UNITS OF ANALYSIS, MEASUREMENT SCALES, AND STATISTICS - A COMMENT

Kerwin and Shaffer recently conducted a study examining the role of dogmatism as a possible moderator of the relationship between judge's instructions and mock jury decisions. The study was deficient in that it inappropriately applied an individual-level construct (dogmatism) to groups; used incorrect scaling procedures and an inappropriate statistical analysis, leading to incorrect conclusions, and used aggregate data to draw conclusions concerning individual behavior, thereby losing information available in the data. Implications of these problems for research design are discussed.