Skip to content
Prev 256758 / 398506 Next

Wilcoxon rank sum in unbalanced design

On Apr 13, 2011, at 17:07 , netrunner wrote:

            
What problem? Wilcoxon tests work quite happily with unequal group sizes. There's some loss of power due to imbalance, but the only way around _that_ is to move observations from one group to the other...

(And exact testing actually solves a slightly different problem than the tie issue, but never mind.)