This paper develops a general method for comparing treatment magnitudes for research employing multiple treatment fixed effects analysis of variance designs, which may be used for main effects with any number of levels without regard to directionality. (Author/BS)

Formulae for estimating individual rater reliabilities from analysis of treatment effects are presented and evaluated. Monte Carlo methods illustrate the formulae. Results indicate that large sample sizes, large true treatment effects, and large differences in the actual reliabilities of raters are required for the approach to be useful. (SLD)