When Bigger Is Not Better: Design Sensitivity in a Sample of Criminal Justice Experiments

The discussion focuses on the problem of statistical power and challenges traditional assumptions about the relationship between research design and experimental results. It explains conventional means of increasing the statistical power of research designs, describes the experimental studies of sanctions, and considers the general methodological characteristics of those experiments. The discussion also examines the relationship between sample size, which is usually seen as a primary determinant of statistical power, and the actual outcomes of experiments. The analysis shows that larger studies that should lead to more powerful research designs do not. Ways are suggested of overcoming the weaknesses of large sample designs. Footnotes, tables, appended list of experiments, and 152 references