Abstract: An outstanding issue in robust parameter design is the choice of experimental plans. Single arrays were proposed as an alternative to the inner-outer arrays advocated by Taguchi. Because factorial effects in parameter design experiments have properties distinctly different from those in traditional fractional factorial experiments, new principles on the relative importance of effects need to be considered. Based on them a new criterion is developed to discriminate among different single arrays. Search methods are proposed to find ``optimal'' single arrays with run size 8, 16 and 32.
Key words and phrases: Effect ordering principle, fractional factorial design, minimum J-aberration criterion.