In this chapter the authors describe the development of closed-form models, which relax the assumption of independent and identically distributed random-error terms in the MNL model to provide a more realistic representation of choice probabilities. The rest of this chapter is structured as follows. Section 2 reviews the properties of the MNL model as a reference point for the development and interpretation for other closed-form models. Sections 3 and 4 describe models that relax the assumption of independence of error terms across alternatives and models that relax the assumption of equality of error terms across cases. Finally, in Section 5 directions for further development of these models are suggested.
Abstract