It is well known that the Haseman-Elston (H-E) sib-pair linkage method does not assume that the genetic model underlying the trait phenotype is known without error, although this assumption is made for marker loci. However, misspecification of allele frequencies at the marker locus decreases power when some or all parental genotypes are unknown. In this study, the power of the H-E sib-pair method was compared for different types of traits when some or all parental data were missing and allele frequencies at the marker loci were misspecified. Data were generated for a quantitative trait and marker loci in nuclear families using G.A.S.P. (V3.3). Three types of traits were simulated with two equifrequent alleles with a random environmental effect (10%, 30%, and 50%). The simulated data were analyzed using (i) one of the parent's marker data, and (ii) no parental marker data, with both correct and incorrect marker allele frequencies. This test is found to be robust in most of the situations considered except for a slight decrease in power when sample size is small and when the marker locus is not very polymorphic.
Copyright 2001 Wiley-Liss, Inc.