Probability models and computational models for ANOVA in multicenter clinical trials.
Multicenter trials are usually analyzed by fixed-effects, two-way analysis of variance. A mixed-effects computation with random treatment-by-center interaction may produce better results even if this interaction is considered truly to be a fixed effect. It may be more appropriate, however, to consider the treatment-by-center interaction to be a random effect, anyway.
Analysis of Variance↗