Multivariate aligned rank test for interactions in multiple group repeated measures designs

This study showed that a multivariate test of interactions for aligned ranks in a split-plot design controlled Type I error rates for non-normal data with non-spherical covariance structures. Furthermore, it performed well in the presence of a strong repeated measures main effect, whereas tests performed on rank transformed scores demonstrated severely inflated Type I error rates. This test also demonstrated more statistical power than parametric tests performed on non-normal data sampled from a skewed, heavy-tailed distribution. Methods for conducting multiple comparisons are proposed.