There are several other methods that take into account pedigree-based or empirically estimated kinship matrices in the statistical test39-42. One of the key differences between these methods and the mixed model methods, including EMMAX, is that the mixed model methods have a procedure for estimating the contribution of the kinship matrix to the phenotypes, whereas the other methods do not. Estimating the phenotypic variance contributed by the sample structure enabled us to avoid undercorrection or overcorrection of the sample structure in the NFBC66 and WTCCC data sets.
