Removing Heterogeneity Bias from Logit Model Estimation

This paper introduces a new estimation procedure for the logit model that is an extension of a conditional estimation procedure developed by Chamberlain (1978, 1980). The method eliminates heterogeneity bias which is present in the estimates produced by traditional estimation techniques. It improves the goodness of fit and explanatory power of these models, while incorporating purchase event feedback.