Nonparametric Version of Williams' Test for a Randomized Block Design

摘要

A large-sample nonparametric version of Williams' test for comparing increasing doses of a substance with a zero-dose control response is given for observations from a randomized block experimental design. The method is based on Friedman-type ranks. Williams' test and the nonparametric equivalent for a one-way design are briefly described. The validity of the new method is shown and a numerical example is given.