Elsevier

Journal of Transport & Health

Volume 5, June 2017, Pages 133-141
Journal of Transport & Health

Identifying destination distances that support walking trips in local neighborhoods

https://doi.org/10.1016/j.jth.2016.08.009Get rights and content

Highlights

  • Network buffers subsume destinations at smaller network buffer distances.

  • 401 m–800 m and 801 m–1200 m donut-buffers count destinations with greater specificity.

  • Destinations within 401 m–800 m could be responsible for associations with walking found at 1200 m.

  • Having a variety of types of destinations may increase walking trips.

  • Donut-buffers warrant further investigation to inform urban planning policy guidelines.

Abstract

When examining associations between local destinations and walking it is common to count local destinations using street network buffers measured at various distances to mitigate spatial data aggregation issues caused by scale and the Modifiable Areal Unit Problem. However, it remains unclear whether a particular buffer size is preferred since large buffers may mask important effects whilst small buffers may not accurately represent a neighborhood area. Furthermore, the use of various buffer distances in measuring destination counts does not yield specific information on distances where destinations could be placed in order to increase levels of walking. This paper extends current methods to address these issues by using a new method to define network buffers to identify threshold distances for walking to seven destination types using multilevel models. Donut-buffers are introduced as a method of counting destinations between distances of 401 m–800 m and 801 m–1200 m which are compared to standard network buffers at distances of 400 m, 800 m and 1200 m respectively. We found that destinations within 401 m–800 m could be responsible for associations found at a network buffer of 1200 m. Specifically, the odds of walking increased when local food outlets including supermarkets, cafés/takeaway stores, and small food stores, were located within 401 m–800 m but not 801 m–1200 m, suggesting that these destinations encourage walking when placed at 401 m–800 m away. Consequently we argue that donut-buffers offer greater specificity than standard network buffers for geographic measurement of destinations. This warrants further investigation to inform urban policy guidelines for designing walkable environments.

Introduction

The quantity, mix and proximity of destinations located in neighborhoods are strongly associated with walking (Ewing and Cervero, 2010) with a growing number of longitudinal studies confirming these findings (, , ). Decisions made by urban planners and designers about the proximity, mix and quantity of local destinations therefore play a key role in determining whether or not residents have the opportunity to, and are likely, to walk.

A best practice approach to designing pedestrian-friendly communities, has been to create the ‘400 m neighborhood’ (Western Australian Planning Commission, 2000), which locates shops and services within a ‘walkable’ distance of 5 minutes from residences (i.e., 400 m). However, there is little evidence to support whether this threshold optimizes levels of walking. Indeed, there is evidence that people walk further than 400 m to access local destinations (Agrawal and Schimek, 2007, , ). Moreover, some studies find that people are willing to walk further than commonly used catchment-buffers of 400 m for buses and 800 m for trains (O׳Sullivan and Morrall, 1996, ). Other studies use public participation or soft geographic information system (softGIS) techniques to record information on distances to destinations nominated by survey participants, with findings again suggesting that people are willing to travel longer distances than the aforementioned planning guidelines (Salonen et al., 2014). To develop urban design guidelines that support sufficient levels of walking that promote health, evidence is needed about what constitutes a threshold walkable distance to increase walking to local destinations.

To address this problem, active living researchers have recommended and used a variety of approaches to define the availability of local destinations, with density and availability of destinations measured using different network buffer sizes around study participants’ homes, ranging from 400 m to 1600 m in size; and, within different administrative units such as Census Collection Districts (CCDs), census tracts, and suburbs (Brownson et al., 2009). These different approaches have contributed to different effect sizes when evaluating built environment associations with walking and can lead to bias in the measurement of spatial proximity measures (Duncan et al., 2014).

The correspondence between differing exposure effect and buffer definitions and buffer sizes is related to the measurement of the built environment in terms of scale, and spatial and social patterning which are related to both the Modifiable Areal Unit Problem (MAUP) and the Uncertain Geographic Context Problem (UCGP) (, ). When measuring spatial factors, the MAUP considers how much area (i.e., scale) is included and its configuration or zoning. The UCGP considers the discord between abstract buffer definitions of areas representing neighborhoods and areas individuals frequent in their daily lives known as ‘life- or activity-spaces’ (Kwan and Weber, 2008). To mitigate some of these issues, the definition of buffers and buffer distances should match the outcome of interest (Clark and Scott, 2014). However, despite advances in GPS technology few papers have investigated scale and differing buffer definitions when measuring the built environment for defining land uses such as destinations and their association with walking (, ).

