fisher table probability - R-help

Wed, Feb 1, 2006 5:21 PM #

Hi, is there a way to generate the table's probability
when doing the fisher's exact test on a 2x2 table? The
fisher's exact test gives the p value, but not the
probability for the table.

Brian Ripley

Wed, Feb 1, 2006 10:52 PM #

On Wed, 1 Feb 2006, array chip wrote:

Call dhyper, as fisher.test itself does.

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

array chip

Thu, Feb 2, 2006 1:53 PM #

Thanks for the suggestion! what if the dimensions of
the table is greater than 2, say 3x4?

--- Prof Brian Ripley <ripley at stats.ox.ac.uk> wrote:

Brian Ripley

Thu, Feb 2, 2006 2:03 PM #

On Thu, 2 Feb 2006, array chip wrote:

Look at the references quoted on the help page for the formula: the 
simulation code for p-values in R-devel makes use of it and it is easy to 
compute via lgamma.

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

array chip

Thu, Feb 2, 2006 2:54 PM #

Thanks for pointing. This is my simple function for
doing this, just like to share if anyone ever needs
it:

   ## x is a rxc table
tabprob<-function (x) {
  tmp<-0
  for (i in 1:dim(x)[1]) {
    tmp<-tmp+lgamma(sum(x[i,])+1)
    for (j in 1:dim(x)[2]) {
      if (i==1) tmp<-tmp+lgamma(sum(x[,j])+1)
      tmp<-tmp-lgamma(x[i,j]+1)
    }
  }
  tmp<-tmp-lgamma(sum(x)+1)
  exp(tmp)
}

--- Prof Brian Ripley <ripley at stats.ox.ac.uk> wrote:

__________________________________________________
Do You Yahoo!?