[R] Interpreting the result of a Wilcoxon (Mann-Whitney U) test

Filipe Correia fcorreia at gmail.com
Thu Oct 3 11:13:00 CEST 2013

Hi Petr,

On Thu, Oct 3, 2013 at 8:01 AM, PIKAL Petr <petr.pikal at precheza.cz> wrote:
>> Well, I guess I'm asking all of the three... :) Not with the intent to
>> choose which one fits, but because I was experimenting and making sure
>> I really understand how I could use R's implementation of the Wilcoxon
>> test. But I'm intrigued that the answer to "are a and b different" is
>> not consistent with "is a less than b"... Maybe my understanding of the
> Your understanding of statistical test is not correct. When asking /a is not equal b/ you are "splitting" probability of alternative hypothesis to both sides, with asking /a is lower than b/ probability of alternative hypothesis is only at one side of distribution. It is really not a question of R but a question of basic statistics.
> If some program gave you same answer to those questions you shall get rid of it immediately.

I see what you mean. Makes sense! Thank you.


More information about the R-help mailing list