15 Spatial experimental design

Abstract

Publisher Summary This chapter focuses on spatial experimental design. The statistical theory of experimental design is concerned with obtaining as much information as possible from the resources available. In some applications, of which agriculture is the most obvious, the experimental material consists of (possibly disjoint) sets of spatial (i.e., two-dimensional) material. Other spatial applications include sheet metal production and paper making. Some spatial designs have arisen for particular circumstances, such as spacing trials and genotype experiments. In these cases, the design is usually chosen on intuitive grounds, rather than for its efficiency under an assumed model. In other cases, efficiency is sought in terms of a prior assumed model for the unobserved data. This chapter presents the design of comparative experiments in which units (or subsets of units) are spatially arranged, and the spatial or directional layout is important. The chapter concludes with a discussion of the designs that are efficient under spatial dependence are discussed, concentrating on the case that all contrasts are of equal interest (treatments of equal status) along with test-treatment designs under dependence and factorial experiments under dependence.