects of factors such as social fragmentation or indicators of neighbourhood quality such as crime and disorder. . .” (Coombes et al., 2010). Relatedly, there remains is a need for studies “. . .to recognise and identify neighbourhood level and individual level factors that may moderate the strength of association between neighbourhood influences and physical activity. . ..” (Li et al., 2005). In a single equation model one could expect this to be operationalised as a three-way interaction effect (e.g. greenspace × physical activity × affability of the neighbourhood). The purpose of this study is to fill this gap in the literature and do so in a manner that circumvents methodological problems reported elsewhere. In doing so for the case of major Australian cities this study extends on earlier research efforts which seek to disentangle the complexity underpinning the links between greenspace, physical activity and wellbeing. Furthermore, the findings reported in this study may prove relevant to land use planners seeking to cultivate healthier and happier cities (Ballas, 2013) given the broader challenges posed by continuing population growth and declining per capita greenspace in cities. In what follows, Section 2 reports the method and data employed. Section 3 provides an account of the results. Finally, Section 4 discusses the findings and concludes. 2. Method This section details the method and data employed to undertake this investigation. Section 2.1 describes Eq. (1). Section 2.2 describes Eq. (2). Section 2.3 explains the fixed effects estimation technique used throughout this study. Finally, Sections 2.4 and 2.5 describe the HILDA data and GIS data respectively. 2.1. Eq. (1) To examine the hypothesised greenspace and physical activity synergy wellbeing is modelled using a cluster-specific fixed effects model as show in Eq. (1): WBr,k = ␻ +

m 

␤j xjr,k + ␥xr,k yr,k + ␬k + ␧r,k

(1)

j=1

Where, dependent variable WBr,k represents a resident’s mental health. xjr,k denotes variables j = 1. . .m. These variables include: physical activity, greenspace and a number of socioeconomic variables. These socioeconomic variables these include for example; age, gender, ethnicity, health; similar to earlier studies of wellbeing

1 Neighbourhood affability is defined as a resident’s perceptions of behaviours in the neighbourhood being friendly or convivial in nature. 2 Neighbourhood incivility is defined as a resident’s perceptions of behaviours in the neighbourhood being characterised as unsociable or offensive.

640

C.L. Ambrey / Land Use Policy 57 (2016) 638–644

in Australia (cf. Ambrey and Fleming, 2014). The inclusion of these socioeconomic variables mitigates the risk that omitted or unobserved variables might confound the triadic relationship between physical activity, greenspace and wellbeing. xr,k yr,k represents the two-way interaction term of greenspace × physical activity which operationalises the hypothesised greenspace and physical activity synergy. ␬k represents the LGA-specific fixed effects. Finally, ␧r,k is the error term. 2.2. Eq. (2) In order to investigate how any hypothesised synergies between greenspace and physical activity might depend on perceptions of neighbourhood affability and incivility Eq. (1) is augmented to include additional interaction terms yielding: WBr,k = ␻ +

m 

j=1

␤j xj r,k +

q 

p=1

␥p xp r,k yp r,k +

b 

␥a xa r,k ya r,k za r,k + ␬k + ␧r,k

(2)

a=1

Where xp r,k ypr,k represents two-way interaction terms p = 1. . .q, for instance, greenspace × physical activity. In addition, lower order interaction terms such as, physical activity × affability of the neighbourhood, are also included to preserve the interpretation of the key three-way interaction terms of interest. Finally, xa r,k yar,k za r,k represents three-way interaction terms, a = 1. . .b. For example, greenspace × physical activity × affability of the neighbourhood. All other terms are as previously defined. 2.3. Estimation strategy There are significant differences between the cluster-specific fixed effects estimator and a conventional ordinary least squares regression. The the cluster-specific fixed effects estimator employed in this study uses the within transformation. That is, it uses deviations of the dependent and independent variables from the broader local area (the LGA more specifically) means to identify the (moderated) association between greenspace and wellbeing (the measures for which importantly vary within the broader local area). In doing so, this technique allows for potentially confounding influences unique to the local area to be differentiated out using multiple observations (multiple residents) per local area. These potential confounders would otherwise lead to biased and inconsistent estimates (Cameron and Trivedi, 2009, 2010). Furthermore, throughout standard errors are adjusted for clustering of residents within the LGA which could otherwise bias standard errors downwards due to similarities among residents and a common exposure to LGA-level shocks (Nichols and Schaffer, 2007). 2.4. Household, income and labour dynamics in australia survey data The socioeconomic data is obtained from wave 12 (2012) of the Household, Income and Labour Dynamics in Australia (HILDA) survey. The HILDA survey is national probability sample, first conducted in 2001 which owes much to other household panel studies conducted elsewhere in the world; particularly the German Socio-Economic Panel and the British Household Panel Survey. The reference population for first wave of the survey was all members of private dwellings in Australia aged 15 years and over following coverage rules broadly in line with the monthly Labour Force Survey supplements undertaken by the Australian Bureau of Statistics. The sampling design of the survey involves the selection of households into the sample by a multi-stage process. To begin with, a random sample of 488 CDs based on the 1996 census boundaries was selected from across Australia, stratified by State, and within the five largest States in terms of population, by metropolitan and non-metropolitan regions, each CD consisting of

