This paper investigates attitude of respondents towards a new transportation alternative namely cooperative ride system (CRS) for elderly people using 1052 samples collected in Asuke, Japan. A multinomial logit (MNL), nested logit (NL) and NL with consideration of covariance heterogeneity (COVNL) across decision makers were estimated. Additionally, three COVNL models with different transformation functions (probit, logit and C log-log) were tested for comparison. The results indicate that by

more »

... orporating the covariance heterogeneity, the COVNL model results in dramatic improvements in terms of better fitting to the data and behaviorally representing the choices of decision makers. Moreover, among three transformation functions, the C log-log function is the best while the probit function is the worst ones for data fitting. Evidence from model estimation results shows that the potential volunteer drivers for the CRS are the people living in a single household, currently using cars and having more frequencies of driving and traveling. Most importantly, the elderly people, whom the CRS is proposing for, are found to significantly participate as the riders. Furthermore, females, students and the people earning low annual income possibly join as the riders. Interestingly, the people who gained experiences of being the volunteer drivers or riders have higher willingness to take part in the CRS.