Statistics for Six Sigma Green Belts with Minitab and JMP

7.7. Kruskal-Wallis Rank Test: Nonparametric Analysis for The One-Way Anova

If the assumptions of the one-way ANOVA F test are not met, you can use a non-parametric test called the Kruskal-Wallis rank test. The Kruskal-Wallis rank test for differences in medians is an extension of the Wilcoxon rank sum test for two independent samples, which is discussed in Section 7.6. The Kruskal-Wallis rank test has the same power relative to the one-way ANOVA F test that the Wilcoxon rank sum test has relative to the pooled-variance t test for two independent samples.

You use the Kruskal-Wallis rank test to test whether independent sample groups have equal medians. To use the Kruskal-Wallis rank test, you need to assume only that the measurements are ranks ...