Analysis of Variance in Small Group Research

Abstract

This paper discusses an ANOVA design appropriate for small group research in which dependent variables are derived from individual member responses. In such instances, subjects are nested within groups and groups are nested within treatments. Conditions under which it would be appropriate to pool the subjects within groups means square with the groups mean square to form the error term for the ANOVA are discussed. Such an error term will have somewhat greater power than that based on the groups means square alone. A common practice of using the subjects within groups mean square alone as the error term is shown to be inappropriate.