Kruskal-Wallis

Return to Behavioral Research Methods

This is the non-parametric alternative to the One-Way ANOVA.

Use this when your data is ordinal AND you have more than 2 groups that you want to compare.

The null hypothesis claims there is no tendency for ranks in any treatment condition to be systematically different from ranks in any other treatment condition.

The alternative hypothesis claims ranks in at least one condition are systematically different than ranks in another condition.

Unlike the One-Way ANOVA, there are no Post Hoc tests so there is no way to statistically identify where those differences may be located. You have to eyeball it.

Kruskal-Wallis in SPSS

 * 1) Click on 'Analyze' -> 'NonParametric Tests' -> 'K Independent Samples'


 * 1) Move your ranking into the 'Test Variable' box


 * 1) Move your group (or category) into the 'Grouping Variable' box


 * 1) Define the range of the grouping variable


 * 1) Choose 'Kruskal-Wallis H' under 'Test Type'


 * 1) Click 'OK'


 * 1) Your output should appear