ABSTRACT
Understanding variation is fundamental to quality improvement and custo mer satisfaction. That was realized early by Shewhart (1931) and later emphasized by, for example, Deming (1986, 1993). While Shewhart and Deming mainly concentrated on the reduction of variation by removing so-called assignable or special causes of variation, Taguchi (1986) suggested a systematic way to make products and processes insensitive to sources of variations (see also Taguchi and Wu (1980)). This strategy is usually called robust design methodology or robust design engineering; see, for instance, Kackar (1985), Phadke (1989), and Nair (1992). An important step is to identify factors, controllable by the designer or process developer, that affect the dispersion of a response variable y of interest.
