Incorporating Probabilistic Choice Rules within Random Utility Models of Brand Choice: Theory and Empirical Illustration

Existing random utility models of brand choice behavior, in the tradition of McFadden (1974), are based on the assumption that consumers employ the decision rule of maximum utility to choose a brand. However, there is a rich body of empirical work in mathematical psychology that suggests that consumers choose probabilistically between brands (see, for example, Luce 1959). In this paper, we propose a random utility model of brand choice that is based on the decision rule of probabilistic choice, and nests the traditional random utility model as a special case. We argue that the proposed random utility model is mathematically more flexible than existing random utility models. We illustrate the empirical implementation of the proposed random utility model using real-world choice data as well as experimental choice data. We compare the empirical performance of our proposed model to the traditional random utility model. Interestingly, our proposed model outperforms the traditional random utility model (that it nests as a special case) in explaining both consumers’ real-world choices and experimental choices.