FOUR METHODS OF ANALYZING BETWEEN VARIATION FOR THE K-GROUP MANOVA PROBLEM

Abstract

Four methods of analyzing between variation in the k-group MANOVA problem are considered. Specifically, they are (1) discriminant analysis, (2) stepdown analysis, (3) contrasts on the classification variable and (4) a two group breakdown. These methods are discussed briefly, with certain basic com- puting equations being given. Then a sample problem is used to illustrate how the methods may be applied and interpreted. The univariate h2 (squared correlation ratio) is related t o a corresponding multivariate measure. Also, a variant of h2 is introduced to indicate the proportion of variance in each of the dependent variables which is accounted for by the classification variable.