Date: Feb 19, 2013 6:17 PM
Author: Luis A. Afonso
Subject: Re: No-Null HST . . .
Reading ?Why Permutation Tests are Superior to t and F Tests? John Ludebrook, Hugh Dudley
(The American Statistician, May 1998,Vol.52,No.2) www.citeulike.org/user/austin987/article/7571252
The A. indicates correctly (for a pair of samples 7, 5 sizes) that the number of potentially different pairs is 792, 12Chose5. However, after insisting at the practical impossibility to set all permutation samples even when the sizes are only moderately large, he does not mention that a much lower sample can be used; the results of a Permutation Tests (Fisher´s style) are in spite well representative of the difference on mean Population values.
Luis A. Afonso