Analyzing designed experiments: Should we report standard deviations or standard errors of the mean or standard errors of the difference or what?

Abstract

AbstractANOVA, one of the most common statistical methods applied in agronomy, offers a variety of results we can report when analyzing designed experiments. The focus, of course, is on treatment means, but what should we report to characterize precision? Should we choose treatment standard deviations (SDs) or standard errors of the mean or standard errors of the difference (SEDs)? We discuss why treatment raw SDs should not be reported as the result of ANOVA, and point out that most of the time it is SEDs that should be provided.