THE DEPENDENT AVAILABILITY LOGIT MODEL AND ITS APPLICATIONS (CHAPTER 26 OF TRAVEL BEHAVIOUR RESEARCH: UPDATING THE STATE OF PLAY)

Often discrete choice situations arise where the choice set is determined by the price of the options and the budgets of the decision makers. In this case, the availability of one option implies the availability of all the cheaper options, or conversely the nonavailability of one option implies the nonavailability of all the more expensive options, thereby introducing a dependency into availability in the choice set. This chapter proposes a model, referred to as the Dependent Availability Logit (DAL) model, where availability and choice is modelled jointly under the assumption of dependent availability. The Davidon-Fletcher-Powell quasi-Newton method is used to fit the model, yielding maximum likelihood estimates and their standard errors. For illustrative purposes, the model is applied to a data set for the choice of mode of travel in an intercity corridor in Egypt.