In understanding appropriate scales for defining buffers, three problems are associated with current approaches to studying distances to destinations. First, in comparing network buffers of different sizes (e.g., ≤400 m, ≤800 m, ≤1200 m) researchers fail to consider that smaller buffers are contained within larger buffers (see Fig. 1), making it difficult to disentangle whether destinations in close proximity or those measured further away are driving the observed associations with walking outcomes. Second, a growing number of studies have considered the independent influence of destination quantity, mix and proximity (, , King and Bentley, 2015) on walking using network buffer measures without establishing threshold distances for walking. These approaches provide limited guidance to planners about the distance from homes to local destinations that may influence health (Giles-Corti et al., 2005). Finally, it is unlikely that any one destination alone encourages walking. Rather, it is the combination of destinations that increases walking (Sallis et al., 2009).

This paper seeks to address this gap in the literature by using a novel approach to evaluate how the built environment around people׳s homes changes with increasing distance and how these changes influence walking. To inform future urban design guidelines on proximity to destinations, we measured access to destinations based on what we have called ‘donut-buffers’, i.e., incremental network buffers around study participants’ homes with destinations counted between distances of 401–800 m and 801–1200 m. We compared this new method with standard network buffer approaches that count the number of destinations based on street network buffers of 400 m, 800 m and 1200 m respectively. This paper extends the work of (King et al., 2015) in examining the relationship between destinations and the frequency of walking trips.

Section snippets

Survey data

Data from the cross-sectional multilevel Victorian Lifestyle and Neighborhood Environment Study (VicLANES) were used. Data were collected in 2003 from 2349 participants aged over 18 years. Survey methods have been described elsewhere (Kavanagh et al., 2005; King et al., 2006). Data were collected on 21 inner local government areas (LGAs) and 50 CCDs located within 20 km of the Melbourne Central Business District. Using 2002 data from the Australian Bureau of Statistics, CCDs were randomly

Number of destinations by buffer type

Summary statistics for survey participants and destination types are presented in Table 1, Table 2, Table 3 respectively. Table 1 shows that a higher proportion of: females, people aged 45 or over, those with secondary or other education levels, people who were not in the labor force, and single adults with no children were more likely to undertake ≥4 trips/week. Table 2, Table 3 show that cafés/takeaway stores, transport stops and small food stores were the most prevalent. Supermarkets and

Discussion

For the network buffer models, all destination types except education and sport were associated with walking trips when counted using network buffers of 800 m and 1200 m. However, when using the network buffers it is not clear which distance is driving the association with walking trips because more proximate destinations measured using smaller network buffers are included within the larger network buffer. Only transport stops, the most prevalent destination present, was associated with walking

Strengths and limitations

This study used data with several strengths and limitations. A strength of the analysis is that a range and mix of destination types were included. However, additional destinations required for daily living that may encourage walking such as banks, pharmacies, and other retail were not available. Following current recommendations for mitigating the MAUP, another strength of the analysis is that buffers were calculated at a variety of scales (, ). Specifically, the finding that walking trips are

Conclusions

Increasing walking trips may assist local residents to increase their physical activity to recommended levels (Department of Health, 2014), which is of increasing importance considering that more than one half of Australians are insufficiently active to benefit their health. Given the importance of physical activity to reduce non-communicable disease risk, providing planners with evidence about how they can create neighborhoods that increase walking is a recognized priority (Rydin et al., 2012

References (37)

  • Australian Bureau of Statistics, 2001. Census of Population and Housing: How Australia Takes a Census: Australian...
  • Australian Electoral Commission

    Measuring the Accuracy of the Electoral Rolls and Testing the Effectiveness of Continuous Roll Update: A Report on the Fieldwork from February–March 2004

    (2005)
  • B.J. Boruff et al.

    Using GPS technology to (re)-examine operational definitions of ׳neighbourhood׳ in place-based health research

    Int. J. Health Geogr.

    (2012)
  • R.C. Brownson et al.

    Measuring the built environment for physical activity state of the science

    Urban Stud.

    (2009)
  • Centers for Disease Control and Prevention

    Physical Activity and Health: A Report For The Surgeon General

    (1996)
  • A. Clark et al.

    Understanding the impact of the modifiable areal unit problem on the relationship between active travel and the built environment

    Urban Stud.

    (2014)
  • Department of Health, 2014. Australia׳s Physical Activity and Sedentary Behaviour Guidelines for Adults (18–64). from...
  • D.T. Duncan et al.

    Examination of how neighborhood definition influences measurements of youths׳ access to tobacco retailers: a methodological note on spatial misclassification

    Am. J. Epidemiol.

    (2014)
  • Cited by (57)

    View all citing articles on Scopus
    View full text