Inferential aspects for the optimal selection of the control parameters in Taguchi method

This article focuses on the estimation of dispersion effects in off-line quality control techniques. In this context, the Taguchi design for the optimal choice of process parameters is one of the most commonly used statistical methods. Starting from Taguchi methodology, we consider that an additive or a multiplicative model defines the relationship between the deterministic component and the variability of the process. We apply a hypothesis testing in order to find the optimal factor combination that minimizes the variability indicator of product quality, using ranking and selection methods of the Bechhofer kind. Moreover, an extensive simulation study shows how the probability of finding the optimal set of factors changes according to the main sampling parameters, in order to provide guidance for practitioners.