In the last years, in Japan, Genichi Taguchi developed some techniques to improve the quality of the products during its design, evolution and manufacture. His ideas have raised the attention of engineers and statisticians around the world who have proposed some alternative methods that are simpler and more efficient than Taguchi's methods. Nelder and Lee observed that Taguchi's methodology lead up to a joint model for the mean and dispersion employing generalized linear models. They showed that this class is general and sufficient to analyze Taguchi's Models. The purpose of this work is to describe joint modeling of mean and dispersion that was proposed by Nelder and Lee and the key points of the theory in a concise manner.
joint modeling of mean and dispersion; extended quasi-likelihood; Taguchi's methods
