ABSTRACT
Alternative approaches making better use of overall clinical trial information, such as the covariance structure, would be useful. The approaches that follow assume multivariate normality. Under this assumption, for two-sided alternatives, Hotelling's T2 test or a X2 test are the classical tests. However, in clinical trials, the alternative hypothesis is one-sided, and the use of this more restricted alternative hypothesis will result in significant improvements in power (Robertson et aI., 1988).
