It is increasingly recognised that park and ride (PnR) is an efficient travel mode joining private car with public transport system for providing low carbon emissions and better social equity. Departure train stations, as a transfer point of the travel mode, are paid more attention by commuters. This paper presents non-linear multinomial logit station choice models for understanding train station choice under travel time unreliability. A research framework about station choice under uncertainty is established based on discrete choice theory, cumulative prospect theory and mean-variance approach. Four weighting functions were tested for the station choice model. The data used to capture PnR users' choice behaviour under uncertainty was collected based on a stated preference experiment designed for D-efficiency and the travel time to the station was obtained from revealed preference data. The results showed that the non-linear MNL model with GE risky weighting function fits the data best. From the model, the respondents' attitude towards travel time variability was identified as risk averse. In addition, PnR users who have experienced greater travel time variations could tend to be more risk averse towards their station choice under travel time variability than those who have experienced less travel time variations. (C) 2017 The Authors. Published by Elsevier B.V.
|Journal||Transportation Research Procedia|
|Publication status||Published - 2017|
|Event||14th World Conference on Transport Research (WCTR) - Shanghai|
Duration: 10 Jul 2016 → 15 Jul 2016