Walkability: Which Measure to Choose, Where to Measure It, and How?

Abstract

The urban structure of neighborhoods has a decisive impact on active mobility, but this impact is hard to evaluate in a mode choice model because of the high collinearity between urban form variables and the uncertainty surrounding adequate spatial measurement parameters. Several composite scores, or walkability measures, have emerged from the literature, each using its own method and including different variables. No consensus has been reached on the size of the catchment area that should be used to measure walkability and most studies only measure walkability at the origin of the trip without considering other spatial units. In this paper, a series of four walkability measures: the Pedestrian Index of the Environment (PIE), the Walkability Index (WI), the Pedestrian Potential Index (PPI) and the Neighborhood Destination Accessibility Index (NDAI), are applied to the Greater Montréal Area to examine their correlation with the choice of walking for short trips. Several definitions of a neighborhood are tested for each measure, as well as several spatial units of measurement. A series of binary logistic regressions are then estimated using observed trip data from the 2013 Montreal Origin–Destination Survey to identify which measure, using which spatial parameters, offers the best performance in a mode choice modeling context. Measures using a diversity of urban form variables, the WI and the PIE, are found to offer the best performance, especially when measured at a scale smaller than a 1.6 km radius, while the spatial unit offering the best model fit varies between trip purposes.

Level of Pedestrian Stress in Urban Streetscapes

“…Apart from early research conducted by Mozer ( 16 ), since 1990, the literature has focused on “walkability” as the generic target to pursue to assess the pedestrian-friendliness of an urban area ( 5 ). However, walkability is a wide and open neologism commonly associated with urban conditions related to land use and density of both the population and destinations ( 3 , 58 ). Given the above definition, walkability incorporates urban characteristics that are hard to intercede (e.g., changing and mixing land uses requires substantial efforts in any city).…”

“…P-LOS has been evolving from its early stages, where user density was the only input, to its current multivariable and multimodal approach. Other indicators, such as the walkability index (2)(3)(4)(5), were introduced before the 2000s and also incorporated measures from both operational and physical attributes related to the infrastructure. However, despite these developments, stress has not been commonly used as an objective variable in the design of SPIs that evaluate the effect of measurable characteristics on this psychological response.…”

Associations between neighbourhood built characteristics and sedentary behaviours among Canadian men and women: findings from Alberta's Tomorrow Project

“…Despite being reliable and able to capture domain and context specific behaviour, self-reported measures may underestimate sedentary time relative to objective measures (e.g., accelerometers) (66). There is no agreed upon operational definition for buffer shape or size for estimating neighbourhood characteristics in relation to sedentary behaviour (28,35,36) however, our 400m Euclidian buffer neighbourhood definition may not fully capture built characteristics accessible to households and may underestimate the influence of the built environment. Similar to previous physical activity (67), travel behaviour (35) and sedentary behaviour studies (19), future research should continue to investigate the sensitivity of estimated associations between built environment characteristics and different context specific types of sedentary behaviour to different residential neighbourhood definitions (28,29).…”

“…We used Geographical Information Systems to estimate neighbourhood characteristics within a 400m Euclidean buffer around each participant's home. Just as there is no universally agreed upon set of built characteristics that should be included in walkability indices (34), so to there is no agreed upon buffer shape (e.g., network vs. Euclidean polygon) or size for estimating neighbourhood built characteristics in relation to health behaviours, including sedentary behaviour (28,35,36). Nevertheless, Koohsari et al (28) suggests that the built environment close to home may be more relevant for leisure-time sedentary behaviour.…”

“…There is no agreed upon operational definition for buffer shape or size for estimating neighbourhood characteristics in relation to sedentary behaviour (28,35,36) however, our 400m Euclidian buffer neighbourhood definition may not fully capture built characteristics accessible to households and may underestimate the influence of the built environment. Similar to previous physical activity (67), travel behaviour (35) and sedentary behaviour studies (19), future research should continue to investigate the sensitivity of estimated associations between built environment characteristics and different context specific types of sedentary behaviour to different residential neighbourhood definitions (28,29). The study included survey and built environment data that were collected in 2008, however, some built environment characteristics and the nature of sedentary behaviour could have changed since 2008 (e.g., greater of use of mobile technologies).…”