Abstract
The development of behaviourally richer representations of the role of well-established and increasingly important influences on modal choice, such as trip time reliability and accounting for risk attitude and process rules, has moved forward at a fast pace in the context of automobile travel. In the public transport setting, such contributions have, with rare exception, not been considered. In this paper, we discuss and empirically illustrate the merits of advanced modelling developments aimed at improving our understanding of public transport choice, namely the inclusion of reliability in extended expected utility theoretic forms, to recognize risk attitude and perceptual conditioning, the consideration of passenger crowding and its inclusion in linear additive models, and the role of multiple heuristics in representing attribute processing as a way of conditioning modal choice. We illustrate the mechanics of introducing these behaviourally appealing extensions using a modal choice data set collected in Sydney.
Acknowledgements
This paper is a contribution to the research programme of the Volvo Research and Education Foundation Bus Rapid Transit Centre of Excellence. We thank the many colleagues who have contributed to our joint research efforts in the areas of activity presented in this paper. We especially thank Bill Greene, Andrew Collins, David Layton, Caspar Chorus, Stephane Hess, Riccardo Scarpa, Corinne Mulley, and Matthew Beck. An earlier substantially different version was presented at the Twelfth International Conference on Advanced Systems for Public Transport (www.caspt.org), Santiago, Chile, July 23–27, 2012. The comments of three referees are appreciated and have been most helpful in improving the paper.
Notes
Even though it has been recognized that travel time reliability is important to public transport users (Cantwell et al., Citation2009; Rietveld, Bruinsma, & van Vuuren, Citation2001) and hence should be given greater emphasis in transport policy and included in patronage forecasting and appraisal studies.
The reliability ratio is defined as the ratio of the value of saving one minute of the standard deviation of travel time to the value of reducing 1 min of average travel time.
This model was first introduced as a contextual concavity model by Kivetz, Netzer, and Srinivasan (Citation2004), who used it to model a specific phenomenon known as extremeness aversion. They made the prior assumption that relative to the worst-performing attribute, utility is concave in the gains. This assumption is empirically testable and we find that it does not always hold (Leong & Hensher, Citation2011). Hence, it may be more useful to label such a functional specification as an NLWLR model instead.