ABSTRACT
The split-plot type design involves a design structure with more than one size of experimental unit where the smaller-size experimental units are nested within the largersize experimental units. Some examples of split-plot design structures were presented in Chapter 5, including the class of hierarchal design structures. Two main problems occur in the design and analysis of split-plot type design structures. The rst problem consists of the selection and/or identi cation of the different sizes of experimental units used in the design structure followed by the assignment of treatments from the treatment structure to the different experimental unit sizes in the design structure. The successful identi cation of the different sizes of experimental units is paramount in the speci cation of an appropriate model that can describe the resulting data. The second problem is constructing the appropriate model that describes the pertinent features of the treatment and design structures. It is important to be able to identify the sources of variation that measure the variability associated with each size of experimental unit. These sources of variability are used to compute the respective error terms which are used to compute estimates of the standard errors of estimated means and for pairwise comparisons among means. Since these design structures involve more than one size of experimental units, the estimates of the standard errors of the xed effect parameters and their comparisons among them involve one or more sources of variation. A very important characteristic of the model for the split-plot type designs is that they are the basic model for starting the construction of the repeated measures models discussed in Chapter 26. Examples of several of the concepts were presented in Chapter 5.
