Box-Cox Transformations and the Taguchi Method

We assess the capability of Box-Cox transformations for simplifying and (statistically) validating a «Taguchi analysis». We consider the possibility when a Box-Cox transformation can induce a mean bias in the variability performance measure which can inhibit the production of clearcut results, and we suggest a safeguard against this. A simple alternative method for choosing a data transformation is suggested, which ensures the determination of simple, valid and independent performance measures for the mean response and the variability in the response