Unbalanced design

To be analysed by Genstat ANOVA a design must have the property of first-order balance This requires that the contrasts of each treatment term must all share a single efficiency factor, in every stratum where the term is estimated. For ANOVA to be able to combination of the information from all the strata where a treatment term is estimated the design must be generally balanced. This has the additional requirement that all the block terms are orthogonal. For further details, see Payne & Tobias (1992, Scandinavian Journal of Statistics 19, 3-23).