Statistically combining independent studies: A meta-analysis of sex differences in conformity research.

Abstract

Traditional (literary) reviews of research in social psychology are compared with a statistical approach. It is concluded on both abstract and practical grounds that adoption of the statistical approach would lead to theoretical progress for the research area covered. A meta-analysis package is described and then applied to the question of whether there are sex differences in degree of conformity. The meta-analysis is yoked to a literary analysis, and conclusions of differing direction and confidence appear. Problems in application are encountered, and appropriate courses of action are discussed. Finally, limitations on the power of the procedure are outlined. The traditional way to review research in social psychology has been to take a literary approach. That is, summary statements about research areas are usually based on impressions gleaned by the reviewer from a reading of related studies. This article takes issue with the efficiency of such an approach. An example of a statistical technique for combining the results of independent experiments is provided, and this approach is contrasted with a literary summary. The quantitative procedures describe numerically the characteristics of a body of evidence and give a probability level related to the observed pattern of results. Although the actual numerical manipulations require little statistical sophistication, the theoretical progress possible through their employment may be great (progress is here denned as more precise and confident statements about segments of our world). The reasons for believing that progress will follow the adoption of quantitative procedures are both abstract and practical.