[R] power 2x3 exact test

Duncan Murdoch murdoch at stats.uwo.ca
Thu May 10 00:00:27 CEST 2007


On 09/05/2007 5:11 PM, Bingshan Li wrote:
 > Hi, all,
 >
 > I am wondering if there is an algorithm for calculating power of 2x3
 > table using Fisher exact test. There is one (algorithm 280) for 2x2
 > Fisher exact test but I couldn't find one for 2x3 table. If we are
 > not lucky enough to have one, is there any other way to calculate
 > exact power of 2x3 table? The reason why I want exact power is
 > because some cells are assumed to be very small and chi square
 > approximation is not valid.

I think there are lots of possible alternatives to the null in a 2x3 
table, so you may have trouble finding a single answer to this question. 
  But assuming you have one in mind, I'd suggest doing a Monte Carlo 
power calculation:  simulate a few thousand tables from the alternative 
distribution, and see what the distribution of p-values looks like.

Duncan Murdoch



More information about the R-help mailing list