Travelers' preferences in multimodal networks: Design and results of a comprehensive series of choice experiments

The modeling of individuals' choice behavior in integrated multimodal transport networks requires the estimation of preference parameters related to the trade-off between uni-modal trips and multimodal combinations of private and public modes as well as relevant attributes of access, main and access stages of the trip. The stated choice method is a well-established method to estimate travel choice models empirically. However, including all relevant elements in a single experiment will not only result in choice tasks that are too complex for respondents but will also lead to choice-sets that include options that are not feasible for a given trip distance. To overcome this problem, this paper develops an approach that involves the use of a series of SP experiments to estimate a single comprehensive multimodal travel choice model. In total, four experiments are designed focusing on particular multimodal (including Park-and-Ride options) and public-transport choices for trips of varying distance. A representative national sample (N = 2746) of individuals from the Netherlands participated in the experiments through an online questionnaire. The data pooled across experiments are used to estimate the model in a scaled error-component-mixed multinomial logit framework. In this way, valuations of time, costs and service-quality attributes could be estimated on a relatively high level of detail concerning modes and trip stages. Comparisons with previous research indicate that the parameter estimates have reasonable values. The estimation results offer rich information on how travelers tradeoff between travel-time, travel-costs and service-quality attributes in travel choice in multimodal networks.