Testing repeated measures hypotheses when covariance matrices are heterogeneous: Revisiting the robustness of the Welch-James test again

The Welch-James and Improved General Approximation tests were examined in b etween-subjects x within-subjects repeated measures designs for their rates of Type I error when data were nonnormal, nonspherical, and heterogeneous and when group sizes were unequal as well. The tests were computed with eit her least squares or robust estimators of central tendency and variability and assessed with critical values that were obtained either theoretically o r through a bootstrapping method. Prior findings indicated that one could o nly obtain a robust test of the interaction effect with the Welch-James pro cedure when sample sizes were very large. This study's results indicate tha t a robust test of the interaction effect can be obtained with reasonable s ample sizes when the Welch-James test is computed with trimmed means and Wi nsorized covariance matrices.