An analysis of the built environment and auto travel in Halifax, Canada

This study tests whether the built environment influences total distance traveled by auto for non-work trips on a weekday. Using cross-sectional data from Halifax, Canada, we identify a strong causal relation between the built environment and auto distance traveled by worker and non-worker for a selection of non-work travel. We apply linear regression and the spatial lag model to control for spatial autocorrelation and find that the presence of built environment variables in linear regression handles the autocorrelation problem. We use attitude variables to control for residential self-selection. The importance of measuring the built environment near home and workplace is demonstrated. Also, we find that an empirically derived geographical scale of measuring the built environment outperforms the commonly used quarter-mile scale. The study demonstrates the importance of selecting a suitable set of trips for travel behavior-built environment analysis and suggests that future studies should classify travel based on trippurpose and examine what types of trips are influenced by the built environment.