2 Multivariate analysis of variance of repeated measurements

Neil H. Timm

DOI: https://doi.org/10.1016/s0169-7161(80)01004-8

1980-01-01

Handbook of Statistics

Abstract:Publisher This chapter discusses the analysis of repeated measurements from a multivariate analysis of variance point of view. The generalization of the analysis of variance procedure to analyze repeated measurement designs utilizing the multivariate analysis of variance approach employs the multivariate general linear model and the testing of linear hypotheses using p-dimensional vector observations. In many applications of repeated measurement designs, subjects receive treatments in low-order factorial combinations, where the sequence of administration is randomized independently for each subject. Designs with multivariate observations on p variates over q conditions are called “multivariate” or “multi-response repeated measurement designs” because the multivariate observations are not commensurable at each treatment condition, but are commensurable over conditions a variable at a time. The chapter discusses the one-sample repeated measurement design, I-sample repeated measurement design, crossover/changeover design, and growth curve analysis. In the analysis of the one-group repeated measurements design from a multivariate analysis of variance point of view, no restrictions are placed on the structure of the variance-covariance matrix Σ. While the standard multivariate analysis of variance (MANOVA) model (SMM) is applicable in many experimental situations, the model has several limitations if an experimenter wants to analyze and fit growth curves to the average growth of a population over time.

English Else