Consumer choice process consists of two sequential stages: first, the consumer forms a consideration set; second, the consumer evaluates each alternative in the consideration set to select the best alternative. Most existing approaches to modeling this choice process are based on the Multinomial Logit (MNL) model (e.g. Fortheringham, 1988; Roberts and Lattin, 1991; Andrews and Srinivasan, 1995; Bronnenberg and Vanhonacker, 1996).; Regardless of the approach used, current models treat consideration sets as crisp; i.e., an alternative is either considered or is not considered. Yet, almost by definition, consideration sets are unlikely to be crisp. We argue that consideration sets are fundamentally fuzzy; that is, every alternative in the universal set belongs to the consideration set to some degree .; To test the consistency of our framework with actual consumer choice behavior, we developed an operational model. We also developed a latent mixture specification that allows for the possibility of different segments of consumers with differing search behaviors and differing sensitivities of response to marketing instruments.; To calibrate our model empirically, we used a new type of data collected from an online supermarket, Peapod Inc. Our empirical analysis, using choice data from two product categories (i.e., soft margarine and liquid detergent), indicates that consumers not only use internal information search from their past purchase experience, but also engage in external information search to reduce the fuzziness in a given choice situation. We also find there is heterogeneity in the capability to process external information. Some consumers search external information (in the online store) to dramatically increase their consideration set sizes, but others do not. This finding has managerial implications for the design of online stores: providing more information does not mean better outcomes for all consumers. For some consumers, a simple shopping environment, without information overload, may be ideal.
展开▼