approximately 200–250 households. The CDs were sampled with probability proportional to their size as measured by the number of dwellings (unoccupied and occupied) recorded in the 1996 Census with some adjustments for population growth since the Census. Within each of these CDs, all dwellings were fully enumerated and a sample of 22–34 dwellings randomly sampled based on the expected response and occupancy rates within each area (Watson and Wooden, 2002). The HILDA survey has progressed a great deal since its inception and has facilitated high quality research (Watson and Wooden, 2012). This data is subset to those residents who reside in Australia’s major cities as defined by the Accessibility/Remoteness Index of Australia3 and those residents who fall within the capital cities of Australia as defined by the Greater Capital City Statistical Areas. By this definition Australia’s major capital cities include: Adelaide, Brisbane, Canberra, Darwin, Melbourne, Perth and Sydney. This subset of the HILDA survey data yields a sample of 3288 responding residents. The sample contains areas that would be regarded as densely populated within Australia. However, Australia’s cities when compared to other cities around the world are relatively less densely populated (Australian Bureau of Statistics, 2014). To be specific, in the sample there are on average 18 (with a range of 0.0049 and 279.9117) residents per hectare and approximately 85% of the dwellings are separate houses. The mental health variable is measured by the SF-36 Health Survey (within the HILDA Survey), an internationally recognised diagnostic tool for assessing functional health status and wellbeing. The Mental Component Summary (MCS) (0–100) used is derived from factor analysis of data on of a subset of eight scales, vitality, social functioning, role-emotional and mental health, derived from 36 items, transformed to a 0–100 index using 1995 Australian Bureau of Statistics population norms (Australian Bureau of Statistics, 1995; Ware et al., 2000). The key measure of physical activity is derived from residents’ responses to the question ‘In general, how often do you participate in moderate or intensive physical activity for at least 30 minutes?’ Where responses include: ‘Not at all’, ‘Less than once a week’, ‘1 to 2 times a week’, ‘3 times a week’, ‘More than 3 times a week (but not every day)’ and ‘Every day’.4 This question is dichotomised to yield the physical activity variable, which takes a value 1 if the resident responded ‘More than 3 times a week (but not every day)’ or ‘Every day’ and 0 if the resident responded ‘Not at all’ or ‘Less than once a week’ or ‘1 to 2 times a week’. This dichotomisation has also been used in other contexts (cf. Perales et al., 2014).5 Residents’ perceptions of neighbourhood affability and incivility are obtained from the question ‘How common are the following things in your local neighbourhood? (a) Neighbours helping each other out. (b) Neighbours doing things together. (c) Loud traffic noises. (d) Noises from airplanes, trains or industry.

3 Defined as those areas with relatively unrestricted accessibility in terms of road distances to a wide range of goods and services and opportunities for social interaction (Australian Population and Migration Research Centre, 2014). 4 It is important to note that the survey data does not provide an indication of where a resident engaged in this physical activity. 5 An anonymous reviewer suggested an alternative dichotomisation of the physical activity variable. In this alternative specification the variable takes a value 1 if the resident responded ‘Every day’ and 0 if the resident responded More than 3 times a week (but not every day)’ or ‘Not at all’ or ‘Less than once a week’ or ‘1 to 2 times a week’. In this case, the greenspace × physical activity interaction term was not found to be statistically significant at conventional levels. However, the results for this categorisation relating to how any hypothesised synergies between greenspace and physical activity might depend on perceptions of neighbourhood affability and incivility are similar in magnitude, sign and statistical significance to those reported in Table 3.

