The analysis of repeated measurements: Univariate tests, multivariate tests, or both?

Abstract

One strategy which has been recommended for examining effects in repeated measures designs combines a degrees of freedom (d.f.)‐adjusted univariate F test and a multivariate test. The results of a simulation study for a groups × trials repeated measures design are presented, and demonstrate that for balanced designs this combined strategy rarely provided superior Type I error control to that afforded by uniformly adopting either a univariate or multivariate test. For unbalanced repeated measures designs, all three strategies typically resulted in very conservative or very liberal Type I error rates when covariance matrices and group sizes were positively and negatively paired, respectively. For such designs, the approximate d.f. Welch‐James (James, 1951, 1954; Welch, 1947, 1951) multivariate statistic developed by Johansen (1980) is preferable to any of the other strategies investigated in this paper. Power comparisons also favour the use of the Welch‐James test.