Abstract
This paper focuses on the dynamics of consumers'consideration sets over a series of purchase occasions and suggests the new model that composes individual-level consideration sets over a series of purchase occasions. Consideration sets are the sets of alternatives from which, in making brand-choice decisions, s/he chooses one, but are not directly observable. We develop a discrete choice model with probabilistic consideration sets, and compose the change of individual-level consideration sets using purchase history data. This estimation approach jointly incorporates the history data of individuals'selected brands and the data of marketing mix. We provide empirical applications of the proposed method using the household level scanner panel data set. The results show that the reason why s/he doesn't buy our brand is whether our brand is included in consideration sets or not, taking into account the marketing mix setting.