C.L. Ambrey / Land Use Policy 57 (2016) 638–644

641

Table 1 Key model variables. Variable name

Definition

Mean (std. dev.)

Dependent variable Mental health

Resident’s SF-36 Mental Component Summary (MCS) (0–100)

49.20 (9.84)

Independent variables Physical activity Greenspace (ha) per capita

Neighbourly interaction and support (Q2) Neighbourly interaction and support (Q3) Neighbourly interaction and support (Q4) Local disamenity (Q2) Local disamenity (Q3) Local disamenity (Q4) Insecurity in the neighbourhood (Q2) Insecurity in the neighbourhood (Q3) Insecurity in the neighbourhood (Q4)

Resident participates in moderate or intensive physical activity for at least 30 min either ‘More than 3 times a week (but not every day)’ or ‘Every day’ The amount of greenspace in a resident’s CD per resident in the CD. Greenspace in this instance, includes for instance, public parks, community gardens, cemeteries, sports fields, national parks and wilderness areas Resident perceives neighbourly interaction and support in the second quartile Resident perceives neighbourly interaction and support in the third quartile Resident perceives neighbourly interaction and support in the fourth quartile Resident perceives local disamenity in the second quartile Resident perceives local disamenity in the third quartile Resident perceives local disamenity in the fourth quartile Resident perceives insecurity in the neighbourhood in the second quartile Resident perceives insecurity in the neighbourhood in the third quartile Resident perceives insecurity in the neighbourhood in the fourth quartile

(e) Homes and gardens in bad condition. (f) Rubbish and litter lying around. (g) Teenagers hanging around the streets. (h) People being hostile or aggressive. (i) Vandalism and deliberate damage to property. (j) Burglary and theft’. Where residents may respond: 1 ‘Never happens’; 2 ‘Very rare’; 3 ‘Not common’; 4 ‘Fairly common’; and 5 ‘Very common’. Following Shields et al. (2009) the scores for (a) and (b) are used to construct a measure of ‘neighbourly interaction and support’, ranging from 2 to 10. Similarly, (c)–(f) are aggregated to obtain a measure of ‘local disamenity’ that ranges from 4 to 20. Finally, aggregating (g)–(j) provides a measure of ‘insecurity in the neighbourhood’ (4–20). To reveal further potential nonlinear relationships, these measures are then dichotomised into quartiles. 2.5. Geographic Information Systems data The data from the HILDA survey is linked to Geographic Information Systems (GIS) data on greenspace through the resident’s CD. For reasons of confidentiality the CD is lowest level of spatial reference for residents in the HILDA survey. While this means that the precise residential address of respondents is unknown this spatial reference also provides an upper bound for the degree of measurement error.6 The use of the CD permits the convenient synthesis of the major factors influencing the use of greenspace, size and proximity (Schipperijn et al., 2010). It is reasonable to expect that residents are more likely to use or experience greenspace that is nearby (cf. Coombes et al., 2010). Using GIS CDs are overlayed with greenspace measured from PSMA Australia Limited Transport and Topography dataset. Greenspace includes for instance; public parks, community gardens, cemeteries, sports fields, national parks and wilderness areas (cf. Bell et al., 2008). The variable is the number of hectares of greenspace per resident in the CD. As noted earlier, this greenspace variable is identified as the CD-level measure varies within the broader local government area fixed effects. Within the sample there are 128 unique local government areas and 1331 unique CDs. The mean (median) local government area is 469 (120) km2 , whereas the CD has a mean (median) area of 1.2 (0.3) km2 . To put this area in context, the

6 The resident is at most on average (or at the median) 615 m (312 m) from the greenspace captured within the CD if it is assumed to take approximately the shape a circle.

%

31.2% 0.18 (3.58)

42.7% 19.9% 13.6% 25.9% 24.0% 15.6% 30.3% 32.2% 13.4%

Table 2 Key mental health model results (Eq. (1)).8 Variable name

Coefficient (standard error)

Greenspace (ha) per capita Physical activity Greenspace (ha) per capita × Physical activity

−0.0147 (0.0076) 2.1334** (0.2999) 0.3592* (0.1499)

Summary statistics Observations Groups

3288 128

Standard errors in parentheses adjusted for clustering at the LGA level. Other controls included: Age, gender, ethnicity, marital status, parenting, health, educational attainment, employment status, manual work, income, social desirability bias, free time, social interaction, household members engaged in physical activity, personality traits, years at current address, years interviewed, proximity to lake, proximity to river, proximity to coastline, ‘neighbourly interaction and support’, ‘local disamenity’, ‘insecurity in the neighbourhood’ and the SEIFA 2011 Index. * p < 0.05. ** p < 0.01.

