ABSTRACT
Repeated measures designs have the same structure as split-plot designs with the main similarity being that they both involve more than one size of experimental unit. The difference between the two types of designs is that the levels of all of the factors associated with a split-plot design are randomly assigned to the respective experimental units, whereas for the repeated measures designs, there are one or more factors whose levels cannot be randomly assigned to their experimental units (Milliken and Johnson, 1992). Because of the non-random assignment process of the levels of a factor to one of the sizes of experimental units, the correlation among those experimental units may be different than the uncorrelated assumption used for the analysis of split-plot models. Thus, this chapter emphasizes specifying or selecting an adequate correlation structure associated with the repeated measurements instead of spending time on specifying the form of the slopes and intercepts, since the forms of the regression models for the repeated measures designs are identical to the forms of the regression models for the split-plot design and was discussed in detail in Chapter 15.
