A simulation study on the choice of transformations in Taguchi experiments

Abstract

In this paper we examine two widely used methods to obtain a transformation in Taguchi experiments, namely the lambda plot and the beta technique. We consider different situations with contrasts in uencing the mean and or the variance of the response. Our simulation study reveals that the variation of the L-Method proposed by Kunert and Lehmkuhl (1998) is a good compromise in terms of meeting the confidence level and identifying active effects.