REPEATED-MEASURES DESIGN WITH ERROR SERIA L DEPENDENCE

Two analysis procedures for a random experimental design with several consecutives measures through time in each of the experimental units, are examinated in this paper. One of them consists in facing data appl ying AVAR mixed model with modelated error structure by AR processes. On the other side, the problem also is faced from a more general persp ective, using a repeated measures multivariate approach. Data were sim ulated by Monte Carlo procedures to investigate the effect that have n o satisfaction of independence and sphericity assumptions on the biase s degree of estimated parametres, on the empirical probability to make type I errors and, on the power of statistical test of the two proced ures.