Multivariate Analysis of Variance

Abstract

Multivariate analysis of variance (MANOVA) is an omnibus procedure that allows for the contemporaneous analysis of more than one dependent variable. Dependent variables are the outcome variables, or criteria, of a research design. Performance on neuropsychological tests—memory scores, reaction time, processing speed, the number of words generated on tests of word fluency—can all serve as dependent variables. Interestingly, these dependent variables are often reversed and used as independent variables, or predictors, when interpreting the results of a significant MANOVA. And, scores on neuropsychological tests occasionally serve as independent variables in their own right. For example, one can empirically (e.g., median split) dichotomize performance on any one measure, say processing speed, and then compare persons who are “slow” and “fast” (the independent, or grouping variable) on a number of other dependent variables.