United States county boundaries have a mean (median) area of 5728 (1614) km2 . A detailed description and summary statistics of the key variables are provided Table 1. 3. Results The key mental health model results for Eq. (1) are reported in Table 2.7 Table 2 lends support to the hypothesised synergistic relationship between greenspace and physical activity as indicated by the coefficient for the greenspace and physical activity interaction term (0.3592), statistically significant at the 5% level. These results also suggest that the benefits of greenspace are highly dependent on whether or not a resident also engages in physical activity as indicated by the imprecisely estimated greenspace coefficient (−0.0147). The key mental health model results for Eq. (2) are reported in Table 3. Given the interaction terms in the model the coefficient estimates in Table 3 require careful interpretation. Table 3 points to a synergy between physical activity and greenspace which is markedly greater in neighbourhoods which are perceived to be friendlier. A result evidenced by the positive coefficient for the

8 Note, variance inflation factors of a base model with no interaction terms shows no sign of worrisome multicollinearity. 7 Note that the results have been adjusted for socioeconomic characteristics detailed at the bottom of Table 2. 9 Note, variance inflation factors of a base model with no interaction terms shows no sign of worrisome multicollinearity.

642

C.L. Ambrey / Land Use Policy 57 (2016) 638–644

Table 3 Key mental health model results (Eq. (2)).9 Variable name

Coefficient

(standard error)

Greenspace (ha) per capita Physical activity Neighbourly interaction and support (Q2) Neighbourly interaction and support (Q3) Neighbourly interaction and support (Q4) Local disamenity (Q2) Local disamenity (Q3) Local disamenity (Q4) Insecurity in the neighbourhood (Q2) Insecurity in the neighbourhood (Q3) Insecurity in the neighbourhood (Q4) Greenspace (ha) per capita × Physical activity Greenspace (ha) per capita × Neighbourly interaction and support (Q2) Physical activity × Neighbourly interaction and support (Q2) Greenspace (ha) per capita × Physical activity × Neighbourly interaction and support (Q2) Greenspace (ha) per capita × Neighbourly interaction and support (Q3) Physical activity × Neighbourly interaction and support (Q3) Greenspace (ha) per capita × Physical activity × Neighbourly interaction and support (Q3) Greenspace (ha) per capita × Neighbourly interaction and support (Q4) Physical activity × Neighbourly interaction and support (Q4) Greenspace (ha) per capita × Physical activity × Neighbourly interaction and support (Q4) Greenspace (ha) per capita × Local disamenity (Q2) Physical activity × Local disamenity (Q2) Greenspace (ha) per capita × Physical activity × Local disamenity (Q2) Greenspace (ha) per capita × Local disamenity (Q3) Physical activity × Local disamenity (Q3) Greenspace (ha) per capita × Physical activity × Local disamenity (Q3) Greenspace (ha) per capita × Local disamenity (Q4) Physical activity × Local disamenity (Q4) Greenspace (ha) per capita × Physical activity × Local disamenity (Q4) Greenspace (ha) per capita × Insecurity in the neighbourhood (Q2) Physical activity × Insecurity in the neighbourhood (Q2) Greenspace (ha) per capita × Physical activity × Insecurity in the neighbourhood (Q2) Greenspace (ha) per capita × Insecurity in the neighbourhood (Q3) Physical activity × Insecurity in the neighbourhood (Q3) Greenspace (ha) per capita × Physical activity × Insecurity in the neighbourhood (Q3) Greenspace (ha) per capita × Insecurity in the neighbourhood (Q4) Physical activity × Insecurity in the neighbourhood (Q4) Greenspace (ha) per capita × Physical activity × Insecurity in the neighbourhood (Q4)

0.0000 1.7118* 0.1752 0.7219 0.8081 −1.0316* −0.4944 −0.0254 0.1174 −0.4178 −3.0153** 1.0966 −0.9307 0.8619 0.2093 −0.3933 0.3681 4.3930* −1.2509 −0.0272 1.4684 0.2816 1.3249 1.6100 −0.6383 0.9507 −2.5236 1.5816 1.5437 −3.9200 −0.2724 −0.9067 −0.6389 1.3087 −1.4446 −0.8459 4.2052 −0.5538 −6.3065*

