Sequential and Full Information Maximum Likelihood Estimation of a Nested Logit Model

Sequential estimation of a nested logit model is in general not an empirically desirable procedure, either as an alternative to full information maximum likelihood (FIML) estimation or as a source of param eter starting values for FIML estimation. Empirical studies which use varying ch oice sets across the sampled population create ambiguity in the link between seq uential and simultaneously estimated nested logit models. FIML estimation is now a computationally feasible strategy as illustrated herein.