Abstract: We use a finite projective geometric approach to investigate the issue of maximum estimation capacity in regular fractions of mixed factorials, recognizing the fact that not all two-factor interactions may have equal importance in such a set-up. Our results provide further statistical justification for the popular criterion of minimum aberration as applied to mixed factorials.
Key words and phrases: Complementary subset, main effect, minimum aberration, projective geometry, subspace, two-factor interaction.