Influences of built environments on commuting and non-commuting walk trip generation in a compact city with selection bias correction

Walkable environments are increasingly important for sustainable transport, and the built environment’s influence on walking behaviour has been widely explored. However, limited studies have examined the impact of the built environment on individual walk trip generation for different purposes in walking-oriented compact cities, where zero walk trip observations may not represent non-walkers, thereby leading to sample selection bias. Thus, this study applied a two-stage modelling approach correcting for selection bias to investigate the associations of individual, household, and built environment characteristics with individual walk trip generation, including ‘non-zero walk trips’ (walking adoption) and ‘walking duration’ for both commuting and non-commuting purposes. Macao, a high-density city where walking is the primary mode of transport, was selected as the case study. The individual trip data (N = 2288) from a government travel survey were combined with built environment features for modelling. Results indicated that household car and scooter ownership discouraged commuting and non-commuting walk trip generation. Moreover, population density and land use mixture were inversely correlated with commuting and non-commuting walk trip generation, respectively. Street density increased commuting walk trip generation, whereas pedestrian bridge density facilitated non-commuting walking adoption and shorter commuting and non-commuting walking durations. Finally, based on the findings and a comparison of walk trip generation in cases with varying urban densities, policy implications for promoting walking in a compact city were presented.