Introduction to Model Comparisons One-Way Between-Subjects Designs
Scott E. Maxwell, Harold D. Delaney, Ken Kelley · 2017
One of the most common questions motivating behavioral research is whether two or more conditions differ from each other in effectiveness. The conditions could be therapies delivered in a clinical setting, or types of masking of stimuli in a computer-administered task employed by cognitive psychologists, or instructional methods being compared by an educational psychologist. In this chapter, and in fact throughout all of Part II of the book, the assumption is that each of the various conditions is experienced by a different group of participants, hence the label “between-subjects” designs, as explained at the end of Chapter 2 . Typically, the question of most interest is whether the difference between groups is statistically significant, that is, is the difference larger than would be expected to occur simply as a result of the variation induced by random assignment of participants to conditions. Typically, one would also like to be able to provide information about the magnitude of the difference between conditions. The current chapter will provide an introduction to methods for conducting statistical tests and characterizing the size of effects that will be generalized in subsequent chapters to apply to more complex designs. In addition, the current chapter will introduce methods for answering the critical question that arises in planning an experiment, namely, how large must the sample sizes be to make it likely that an effect of a given projected size will be detected.