Abstract
Tokyo Metropolitan Area (TMA) has one of the densest railway networks. Railway passengers can choose one route from among several alternative mutes in consideration of the level of whole railway services from origin to destination. On the other hand, it is found that there is a similarity among the alternative routes. It suggests that the non-IIA route choice models (e. g., Multinomial Probit model with structured covariance matrix, Mixed Logit model, C-Logit model etc.) need to be applied to the analysis of the railway route choice behavior in the TMA. The characteristics of the models, however, have not been fully examined This study (1) compares with three non-IIA route choice models mentioned above in order to illustrate the characteristics and (2) proposes C*-Logit model on the basis of the characteristics by improvement of C-Logit model.