(0.0000) (0.8306) (0.4799) (0.6703) (0.5708) (0.4563) (0.5824) (0.7058) (0.5849) (0.6244) (0.8273) (0.9276) (0.9491) (0.9466) (1.5132) (1.3297) (0.9949) (1.7019) (0.9844) (0.9739) (1.6874) (0.9393) (0.6757) (1.3360) (1.2041) (0.9202) (1.9483) (3.4769) (1.2168) (3.2527) (0.9369) (0.8227) (1.4065) (1.0571) (0.9399) (1.4578) (2.9169) (1.1858) (2.9371)

Summary statistics Observations Groups

3288 128

Standard errors in parentheses adjusted for clustering at the LGA level. Controls are the same as those detailed in Table 2. * p < 0.05. ** p < 0.01.

greenspace, physical activity and neighbourly interaction and support (Q3) interaction term (4.3930), statistically significant at the 5% level. In stark contrast, the physical activity and greenspace interaction with neighbourhoods characterised by incivility is characterised by much lower levels of mental health. A point illustrated by the negative coefficient for the greenspace, physical activity and insecurity in the neighbourhood (Q4) interaction term (-6.3065), statistically significant at the 5% level. None of the other interaction terms are found to be statistically significant at conventional levels. 4. Discussion The results provide evidence which lends support to the hypothesised greenspace and physical activity synergy, robust to a number of controls. Furthermore, the results also suggest that this synergy depends greatly on perceptions of neighbourhood affability and incivility. Specifically, the results suggest that a friendlier and more supportive neighbourhood amplifies (by more than 10 times) the greenspace and physical activity synergy. In contrast, the results suggest that residents who engage in physical activity and have higher levels of greenspace in their local area suffer a wellbeing

burden where the resident perceives higher levels of incivility in the neighbourhood. While not without their own limitations, these new findings build on earlier research efforts and shed new light on the greenspace and physical activity synergy. Furthermore, the findings reported here may prove useful to land use planners and policy makers seeking to reconcile residents’ wellbeing in the face of seemingly pernicious neighbourhood perceptions, population growth and declining greenspace per capita in cities. The results point perceptions of neighbourhood affability moderating the greenspace and physical activity synergy. This result may be explained by the greater fluidity with which social networking and companionship (thought to be associated with ‘green exercise’ (cf. Thompson Coon et al., 2011)) can be cultivated in places that are already perceived as being more inviting and neighbourly. Planners and policy makers can help nurture this through the design of more walkable neighbourhoods, which are thought to encourage informal meetings, chance encounters and social ties (Barton, 2009). Highlighting the recursive and self-perpetuating nature of designing more walkable neighbourhoods Foster and Giles-Corti (2008) have even suggested that community initiatives which create a supportive, inviting, aesthetically pleasing public realm and that encourage social interaction between residents may promote walking behaviours. The “greenness” of common places

