ABSTRACT
While the block designs so far were complete in the sense that all treatments could be applied in each block, we now present the case of incomplete block designs where only a subset of all treatments can be applied in each block. We discuss which approaches can be used to select the different subsets of treatments and what properties these designs have. One example is the so-called balanced incomplete block design, where all treatment differences can be estimated with the same precision. Using example data, we show how the classical ANOVA models can be used for model fitting and what happens if we use a mixed model approach.
