Multivariate Analysis of Variance (MANOVA)

Abstract

Some of the techniques of simultaneously analyzing and interpreting multi-response data under Model I of Multivariate Analysis of Variance are illustrated using a biochemical problem. The problem is to determine if one can differentiate between weight groups using various biochemical measurements. The paper deals mainly with tests of significance. Only one type of confidence interval estimation is shown. The main theoretical results in MANOVA are summarized in section 2 of the paper and applied in section 3. Some consideration is given to the choice of a subset of all the measurements for the purpose of detecting differences in weight groups. Finally, the paper points out several unsolved problems in this field.