A New Logit Model for Decision Making and its Application

Logit models attempt to answer the following question. Given a set of alternatives, each with multiple attributes, what is the probability that a decision maker will choose a certain alternative? Initially psychologists, then economists and marketing strategists, more recently transportation policy makers and design engineers have employed logit models. A brief review of these models is given in an attempt to show their relevance to more general decision-making situations, and their descriptiveness of human decision-making processes. A new model which synthesizes the desirable features of earlier models is then presented. The new model is applied to an actual case of predicting customer purchase probabilities for a particular brand of an industrial product. The prediction capability of the model is externally validated statistically, using real data, by comparing model results with the actual market share of the company.