SOME ALTERNATIVE APPROXIMATE TESTS FOR A SPLIT-PLOT DESIGN

Alternative tests are presented for the between x within interaction null hypothesis and for two within-subjects main effects null hypotheses in a split plot design. Let mu(j) (j = 1, ..., J) denote a (K x 1) mean vector for group j, n(j) the number of subjects in Group j, [GRAPHICS] and One main effect null hypothesis was H01: C muBAR = 0 where C is a (K - 1) x K matrix with full row rank and rows that are contrast vectors; the second was H02: C muBAR* = 0. These tests do not require the assumption that the dispersion matrices for the several groups are equal. Estimated Type I error rates for the interaction tests and for several tests of H02 are reported.