Message-ID: <4B5DEBB3.706@ucalgary.ca>
Date: 2010-01-25T19:06:27Z
From: Peter Ehlers
Subject: z value from wilcox.exact
In-Reply-To: <1264435279784-1289481.post@n4.nabble.com>
netrunner wrote:
> Hi,
> does anibody know if is possible to have as ouptut from the wilcox.exact
> function also a z value to "link" the U value at the normal distribution?
I don't think that wilcox.exact() provides this directly.
I would use wilcox_test() in pkg 'coin'.
If you do want to restrict yourself to exactRankTests::wilcox.exact,
then you could use the call with 'exact=FALSE' and invert the P-value:
pv <- wilcox.exact(y ~ x, exact = FALSE)$p.value
Z <- qnorm(pv/2) # for the default 2-sided test
You will have to provide the appropriate sign for Z.
-Peter Ehlers
>
> thank you!
>
> netrunner
--
Peter Ehlers
University of Calgary