Disaggregate choice models of travel behaviour in Australian cities

The methodology and structure of disaggregate choice models - discriminant functions and multinomial logit models is described. Using home interview survey data of over 2000 households in Canberra, a number of models of vehicle ownership and travel mode choices have been developed. The modelling exercise identifies the relative effect of factors influencing choice or variables which contribute most to differentiate individuals between alternative choice groups. This is a crucial input to policy design. The statistical analysis includes tests for quality of all models developed in this study by indicating how well each model is capable of predicting the right choice group. The results show that these models are useful in urban transport planning (a).