Influences of built environment characteristics and individual factors on commuting distance: a multilevel mixture hazard modeling approach

Concerns over transportation energy consumption and green-household gas (GHG) emissions have prompted a growing body of research into the influence of built environment on travel behavior. Studies on the relationship between land use and travel behavior are often at some aggregated spatial unit such as traffic analysis zone (TAZ), spatial issues occurs among individuals clustered within a zone because of locational effects. However, recognition of spatial issue in travel modeling was not sufficiently investigated. The object of this study is twofold. First, a multilevel hazard model was applied to accommodate the spatial context in which individuals generate commuting distance. Second, this research provides additional insights into examine the effects socio-demographics and built environment on commuting distance. Using Washington metropolitan area as the case, the built environment measures were calculated for each TAZ. To estimate the model parameters, the robust maximum likelihood estimation method for a partial function was used, and the model results confirmed the important roles played by the neighborhood and individual level factors in influencing commuting distance. Meanwhile, a comparison among the general multilevel model, single level and multilevel hazard models was conducted. The results suggest that application of the multilevel model obtains significant improvements over traditional model. The significant spatial heterogeneity parameter indicates that it is necessary to accommodate the spatial issues in the context of commuting distance. The results are expected to give urban planners and policy makers a better understanding on how the neighborhood and individual level factors influence the commuting distance, and consequently develop effective and targeted countermeasures.