Hi All
I have some paired data that I would like to compare (they are actually percentages) that are clearly not Normal so I was planning on using a Wilcoxon signed rank for paired data. I am interested in doing a power/sample size analysis, however was unable to find any documentation or formulas or code to do this.
Doe anyone have any docs or references or code to do this? I am familiar with the power and sample size calculations for a paired t test and the like and was wondering if there was a way to apply a similar approach to the Wilcoxon test statistic W as for larger N under the null hypothesis W is Normally distributed with mean 0, however what would the distribution of W be under the alternative hypothesis?
Any thoughts or pointers most welcome
Best Regards,
Richard
_________________________________________________________________
It’s a talkathon – but it’s not just talk.
http://www.imtalkathon.com/?source=EML_WLH_Talkathon_JustTalk
|