SOME IMPROVEMENTS TO CURRENT PRACTICES OF ESTIMATING INDIVIDUAL TRAVEL CHOICE MODELS WITH EXISTING DATA

This paper was presented at Session 7: Modelling Workshop. Much money has been spent on the collection of comprehensive household travel (HIS - home interview survey) data in Australian cities which has in turn been used, and will continue to be used in the foreseeable future, for the estimation of travel demand models by transport planning authorities. The proliferation of multinomial logit (MNL) and other estimation/application packages also means that individual travel choice models will continue to be prominent. In this event it is worth ensuring that maximum value is obtained. Improvements to current individual travel choice modelling may be achieved in at least three aspects: first, if the Lancastrian paradigm of consumer behaviour is strictly followed and only characteristics of the transport system are included in the utility function, but the influence of different individual attributes is allowed for by appropriate market segmentation; second, if the unit of analysis adopted is the journey (ie the round trip from home back to home) and not the trip; third, if choice sets are carefully selected, which in the case of the MNL model could result in a hierarchical decision process for the journey-to-work modal choice. The Ballarat HIS data was used for the exercise and resulted in following current practices. (TRRL)