C.L. Ambrey / Land Use Policy 57 (2016) 638–644

represents another mechanism through which social ties in the neighbourhood may be nurtured (Kuo et al., 1998b). The results also point to a moderating role for perceptions of incivility in the neighbourhood. This result suggests that residents who engage in physical activity and have higher levels of greenspace in their local area suffer a greater wellbeing burden where the resident perceives higher levels of incivility in the neighbourhood. It is reasonable to expect that perceiving one’s neighbourhood as uncivil may lead a resident to avoid engaging in social interactions and avoid greenspace where it has become uncivilised. For instance, residents may avoid greenspace that has become the refuge of bored loitering youths, vandalism and property destruction (e.g. ‘Broken Windows’) with implications for their personal wellbeing (Ambrey et al., 2014; Becker, 1968; Fleming et al., 2016; Manning et al., 2015; Wagers et al., 2008). Planners and policy makers through the physical attributes of greenspace have the potential to send a positive signal to residents and potential offenders that a neighbourhood is a pleasant, a civilised, and a cared-for place with civilised standards of behaviour (Kuo et al., 1998a). Practical steps may be taken to remedy perceptions of incivility using crime prevention through environmental design. To the extent that residents are unnecessarily burdened by misperceptions (cf. Ambrey et al., 2014) targeted education and upto-date information about crime in the neighbourhood may prove effective (Weatherburn and Indermaur, 2004). It should be noted though that the results gleaned from this study while intriguing are not unimpeachable. Despite the inclusion of a number of controls which account for many local area factors, these results do not identify causal relations. In this regard and in line with the conclusions of the more generous earlier reviews of the evidence (cf. Bowler et al., 2010; Gladwell et al., 2013) there ‘may be’ or there ‘could be’ synergistic wellbeing benefits associated with physical activity in greenspace. Further, these benefits ‘may be’ or there ‘could be’ moderated by affability and incivility in the neighbourhood. The insights of Mytton et al. (2012) and Mitchell (2013) remain relevant, precisely where residents in the sample engage in their physical activity is unclear as is the usability and accessibility of local greenspace. In addition, it is not difficult to imagine that a resident’s decision to use greenspace depends on the attributes of neighbourhood greenspace. In this regard, there is scope to extend on the findings presented here and to identify how particular physical characteristics of greenspaces may be related to the greenspace and physical activity synergy. For instance, the degree of informal surveillance in greenspaces, the diversity of flora and fauna or the presence of walking trails. Characteristics such as these are not explicitly operationalised in this study although they may have important implications for the useability of greenspace and hence the greenspace and physical activity synergy. Future researchers continuing the quest to understand the greenspace and physical activity synergy may glean new insights by exploring the role of these attributes. Furthermore, future research effort can be usefully directed towards achieving the ideal research design epitomised by the random assignment of participants to treatment (greenspace) and control groups. Although less than ideal other well-controlled approaches include, quasi-experimental designs and the use of difference-in-difference models (Gibbons and Overman, 2012). Coupled with these broad methodological approaches, there also remains a need for further research which takes advantage of more refined and objective measures. In all, this study draws attention to the complexity underlying the links between greenspace, physical activity and wellbeing. In particular, the results point to a sizable conditionality between the hypothesised synergistic benefits and neighbourhood perceptions. In addition to advancing existing knowledge on the greenspace and

643

physical activity synergy it is hoped that these findings prove useful for land use planners and policy makers seeking reconcile residents’ wellbeing with continued population growth and pervasive neighbourhood perceptions in urban centres.

Acknowledgements This paper uses unit record data from the Household, Income and Labour Dynamics in Australia (HILDA) survey. The HILDA Project was initiated and is funded by the Australian Government Department of Social Services (DSS) and is managed by the Melbourne Institute of Applied Economic and Social Research (Melbourne Institute). The findings and views reported in this paper, however, are those of the author and should not be attributed to either DSS or the Melbourne Institute. The author would like to thank the anonymous referees for their comments on an earlier draft. All errors and omissions remain those of the author.

