For any genetic model, Holmans showed that the proportions of affected sibs sharing 2, 1, or 0 identical-by-descent parental marker alleles are constrained to belong to a specific triangle. The triangle constraints do not hold when the sib phenotypes are determined by different models. We test the rejection of triangle constraints on affected sib pairs discordant for severity, to determine whether different models control the severe and mild forms of the disease in the simulated data. With this method we show that a locus on chromosome 5 plays a different role in the two forms of the disease.