The estimation equation is essentially a procedure to determine weights of samples. When samples are correlated, their weights are lower than 1. It was shown in Hanley et al., (2003) that the weight for sibpairs w that minimizes the variance is exactly equal to the Eq.(3) used in this paper. We expect that in general, the weight of related samples determined by minimizing the variance will be equal to the sample size reduction α if the weight for independent individuals is set to 1.
