ABSTRACT
This chapter considers designs in which more than one assignable source of variation is computed. It considers two multiway classifications that are the randomized complete block (RCB) and the Latin square designs. Since in the simple multiway ANOVA models, interaction effects cannot be estimated, these models should not be used when interaction might reasonably occur. The chapter discusses the process of blocking and the random assignment of treatments within blocks on experimental design. It also discusses the effect of interactions on the expected mean squares for the RCB ANOVA. When both block and treatment effects are regarded as fixed, then an additional assumption is required for an exact test of treatment effects. The assumption is that the treatment effects are the same in each block, that is, the relative performance of the treatments does not vary from block to block.
