[R] How to calculate the power of Wilcoxon signed rank test

David Winsemius dwinsemius at comcast.net
Wed Sep 14 00:48:40 CEST 2011


On Sep 13, 2011, at 4:51 PM, Tianchan Niu wrote:

> Thank you David for your reply. But I am still confusing.
>
> I am using Wilcoxon test to compare two samples to assess whether  
> their means differ because neither of them is normally distributed.
> Is it possible to compute the power of the Wilcoxon test similar to  
> that of t test? Or is it just a wrong question?

If you formulate a null hypothesis and alternative that is appropriate  
for the test, then yes, you can ask the question "what is the power of  
the Wilcoxon signed rank test". But you need to look at the  
assumptions of the test, so you can be precise about what is being  
tested. I'm thinking you may want to pose your general, non-R question  
in a forum that is advertised to consider such tutoring. http://stats.stackexchange.com/ 
  comes to mind immediately.

>
> Thank you!
> ________________________________________
> From: David Winsemius [dwinsemius at comcast.net]
> Sent: Tuesday, September 13, 2011 4:24 PM
> To: Tianchan Niu
> Cc: r-help at r-project.org
> Subject: Re: [R] How to calculate the power of Wilcoxon signed rank  
> test
>
> On Sep 13, 2011, at 3:34 PM, tn85 wrote:
>
>> Hello All,
>>
>> I perform a Wilcoxon signed rank test for two sets of data to test
>> whether
>> they two have significantly different means. I would also like to
>> know the
>> power of this test.
>>
>
> Given that none of the various "Wilcoxon tests" are for differences in
> means, you are proposing it for the wrong purpose.
>
>
>> The third part of this tutorial is similar to what I want except  
>> the t
>> distribution.  http://www.cyclismo.org/tutorial/R/power.html#multiT
>>
>> Could anyone help? Or let me know if my question is a nonsense one.
>> Thanks
>> in advance.
>>
>
> --
>
> David Winsemius, MD
> West Hartford, CT
>

David Winsemius, MD
West Hartford, CT



More information about the R-help mailing list