One can also apply the ESS method to each pedigree-type specifically. We count the T and C alleles in pedigrees with only affected sibpairs, then reduce the count by the factor 1/(1 + 0.5) = 2/3. Similarly, the allele counts in pedigrees with three affected sibs are reduced by the factor of 1/(1 + 2 × 0.5) = 1/2, etc. The pedigree-type-specific allele count reduction leads to p-value of 0.00467. We can partially explain why this p-value is not as good as the one derived by the global sample size reduction: the association signal is largely due to an enrichment of the major allele T in the case group; however, the largest pedigree with 8 affected members with 13 counts of the T allele leads to an effective contribution in the “local method” of 3.3 counts, as versus the 9.7 counts in the “global method”.
