Power comparison of ANOVA and Kruskal‒Wallis tests when error assumptions are violated

The effects of the violations normality and homogeneity variances assumptions on power one-way ANOVA F-test is studied in this paper. Simulation experiments were conducted to compare parametric with non-parametric Kruskal‒Wallis (KW) test normal/non-normal, equal/unequal scenarios sample group means. Each these 184 simulation was replicated N = 1000 times obtained for both F KW tests. Shapiro‒Wilk's Bartlett's/Levene's tests each experiment. Results show that outperformed those F-tests 92 (85/92) non-normal cases. Although higher than 85 out under assumptions, differences, all cases study are not significant (p > 0.05) using t sign Based results, favours as a more robust safer use rather especially when distributional data sets doubt.