Analysis of Variance II: Factorial Designs
Many experiments in psychology involve the simultaneous study of the effects of two or more factors on a response variable. Such an arrangement is usually referred to as afactorial design. Consider, for example, the data shown in Table 4.1, taken from Lindman (1974), which gives the improvement scores made by patients with one of two diagnoses, being treated with different tranquillizers. The questions of interest about these data concern the equality or otherwise of the average improvement in the two diagostic categories, and similarly for the different drugs. Because these questions are the same as those considered in the previous chapter, the reader might enquire why we do not simply apply a one-way ANOVA separately to the data for diagostic categories and for drugs. The answer is that such analyses would omit an aspect of a factorial design that is often very important, as we shall demonstrate in the next section.