Studying Consideration Effects in Empirical Choice Models Using Scanner Panel Data

The authors develop a dynamic consideration set formation model, complementary to the static model developed by Roberts and Lattin (1991), which appears to improve the multinomial logit model significantly.

Abstract

The authors develop a dynamic consideration set formation model, complementary to the static model developed by Roberts and Lattin (1991). The model does not require direct consumer reports of consideration and therefore can be applied to scanner data. Explicitly incorporating consideration effects appears to improve the multinomial logit model significantly.