Existing stated preference models in the transportation literature focus principally on measuring preferences for travel alternatives. Choices are predicted by making ad hoc and possibly incorrect assumptions regarding the relationship between preference structures and choice behaviour. In contrast, stated choice models are derived from choice data observed under hypothetical conditions. These models provide a powerful approach to testing simultaneously the assumed choice model and specification of the implied utility function. Nevertheless, conventional stated choice models are based on the rigorous assumption that the nonavailability of a particular travel alternative does not affect the utility and relative choice probability of any other travel alternative included in a choice set. How designs that permit the estimation of such availability effects can be constructed is indicated. A case study on mode choice behaviour in the Eindhoven region, the Netherlands, suggests that choice models incorporating such availability effects can improve the predictive success of mode choice models. The results suggest that people's preferences for choosing the car to commute are only slightly influenced by the availability of modes of public transportation.
Abstract