Compound Symmetric Error Structure
The experiments considered in the sequel of this book possess a compound symmetric error structure. This is due to the fact that the experiments, which fall into the category of bi-randomization designs or two-stratum designs, suffer from a restricted randomization. The category of bi-randomization designs or two-stratum designs is a subset of the so-called multi-stratum designs. In this chapter, we provide the reader with a number of examples and give a general description of this type of design. Next, we show how to write the information matrix of a bi-randomization experiment as a sum of outer products without using Cholesky decomposition. This is important for the design construction algorithms in the next chapters. Finally, we will demonstrate that the asymptotic information matrix of a bi-randomization experiment is a reliable approximation to the finite sample information matrix.
KeywordsDesign Point Information Matrix Generalize Little Square Plot Variable Plot Factor
Unable to display preview. Download preview PDF.