References Ambrey, C., Fleming, C., 2014. Life satisfaction in Australia: evidence from ten years of the HILDA survey. Soc. Indic. Res. 115, 691–714. Ambrey, C., Fleming, C., Manning, M., 2014. Perception or reality, what matters most when it comes to crime in your neighbourhood? Soc. Indic. Res. 119, 877–896. Astell-Burt, T., Feng, X., Kolt, G.S., 2013. Mental health benefits of neighbourhood green space are stronger among physically active adults in middle-to-older age: evidence from 260,061 Australians. Prev. Med. 57, 601–606. Australian Bureau of Statistics, 1995. 4399.0-National Health Survey: SF36 Population Norms. Australian Bureau of Statistics, Canberra. Australian Bureau of Statistics, 2014. Sydney Is Our Densest City, but We Are Lightweights Internationally, 1270.0.55.007-Australian Population Grid, 2011. Australian Bureau of Statistics, Canberra. Australian Population, Migration Research Centre, 2014. ARIA (Accessibility/Remoteness Index of Australia). The University of Adelaide, Adelaide. Ballas, D., 2013. What makes a ‘happy city’? Cities 32, 539–550. Barton, H., 2009. Land use planning and health and well-being. Land Use Policy 26S, S115–S123. Barton, J., Hine, R., Pretty, J., 2009. The health benefits of walking in greenspaces of high natural and heritage value. J. Integr. Environ. Sci. 6, 261–278. Becker, G., 1968. Crime and punishment: an economic approach. J. Political Econ. 76, 169–217. Bell, S., Hamilton, V., Montarzino, A., Rothnie, H., Travlou, P., Alves, S., 2008. Greenspace Scotland Research Report: Greenspace and Quality of Life: A Critical Literature Review. In: Scotland, G. (Ed.). Greenspace Scotland, Stirling. Bodin, M., Hartig, T., 2003. Does the outdoor environment matter for psychological restoration gained through running? Psychol. Sport Exerc. 4, 141–153. Bowler, D.E., Buyung-Ali, L.M., Knight, T.M., Pullin, A.S., 2010. A systematic review of evidence for the added benefits to health of exposure to natural environments. BMC Public Health 10, 456. Cameron, A., Trivedi, P., 2009. Microeconometrics: Methods and Applications. Cambridge University Press, Cambridge. Cameron, A., Trivedi, P., 2010. Microeconometrics Using Stata, Revised Edition. Stata Press, College Station, Texas. Carmichael, L., Barton, H., Gray, S., Lease, H., 2013. Health-integrated planning at the local level in England: impediments and opportunities. Land Use Policy 31, 259–266. Coombes, E., Jones, A., Hillsdon, M., 2010. The relationship of physical activity and overweight to objectively measured green space accessibility and use. Soc. Sci. Med. 70, 816–822. Fleming, C.M., Manning, M., Ambrey, C.L., 2016. Crime, greenspace and life satisfaction: an evaluation of the New Zealand experience. Landscape Urban Plann. 149, 1–10. Foster, S., Giles-Corti, B., 2008. The built environment, neighborhood crime and constrained physical activity: an exploration of inconsistent findings. Prev. Med. 47, 241–251. Gibbons, S., Overman, H.G., 2012. Mostly pointless spatial econometrics. J. Reg. Sci. 52, 172–191. Giles-Corti, B., Broomhall, M., Knuiman, M., Collins, C., Douglas, K., Ng, K., Lange, A., Donovan, R., 2005. Increasing walking: how important is distance to, attractiveness, and size of public open space? Am. J. Prev. Med. 28, 169–176. Gladwell, V.F., Brown, D.K., Wood, C., Sandercock, G.R., Barton, J.L., 2013. The great outdoors: how a green exercise environment can benefit all. Extreme Physiol. Med. 2, 3. Hillsdon, M., Panter, J., Foster, C., Jones, A., 2006. The relationship between access and quality of urban green space with population physical activity. Public Health 120, 1127–1132.

644

C.L. Ambrey / Land Use Policy 57 (2016) 638–644

Hoehner, C.M., Brennan Ramirez, L.K., Elliott, M.B., Handy, S.L., Brownson, R.C., 2005. Perceived and objective environmental measures and physical activity among urban adults. Am. J. Prev. Med. 28, 105–116. Hug, S.-M., Hartig, T., Hansmann, R., Seeland, K., Hornung, R., 2009. Restorative qualities of indoor and outdoor exercise settings as predictors of exercise frequency. Health Place 15, 971–980. Hurvitz, P.M., Moudon, A.V., Kang, B., Fesinmeyer, M.D., Saelens, B.E., 2014. How far from home? The locations of physical activity in an urban U.S. setting. Prev. Med. 69, 181–186. Kaplan, S., 1995. The restorative benefits of nature: toward an integrative framework. J. Environ. Psychol. 15, 169–182. Kuo, F., Bacaicoa, M., Sullivan, W., 1998a. Transforming inner city landscapes: trees, sense of safety and preference. Environ. Behav. 30, 28–59. Kuo, F., Sullivan, W., Coley, R., Brunson, L., 1998b. Fertile ground for community: inner-city neighborhood common spaces. Am. J. Community Psychol. 26, 823–851. Li, F., Fisher, K., Brownson, R., Bosworth, M., 2005. Multilevel modelling of built environment characteristics related to neighbourhood walking activity in older adults. J. Epidemiol. Community Health 59, 558–564. Maas, J., Verheij, R.A., Spreeuwenberg, P., Groenewegen, P.P., 2008. Physical activity as a possible mechanism behind the relationship between green space and health: a multilevel analysis. BMC Public Health 8, 1–13. Manning, M., Fleming, C.M., Ambrey, C.L., 2015. Life satisfaction and individual willingness to pay for crime reduction. Reg. Stud. http://www.tandfonline. com/doi/abs/10.1080/00343404.2015.1082030?journalCode=cres20. McNeill, L., Kreuter, M., Subramanian, S., 2006. Social environment and physical activity: a review of concepts and evidence. Soc. Sci. Med. 63, 1011–1022. Mitchell, R., 2013. Is physical activity in natural environments better for mental health than physical activity in other environments? Soc. Sci. Med. 91, 130–134. Mytton, O.T., Townsend, N., Rutter, H., Foster, C., 2012. Green space and physical activity: an observational study using Health Survey for England data. Health Place 18, 1034–1041. Nichols, A., and Schaffer, M., 2007. Clustered Errors in Stata. Retrieved from: http:// repec.org/usug2007/crse.pdf, accessed: 14.12.10. Peacock, J., Hine, R., Pretty, J., 2007. Got the blues, then find some greenspace: The mental health benefits of green exercise activities and green care, Mind Report No. 1, Colchester. Perales, F., del Pozo-Cruz, J., del Pozo-Cruz, J., del Pozo-Cruz, B., 2014. On the associations between physical activity and quality of life: findings from an Australian nationally representative panel survey. Qual. Life Res. 23, 1921–1933. Pretty, J., Peacock, J., Hine, R., Sellens, M., South, N., Griffin, M., 2007. Green exercise in the UK countryside: effects on health and psychological well-being, and implications for policy and planning. J. Environ. Plann. Manage. 50, 211–231.

