Built environment in local relation with walking: Why here and not there?

Walking, as both a major mode of transport and the most common form of every-day physical activity, deserves further attention in health-related transportation studies. In this paper, we focused on the built environmental correlates of walking for errands and leisure in a sample of 4979 adults (Paris, France) through a cross-sectional study based on an internet survey. The main aims were (i) to delineate places with contrasting relationships between the built environment and walking, using geographically weighted regression models and (ii) to determine what differentiated the contexts we uncovered, in terms of both environmental and individual characteristics, using canonical discriminant analyses. Our results showed that the spatial heterogeneity of relationships between walking and the built environment occurred across the entire studied area and concerned the two walking outcomes, with odds-ratios (ORs) ranging from 1.01 to 1.31 and from 1.07 to 1.35 for walking for errands and leisure, respectively. We suggest that the spatial patterning of convergent relationships is due to contextual effects, i.e., the effect of places with specific intrinsic arrangement of environmental and individual features. Data-driven identification of local contexts should be a key step in future contextual analyses of walking and health-related outcomes. •Spatial nonstationarity characterizes relations between built environment and walking.•We assessed local contexts where built environment-walking relationships vary.•Most discriminant variables for leisure walking were facility density and individual housing.•Most discriminant variables for errand walking were unemployment and education.•Built environment effects on individual behaviors should be examined at local scale.