Multifactor Designs: Designing and Interpreting Experiments with Multiple Influences
The second rule for research states: With one factor there must be only one difference distinguishing different conditions of the experiment, that difference being variations in the factor. Most experiments actually contain two or three factors, and occasionally even more. Theorists realize that most behavior, and certainly most human behavior, is multiply determined. Factors, in the form of manipulated variables, are prospective influences on people's behavior (some dependent variable) that an experimenter wishes to examine. There are many influences on our decisions, and our preferences, and even on the speed or zest with which we do things. Therefore, psychological questions are more realistically inquiries about multiple influences, multiple factors. Designs examining the influence of more than one factor, and that can answer questions about interactive influences of factors, are called multifactor designs. In this chapter, multifactor designs are discussed, along with multifactor analysis of variance, which is a common statistical procedure used with multifactor designs.
Each factor in a multifactor design is a dimension along which experimental conditions (or subject qualities) are deliberately varied. Each subject must be exposed to (or classified according to) just one level of each of the factors. Thus, the second rule for research can be restated as, "All differences between conditions must be specified factor differences."