A note on Cannings and Thompson's sequential sampling scheme for pedigrees.

Abstract

We consider sequential sampling of pedigrees for genetic analysis. Cannings and Thompson (1977) gave simple, general guidelines for valid sequential sampling schemes. We show that their formulation of the likelihood contains an error, which is, however, easily corrected so as to maintain the validity of the sequential scheme. We also point out that although sequential and fixed-structure pedigree sampling do have the same likelihoods (as Cannings and Thompson showed), and therefore yield the same maximum likelihood point estimates of genetic parameters, they do not necessarily yield the same significance tests or confidence intervals.