A Bayesian multidimensional scaling procedure for the spatial analysis of revealed choice data

We present a new Bayesian formulation of a vector multidimensional scaling procedure for the spatial analysis of binary choice data. The Gibbs sampler is gainfully employed to estimate the posterior distribution of the specif ied scalar products, bilinear model parameters. The computational procedure allows for the explicit estimation of a covariance matrix which can accomm odate violations of IIA due to context effects. In addition, posterior stan dard errors can be estimated which reflect differential degrees of consumer choice uncertainty and/or brand position instability. A marketing applicat ion concerning the analysis of consumers' consideration sets for luxury aut omobiles is provided to illustrate the use of the proposed methodology. (C) 1999 Elsevier Science S.A. All rights reserved.