Minimum aberration blocked designs with multiple block variables (Q2227199)

Blocked fractional factorial designs are used to treat inhomogeneity of the experimental units. Mostly (and for simplicity) blocked designs with a single block variable can be found in the literature. In practice, however, inhomogeneity often comes from multiple block variables. The minimum aberration (MA) criterion is a well-accepted criterion for selecting optimal blocked designs. The authors extend the single block variable results by \textit{S. W. Cheng} and \textit{C. F. J. Wu}, ``Choice of optimal blocking schemes in two-level and three-level designs'', Technometrics 44, No. 3, 269--277 (2002; \url{doi:10.1198/004017002188618455})] to the case of multiple block variables, based on two different wordlength patterns \(W_1\) and \(W_2\). MA \(W_1\) and \(W_2\) blocked designs are sometimes different. Using a relationship between the blocking wordlength pattern of a general blocked design and a relatively small blocked design with the same block factors, they establish rules for finding MA blocked designs. Finally, they construct MA \(W_1\) and \(W_2\) blocked designs under different situations.