Impacts of spatial accessibility on household mobility: a customized spatial Lag Model for large-scale analysis of household mobility behaviors

Abstract

Reducing Vehicle Miles Traveled (VMT) has become a crucial policy objective for mitigating transportation-related gas emissions and fostering sustainable urban transportation. This thesis applies and refines spatial statistics to understand, model, and predict household-level daily VMT. The thesis contributes to existing literature by (1) investigating the impacts of spatial accessibility to resources as well as household characteristics on VTM, and (2) capturing spatial dependencies and contextual heterogeneity across urban, suburban, and rural settings. The thesis utilizes the Travel Behavior Inventory (TBI) collected by the Metropolitan Council within Minnesota, U.S., to demonstrate the method. It integrates TBI data with spatial accessibility indicators to relate daily VMT to the levels of access to resources by bike, transit, and automobile as well as the socioeconomic characteristics of the household. The thesis developed a customized spatial weight matrix to capture varying neighborhood structures across urban, suburban, and rural settings, enabling the application of spatial statistical models that account for both spatial autocorrelation and localized behavioral patterns. It calibrates the customized spatial weight matrix using the TBI survey data to demonstrate how the customized matrix can reflect contextual variations in the study area. The methods and procedures developed in this thesis provide a scalable tool for urban planners to evaluate the impact of household characteristics and statistically predict the scale and location of changes in VMT resulting from alterations of spatial accessibility. By bridging the gap between VMT estimation and spatial accessibility evaluation, this thesis provides actionable insights for urban planning and management.