Market Simulation Using Bayesian Procedures in Conjoint Analysis

New procedures for predicting market shares of competing products in assumed possiblemarket scenariosare proposed: Weuse Gibbs sampling in a first phase to obtain the posterior distribution of preference model parameters w.r.t. product choice-relevant attributes and levels. Then, in a second phase, this distribution is used for market simulation. We discuss advantages of the new procedures in an empirical comparison with traditional ones.