The derivation is described of a probability model of travel mode choice for the work journey in terms of the differences in costs and times between the modes available for each individual's journey. The work is based on surveys carried out at two offices in central london, with some test data from a survey of workers in central leeds. Regression techniques were used to establish simple linear relationships between the probability of using a car and the cost and time differences between a car and the best available public transport route. Reasonable correlation coefficients were obtained. Limited test of the models have reproduced existing mode choices to a high degree of accuracy. Relationships were also established between the regression coefficients and income. A generalized model is derived from these relationships. The basic model has a serious fault in that because it is linear, it permits values of probability to be obtained that exceed unity, or are less than zero. A logistic transformation is put forward to correct this and is used on the generalized model. The resulting model still appears to reproduce existing conditions to a similarly high degree of accuracy. The model appears to be reasonably behavioral, and could form the basis of a new and more accurate mode choice procedure. /author/.
Abstract