Comment on “Statistical Power with Group Mean as the Unit of Analysis”

Abstract

Barcikowski (1981) has provided tables for use in situations where unit (e.g., class or school) means are to be used as the unit of analysis. Apparently unrecognized by Barcikowski, and perhaps by the research community at large, is the fact that under the conditions specified for use of these tables (i.e., knowledge of the intraclass correlation), one would not generally wish to carry out analyses based on group means since a superior analytic strategy is available in this situation. This paper explicates this methodology.