A walk trip generation model for Portland, OR

This study proposes a home-based walk trip generation model, based on the built environment around households, controlling for sociodemographic influences. Two-stage hurdle models are estimated based on a household travel survey in Portland, Oregon. The first stage predicts the probability of households making any home-based walk trips. The second stage predicts the number of home-based walk trips for the subset of households that make such trips. The study also tests built environment variables for three different buffer widths around household locations to see which scale best explains walking behavior. The results show that sociodemographic characteristics are strong predictors of walk trip generation. Specifically, household size, income, and number of workers in the household influence the probability of a household having any walk trips, while household size and number of children in the household affect the number of walk trips made by the subset of households making walk trips. Characteristics of the built environment are also significant. Activity density, transit stop density, employment accessibility, intersection density, and most interestingly, sidewalk quality are associated with the decision to walk as a mode of travel, while land-use entropy, transit stop density, employment accessibility, sidewalk quality, and traffic calming and signal are predictors of the number of walk trips made by households making walk trips. Sidewalk quality is represented by a single principal component that neatly captures the common variance in an array of sidewalk variables. To our knowledge, this is the first walk trip generation model to include a measure of sidewalk quality. (C) 2017 Elsevier Ltd. All rights reserved.