TYPE-I ERROR RATES AND STATISTICAL POWER FOR THE JAMES 2ND-ORDER TESTAND THE UNIVARIATE F-TEST IN 2-WAY FIXED-EFFECTS ANOVA MODELS UNDER HETEROSCEDASTICITY AND OR NONNORMALITY/

Type I error rates and statistical power for the univariate F test and the James second-order test for comparing population means were estim ated for the two-factor fixed-effects completely randomized design. Wh en variances were heterogeneous, the univariate F test was shown to ha ve actual Type I error rates greater than the nominal value even when sample sizes were equal, The James second-order test controlled the Ty pe I error rate under heteroscedasticity when population distributions were normal or platykurtic. However, the James second-order test had increased the Type I error rates when the population distributions wer e skewed, When variances were homogeneous and distributions were norma l, the univariate F test and the James second-order test had similar s tatistical power for testing main effects, but large power differences favoring the univariate F were found for testing interaction effects.