International Choice Modelling Conference, International Choice Modelling Conference 2015

Font Size: 
Modelling travellers’ choice between park-and-ride and other modes of travel to work in the context of uncertainty
Ying Huang, Brett Smith, John Taplin, Doina Olaru

Last modified: 11 May 2015



Park-and-ride (PnR) is seen as an attractive public transport alternative in a city with low densities as it combines the efficiency of a mass transit system with the flexibility of the car (Holguin-Veras et al., 2012). Perth's newest rail lines were developed with the expectation that PnR would play a major role. However, the system is already under stress with commuters having trouble securing a bay at the railway stations (Martinovich, 2008). Moreover, commuters experience non-trivial day-to-day travel time variation in private motor vehicles and crowding in public transport depending on their departure time and unexpected occurrences during their trips. This paper explores the way in which uncertainty influences travel choices in terms of the complexity of the choice and the interaction between main travel mode, access mode and departure time, along with capturing the role of parking facilities at key railway stations.

Data collection

A household revealed preference survey was conducted in September 2013 in the catchment areas of seven train stations with PnR facilities in Perth (Western Australia). The survey included the conventional questions about commuters' origin (O) and destination (D), choices of departure time, access mode and of main mode, flexibility of work schedule, generic socio-demographic information, as well as details of travel experience in the last 20 commuting trips, regarding day-to-day variation in commute times, parking space availability at PnR stations, and the level of crowding on public transport modes. For each of the 755 responses, choice sets were derived post-survey, based on information of the transport services available for the O-D pair and the mobility restrictions faced by the commuter.

Modelling approach

Expected utility theory choice models were used to assess the importance of adding attributes associated with uncertainty in the utility functions, in order to capture commuters' reaction to travel time variation, crowding and chance of no available parking bays at railway station. 


Empirical results indicated substantial variability in commuting times with an expected value of 29min and a coefficient of variation of 0.19, varying inversely with the commuting distance. Unsurprisingly, the coefficient of variation is significantly higher for car driving and bus than for train and active transport and their combinations. Crowding is significantly related to departure time, affecting the transit conditions only between 7 and 8:30am. A similar pattern was observed for parking availability, with less than 3% chance to find a parking bay after 8am at the train stations.

The results from the discrete choice models showed that measures of uncertainty significantly improved the model fit, with the variation of travel time being significant in the utility expression for car. For public transport users, all components of travel time (access, in-vehicle, waiting, and egress) were significant in commuters' preference for train and buses. Moreover, access time had a significantly higher value of travel time savings than travel time on the public transport mode. Yet, crowding, measured as possibility of finding a seat, was not significant in choosing public transport.

For PnR users, the results indicated a negative relationship between parking space availability and the utility function. Moreover, the estimated parameter for expected parking cost at the train station was much higher than parking cost at destination, which suggests that PnR facilities play an essential role in the travel of Perth's commuters.

The investigation of socio-economic characteristics revealed that car availability and family obligations had positive effects on choosing car travel, as well as car as access mode. This may also explain the finding that women showed a stronger preference for car use. Where incentives for public transport exist (concessions for students and adults over 60 years of age), this was reflected in the choice of travel mode. Finally, departure time shapes the choice of mode for Perth commuters. Residents living in the catchment area of stations, at which the chance for finding a bay is highly competitive, were more likely to choose PnR if they departed during the pre-peak hour.


Although more complex discrete choice models are currently being estimated, results so far indicate a significant influence of uncertainty on travellers' mode choices for daily commuting in Perth. Empirical outcomes also indicate that sufficient parking space at railway stations may induce mode shifts for individuals. Notably, crowding has not been shown a significant deterrent from public transport, but this may be a reflection of the relatively low crowding in Perth, compared to other cities. Finally, improving the accessibility of railway stations helps public transport become a competitive substitute for private motor vehicles.


Holguin-Veras, J., Reilly, J., and Aros-Vera, F. (2012). New York City Park and Ride Study, Project C-07-66, Available at:

Martinovich, P. (2008). The Integration of Rail Transit and Land Use in Western Australia, Paper presented at the Conference on Railway Engineering (CORE), 7-10 September, Perth, Australia.

Conference registration is required in order to view papers.