Size and power estimation for the Wilcoxon-Mann-Whitney test for ordered categorical data

We derive an exact variance method for the size and power calculation for the Wilcoxon-Mann-Whitney test for ordered categorical data. The O'Brien-Castelloe approximation implemented in SAS version 9.2 (SAS Institute Inc., Cary, NC, USA) is simplified. Simulation studies show that the exact variance approach tends to be more accurate than the O'Brien-Castelloe approximation and the Zhao-Rahardja-Qu method derived under local alternatives. Copyright (C) 2011 JohnWiley & Sons, Ltd.