R help - Aug 2006 - exact Wilcoxon signed rank test with ties and the "no longer under development" exactRanksumTests package

Dear List,

after updating the exactRanksumTests package I receive a warning that
the package is not developed any further and that one should consider
the coin package.

I don't find the signed rank test in the coin package, only the Wilcoxon
Mann Whitney U-Test. I only found a signed rank test in the stats
package (wilcox.test) which is able to calculate the exact pvalues but
unfortunately the procedure cannot calculate the exact values with ties.

Is there any other package that is providing a similiar test? Or is
there an easy work out how to take the ties into account? (Or a chance
that the correction is taken into account for the stats package?)

Stefan Grosse

Take the following example from Bortz/Lienert/Boehnke:
> x1<-c(9,14,8,11,14,10,8,14,12,14,13,9,15,12,9)
> x2<-c(13,15,9,12,16,10,8,13,12,16,9,10,16,12,9)
# exactRankTests package:
> wilcox.exact(x1,x2,paired=TRUE)
        Exact Wilcoxon signed rank test

data:  x1 and x2
V = 13, p-value = 0.1367
alternative hypothesis: true mu is not equal to 0

# wilcox.test by stats package:
> wilcox.test(x1,x2,paired=TRUE,exact=TRUE)
        Wilcoxon signed rank test with continuity correction

data:  x1 and x2
V = 13, p-value = 0.1436
alternative hypothesis: true mu is not equal to 0

Warning messages:
1: cannot compute exact p-value with ties in: wilcox.test.default(x1,
x2, paired = TRUE, exact = TRUE)
2: cannot compute exact p-value with zeroes in: wilcox.test.default(x1,
x2, paired = TRUE, exact = TRUE)

On Fri, 25 Aug 2006, Stefan Grosse wrote:
> Dear List,
>
Stefan,
> after updating the exactRanksumTests package I receive a warning that
> the package is not developed any further and that one should consider
> the coin package.
>
> I don't find the signed rank test in the coin package, only the
Wilcoxon
> Mann Whitney U-Test. I only found a signed rank test in the stats
> package (wilcox.test) which is able to calculate the exact pvalues but
> unfortunately the procedure cannot calculate the exact values with ties.
>
indeed, this is the only gap to be filled in `coin', I just haven't had 
the time to implement this. And of course, `exactRankTests' is still 
available and will be available in the future. So you _can_ use it!

The message just means that I'm not going to add _new_ features
to `exactRankTests' and that we as the authors of both packages
believe that the `coin'-way of doing things is more appropriate.

Hope that helps & sorry for the confusion!

Torsten

ps: please cc emails about packages to the maintainer.
> Is there any other package that is providing a similiar test? Or is
> there an easy work out how to take the ties into account? (Or a chance
> that the correction is taken into account for the stats package?)
>
> Stefan Grosse
>
> Take the following example from Bortz/Lienert/Boehnke:
>
>> x1<-c(9,14,8,11,14,10,8,14,12,14,13,9,15,12,9)
>> x2<-c(13,15,9,12,16,10,8,13,12,16,9,10,16,12,9)
>
> # exactRankTests package:
>
>> wilcox.exact(x1,x2,paired=TRUE)
>
>        Exact Wilcoxon signed rank test
>
> data:  x1 and x2
> V = 13, p-value = 0.1367
> alternative hypothesis: true mu is not equal to 0
>
> # wilcox.test by stats package:
>
>> wilcox.test(x1,x2,paired=TRUE,exact=TRUE)
>
>        Wilcoxon signed rank test with continuity correction
>
> data:  x1 and x2
> V = 13, p-value = 0.1436
> alternative hypothesis: true mu is not equal to 0
>
> Warning messages:
> 1: cannot compute exact p-value with ties in: wilcox.test.default(x1,
> x2, paired = TRUE, exact = TRUE)
> 2: cannot compute exact p-value with zeroes in: wilcox.test.default(x1,
> x2, paired = TRUE, exact = TRUE)
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
>