In a first part, the concepts and theory of exact randomization tests are reviewed, together with their implementation for each of a number of customary test situations including simple anova designs. Approximate (or incomplete) randomization tests are considered in the second part, as manageable alternatives to exact tests. We propose a model to calculate the relative power of approximate randomization tests and sketch out some guidelines for the user.

UR - http://www.tqmp.org/RegularArticles/vol08-3/p127/p127.pdf RP - IN FILE DO - 10.20982/tqmp.08.3.p127 DA - 2012-10-03 ER -