Abstract
Overlap in route choice is a well-known problem in travel analysis. Because of overlap, alternatives are no longer independent (statistical effect), whereas little is unknown about how overlap may influence travelers' choice behavior. In this paper, route overlap is dealt with by considering multiple path size factors. Instead of a single path size factor for a complete route (called the basic path size factor), separate path size factors for subroutes are proposed to account for differences in valuation of the overlap between them. Three possible specifications of subroute path size factors are considered. To study the impact of separate path size factors for subroutes, path size logit models, including subroute path size factors, are analyzed and compared with estimation results obtained with multinomial logit, generalized nested logit, and basic path size logit models. It appears that including separate path size factors for subroutes significantly improves modeling performance. Furthermore, travelers appear to value overlap differently depending on when the overlap occurs during the trip. Overlap in the home end and the activity end of the trip are valued negatively, whereas overlap in the train line-haul part is valued positively, improving the trip's attractiveness. The similarity in changes in values for all parameter estimates indicates that the new path size specifications capture additional behavioral effects. The best specification of the subroute-specific path size factor is based on the length of the full route and the total number of full routes that share the overlapping legs.