Error-Purifying Designs

• Published in: Analysis of Variance and Functional Measurement
• Main Author: Weiss, David J
• Format: Book Chapter
• Language: English
• Published: United States Oxford University Press 24.11.2005; Oxford University Press, Incorporated
• Subjects: Cognitive Psychology; Developmental Psychology; Psychological testing & measurement; Social Psychology; Design; Situations; Second; Conditions; Simplest
• ISBN: 9780195183153; 0195183150
• DOI: 10.1093/oso/9780195183153.003.0007

A repeated-measures design is one in which each of several subjects serves in more than one of the experimental conditions. In the simplest case, each subject gets all of the treatment combinations once. This new structure brings about two situations we have not encountered before. First, “subjects” comprises one of the factors in the design. Second, because there is only one score per cell, no within-cells term is available. (If each subject is in some but not all of the conditions, a repeated-measures nested design is being used—see chapter 11.) The repeated-measures design has two important advantages for the researcher. One is that the design is economical. Once an experimental participant has been prepared or trained, it is obviously efficient to get more than one score from him. With human subjects, just getting them to the laboratory can be expensive. On the other hand, sometimes it is not feasible to get multiple observations from a subject because performance in one condition may affect performance in another condition. But when lack of independence between conditions is not a problem, the repeated-measures design can dramatically reduce the effort required of the experimenter.