Attribute thresholds and logit mode-choice models

In this paper the concept of minimally perceived attribute differences is introduced into a logit choice model. For estimating the parameters of the model, maximum likelihood is employed and an experimental test is carried out on a sample of trip makers going to the Melbourne central business district. It was found that the average respondent required a 12-min (22 percent) difference in travel time or a 12-cent (32 percent) difference in travel cost before he would react to the variation in attribute ratings. The model is compared with a more traditional logit model with a linear additive measure of utility.