| Abstract: | With the increasing complexities involved in analyzing marketing problems, it is often necessary to use complex multivariate-statistical techniques. Although canonical analysis in many respects is similar to other multivariate techniques, it is not as easy to interpret the results as in the case of the more familiar techniques of multiple regression and discriminant analysis. This article provides insights into the problems which arise in interpreting the results of canonical analysis. Specifically, it deals with the two tests of significance and the three analytical phases–canonical correlation, extracted variance, and redundancy. |
