A Two-State Model of Purchase Incidence and Brand Choice

Randolph E. Bucklin and James M. Lattin

The authors develop and test a probabilistic model of purchase incidence and brand choice for frequently purchased consumer products. The model incorporates two ways of shopping in a category. Shoppers who have planned their purchasing (made a decision before entering the store) do not process in-store information and show no response to point-of-purchase promotions. Consumers who have not planned their purchasing in a category (deciding at the point of purchase) may process in-store information and may be strongly influenced by promotions. The two modes of information processing are called decision states and are labeled planned and opportunistic, respectively. The two-state model is calibrated on IRI scanner purchase records for saltine crackers. The model yields a significantly better fit than a one-state nested logit model and provides new insights into the relationship between shopping behavior and consumer purchase response.

(Brand Choice; Purchase Incidence; Nested Logit; Promotion; Shopping Behavior)