Modeling two-stage choice process

Consumers facing a large number of brands to choose from are known to use simplified heuristic to screen a set of relevant brands called the consideration set from the whole alternatives. Purchase decisions are then made from the brands in the consideration set. Two approaches have been suggested to model the two-stage choice process. One is to treat the consideration set as a crisp set. The other is to treat the set as a fuzzy set. The paper empirically compares the two types of models using data for soft drinks, sneakers, and departments. The results show that a model employing the crisp set approach fits the data better than that with the fuzzy set approach and better than a single-stage choice logit model.