摘要
In this paper, we give a modification of the well-known minimum aberration (MA) criterion for finding “good” regular 2m−k fractional factorial designs. In particular, for given values of m and k, we consider a two-stage criterion where in stage one all regular 2m−k designs that yield the maximal number of contrasts for estimating main effects and two-factor interactions are determined and then within this subclass, the MA design is found. Examples are given to compare and contrast the MA criterion with the modified MA criterion and then tables are obtained for selected values of m and k where the best designs under both criteria are given.