Testing Brand - Switching Models

In most proposed models of consumer behavior, analytical methods of parameter estimation have proven intractable. In this article a computational method for obtaining minimum chi-square estimates is presented. In this method a goodness-of-fit statistic that can be used for testing hypotheses is a by-product of the parameter estimation. A quantitative approach for discriminating among competing models of consumer behavior is also developed. Some empirical results are presented to illustrate the concepts developed.