Abstract
Repeated measures designs with autocorrelated errors.- The repeated measures designs lave been traditionally analyzed using the univariate mixed model of AVAR. however when the treatments are not presented in a randomized order, the sequence of observations introdoce correlation in the errors of model. Under the assumption that the serial correlation can be modeled through relatively simple ARMA processes, this paper considers the problem of obtaining efficient and consistent estimators. Finally, a example that illustrates the process that has been described is presented.