A power‐analytic examination of contemporary communication research

Abstract

While Type I error has enjoyed widespread attention in communication research, the interrelated concepts of statistical power and Type II error have been generally neglected. In order to assess the extent of the disregard for these considerations, a statistical power analysis of the articles published in the 1973 volumes of the nine major communication journals was conducted. The results indicated that published communication research was generally lacking in statistical power. The average power estimates for small, medium, and large effects were .18, .52, and .79, respectively. These estimates correlate with previously conducted power analyses by psychological and educational researchers, thus evidencing the fact that low statistical power pervades behavioral research. Recommendations for the elimination of this problem were advanced, and the role of power analysis as an important consideration in experimental design was explicated.