Schipperijn, J., Stigsdotter, U., Randrup, T., Troelsen, J., 2010. Influences on the use of urban green space: a case study in Odense, Denmark. Urban For. Urban Green. 9, 25–32. Shields, M., Price, S., Wooden, M., 2009. Life satisfaction and the economic and social characteristics of neighbourhoods. J. Popul. Econ. 22, 421–443. Sreetheran, M., van den Bosch, C.C.K., 2014. A socio-ecological exploration of fear of crime in urban green spaces—a systematic review. Urban For. Urban Green. 13, 1–18. Sugiyama, T., Ward Thompson, C., 2008. Associations between characteristics of neighbourhood open space and older people’s walking. Urban For. Urban Green. 7, 41–51. Thompson Coon, J., Boddy, K., Stein, K., Whear, R., Barton, J., Depledge, M.H., 2011. Does participating in physical activity in outdoor natural environments have a greater effect on physical and mental wellbeing than physical activity indoors? A systematic review. Environ. Sci. Technol. 45, 1761–1772. Ulrich, R., Simons, R., Losito, B., Fiorito, E., Miles, M., Zelson, M., 1991. Stress recovery during exposure to natural and urban environments. J. Environ. Psychol. 11, 201–230. US Department of Health H.S, 1996. Physical Activity and Health, A Report of the Surgeon General. US Department of Health and Human Services, Centres for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Atlanta, Georgia. Wagers, M., Sousa, W., Kelling, G., 2008. Broken windows. In: Mazerolle, L., Wortley, R. (Eds.), Environmental Criminology and Crime Analysis. Willan, Cullompton. Ward Thompson, C., Roe, J., Aspinall, P., Mitchell, R., Clow, A., Miller, D., 2012. More green space is linked to less stress in deprived communities: evidence from salivary cortisol patterns. Landscape Urban Plann. 105, 221–229. Ware, J., Snow, K., Kosinski, M., 2000. SF-36 Health Survey: Manual and Interpretation Guide. Quality Metric Incorporated, Lincoln, RI. Watson, N., Wooden, M., 2002. The Household, Income and Labour Dynamics in Australia (HILDA) Survey: Wave 1 Survey Methodology, Project Technical Paper Series No. 1/02. Melbourne Institute of Applied Economic and Social Research, The University of Melbourne, Melbourne. Watson, N., Wooden, M., 2012. The HILDA survey: a case study in the design and development of a successful household panel study. Longitud. Life Course Stud. 3, 369–381. Weatherburn, D., Indermaur, D., 2004. Public perceptions of crime trends in new South Wales and Western Australia, crime and justice bulletin: contemporary issues in crime and justice No. 80. N. S. W. Bur. Crime Stat. Res. 80, 1–8. World Health Organization, 2010a. The Parma Declaration on Environment and Health. World Health Organization, Europe, Copenhagen. World Health Organization, 2010b. Urban Planning, Environment and Health: From Evidence to Policy Action. WHO, Europe, Copenhagen.