Abstract
In order to begin to explore the usefulness of the new neural networks methodology for both academics and practitioners, this paper compares its results to multiple regression's. Nine consumer behavior and demographic variables (opinion leadership, enduring involvement, perceived knowledge, objective knowledge, age, social class, marital status, race, and gender) are used to predict rock music shopping behavior of college students using both regression and neural networks. Results indicate that while the consumer behavior variables are consistently better predictors than the demographic variables as expected, the two different methods do not agree completely on which of these variables best predict shopping behavior. Furthermore, while neural networks do not provide measures of statistical significance like multiple regression, the R 2 obtained by neural networks is found in this study to be higher than that obtained by multiple regression.