← All apps

Wilcoxon rank-sum test

Rank all observations together and add up the ranks of group A: that sum is the statistic W. If the groups do not differ, the group labels are just a random shuffle. Shuffle them yourself to build up the distribution of W under the null hypothesis.

Observed W
Shuffles
Proportion with W ≤ observed
estimated one-sided p-value

Observations and their ranks

group A group B

Distribution of W under the null hypothesis

shuffles with W ≤ observed other shuffles