Calibrating walkability indicators for commute walk trips: A structural equation modeling approach

Abstract

Walking plays a significant role in promoting public health, decreasing air pollution, and vitalizing cities. Walkability is an important notion that transportation planners and policymakers often use to quantify the suitability of and to promote travelers’ propensity for walking in an urban area. This paper adopts a data-driven approach using structural equation modeling for simultaneous estimation of 1) a latent construct of walkability using different walkability indicators (thus estimation of weights of the indicators), and 2) the effect of walkability on commute walking behavior for an urban area. Doing so accounts for the effects of various indicators of walkability as well as socio-economic and security characteristics. A Partial Least Square Structural Equation Model is developed to estimate both direct and indirect effects of walkability on the share of commute walk trips in total commute trips for the city of Chicago using census tract-level data. We find that walkability is the most important factor affecting the share of commute walk trips. Among the indicators of walkability, transit- and job density-related indicators are the most important ones. Some socio-economic factors, especially travel time, vehicle ownership, and college student share, also critically influence the propensity for walking to commute. We further find that our calibrated walkability score results in a different hotspot distribution in Chicago than using the uncalibrated walkability score, as is currently practiced by the Chicago Metropolitan Agency for Planning, but is more consistent with hotspot distribution of the actual commute walk trips in the city. These findings advance the understanding and characterization of walkability and actual walking behavior in urban areas, which in turn helps inform pedestrian infrastructure investment and policy making.