Authors: Dr. Mingyue Sheng and Prof. Basil Sharp
As well documented in the literature, urban form plays an essential role in determining transit ridership. However, among these studies, the majority of empirical work has not considered space as a relevant factor. Instead, most of the findings are based on a strong assumption that there is no spatial effect across the research area. This general negligence of spatial effects will, in turn, produce biased estimators if substantial geographical patterns exist. Given the observational heterogeneous distribution of transit patterns in the Auckland region, it is exceedingly doubtful whether the assumption of no spatial interdependence is valid. Based on cross-sectional data, mainly extracted from the New Zealand (NZ) 2006 census with additional geographical information compiled by ArcMap for the Auckland region, this paper contributes to the existing literature by offering insight into the spatial structure of the current public transport sector. The use of a spatial Durbin model provides a better understanding of the urban form factors that influence bus mode share by decomposing the total effect of one explanatory variable into direct and indirect effects. The results show that the total effects are comprised mostly of spatial spill-over impacts. In addition to urban form variables, several other dimensions of potential bus mode share predictors are considered, including transit supply quality, accessibility to other modes of public transport, plus variables that describes household characteristics.
Keywords: Spatial dependence, spatial Durbin model, spill-over effect, urban form, transit behavior