The meaning of probability in relation to Fisher's exact test (Q1093282)

The problem of finding the second tail in Fisher's exact test is analysed with an experimental approach. Equations are derived to predict what would happen if Fisher's model were simulated experimentally and the total P found by applying \textit{J. O. Irwin}'s rule [Tests of significance for differences between percentages based on small numbers. ibid. 12, 83- 94 (1935)] after a varying number of trials, M. These equations reveal a crucial property of the test. For most significance tests P values determined experimentally in this way do not change systematically with M, but they do in Fisher's test..