UC Irvine Institute of Transportation Studies

A generalization of the multinomial logit (MNL) model is developed for cases in which discrete alternatives are ordered so as to induce stochastic correlation among alternatives in close proximity. The model belongs to the Generalized Extreme Value class introduced by McFadden, and is therefore consistent with random utility maximization. If the true model is nearly MNL, iterative estimation on an ordinary MNL computer package provides approximate parameter estimates and a test for the hypothesized failure of the MNL'S "independence from irrelevant alternatives" assumption. A straightforward extension can handle cases where observations have been selected on the basis of a truncated choice set. The model's properties are investigated through a numerical example, and through two empirical applications whose rather unsatisfactory results are very briefly described.