ABSTRACT

In Chapters 13 and 14 we discussed the analysis of two-way treatment structures in a completely randomized design structure when there are missing treatment combinations. In this chapter we illustrate how to analyze a two-way treatment structure in a randomized complete block design when some treatment combinations are not observed. Consider the data in Table 15.1, which is obtained from the experiment described in Chapter 12, but with a few treatment combinations not being observed in any of the blocks. Figure 15.1 shows the treatment combinations observed at least once. Any hypothesis that involves treatment combinations (fat 1, surfactant 3) or (fat 2, surfactant 2) cannot be tested unless additional assumptions are made. In this discussion, let FSij represent the response expected when fat i and surfactant j are assigned to a randomly selected experimental unit.