Exploration of Walking Behavior in Vermont Using Spatial Regression.

摘要

This report focuses on the relationship between walking and its contributing factors by applying spatial regression methods. Using the Vermont data from the New England Transportation Survey (NETS), walking variables as well as 170 independent variables are derived including some through spatial analysis with Geographic Information Systems (GIS). Among those independent variables, people’s lifestyle and perception of the built environment variables are included. A linear regression model is first established to serve as a base model for comparisons with spatial regression models. The results reveal that people’s lifestyle and perception of the built environment are significant variables explaining Vermonters’ walking behavior. Methodologically, the results reveal that no spatial effect is found and that there are no significant differences between the linear and spatial regression models. Therefore, the study concludes that it may be appropriate to apply traditional non-spatial statistical tools to analyze the relationship between walking and its contributing factors. However, the study suggests that researchers examine whether spatial effect exists in these inherently spatial behaviors before using only traditional statistics. This caution is particularly relevant as methods to estimate spatial models become more commonplace and easily available. The two spatial methods used in this report both reveal small but different challenges.