There are many criteria in experimental design such as balance, randomness, orthogonality, efficiency, robustness, etc. In this paper we review the usefulness of uniformity in experimental design. It turns out that the uniformity should be an important criterion, which can provide additional information to all existing criteria, for comparing designs. Connections between uniformity, orthogonality, aberration, confounding, and isomorphism of fractional factorial designs are discussed. A new concept, uniformly orthogonal design, is proposed.
|Title of host publication||Multivariate Statistics|
|Editors||T. Kollo, E.M. Tiit, M. Srivastava|
|Number of pages||9|
|Publication status||Published - 1 Jan 2000|