Single Degree of Freedom Orthogonal Components of a Factor at 2k Levels in Terms of Linear Combinations of the 2k Contrasts of K Factors at 2 Levels

摘要

A computational procedure is developed which extends the use of Yates algorithm (for contrasts of two-level factors in an experimental design) to factors at 2k levels. By use of K dummy factors at two levels, the 2k levels of an experimental factor are represented as treatment combinations of the dummy variables. The components of the experimental factor are then related to linear combinations of the contrasts of the K dummy factors at two levels. The dependence of the linear combination on the original assignment of treatment combinations of the dummy factors is illustrated for two assignments for K = 2. (Author)