Simplification procedures for a probabilistic choice model

A standard assumption of the classical multinomial logit model is that an individual who has to choose an alternative among a set of available alternatives compares the utilities of all these choices. However, in some situations this procedure can be costly in time and money. In order to account for this feature, the model is generalized to the case in which the individual decides to simplify his decision by drawing a limited number of alternatives from the full choice set. The conditional and marginal choice distributions are derived, and some connections with other generalizations of the model are analyzed.

Keywords::

• multinomial logit model
• Gumbel distribution
• multivariate multinomial logit model
• general extreme value model