Examining the difference between park and ride and kiss and ride station choices using a spatially weighted error correlation (SWEC) discrete choice model

Abstract

This paper presents a novel discrete choice model formulation: the spatially weighted error correlation (SWEC) logit model for spatial location choices. This model captures both the correlation between spatially distinct alternatives based on the relative distance between them and the heteroskedasticity of the errors of the alternative as a function of their relative distance to the decision maker. The SWEC model is applied for the estimation of models of transit station choice for P-and-R as well as kiss and ride (dropped off at transit station) transit commuters in the Greater Toronto and Hamilton Area (GTHA). The kiss and ride model has particular relevance due to its ability to capture household tradeoffs made by both the driver and the passenger. The tradeoffs are captured by utilizing the subsequent trip made by the driver in the utility function specification. The proposed model structure provides additional insights into how station choice occurs for such complex trips. Finally, the application of the SWEC model for both choice contexts provides a fundamental improvement over conventional approaches, as it is able to capture non-proportional substitution patterns and heteroskedasticity inherent with spatial choices.