As found by Jim Fry (University of Rochester), Table 25.1 contains numerous (minor) numerical errors. pdf copy of the corrected table.

The following comments are from Gary Snowder (Animal Geneticist, USDA, ARS, U.S. Sheep Experiment Station, Dubois, ID 83423).

The topic of multiple thresholds only considers the approaches of Reich et al. 1972. While there is nothing incorrect with this approach there are other approaches that allow one to consider all thresholds without having to combine classification groups.

1. W. E. Vinsen, J. M. White and R. H. Kliewer. 1976. Overall classification as a selection criterion for improving categorically scored components of type in Holsteins. J. Dairy Sci. 59:2104-2114.
   Buried in this manuscript is a simple formula for correcting a heritability estimate of any polychotomous trait to the underlying continuous scale. See pg 2106, equation 3.

   I have used this method for subjective traits with 3 and 4 levels to obtain reasonable adjustments. For example, with sheep data I have estimated that the heritability of a subjective milk score (low, average, high) for range ewes at the observed level is .10. After correcting to the underlying scale, the estimate increases to .26. This latter estimate agrees with heritability estimates for the quantitative trait of milk production in milking sheep.

   Anyway, it provides a useful and simple procedure for a qualitative trait whose distribution is greater than binomial.

2. D. Gianola. 1979. Heritability of polychotomous characters. Genetics 93: 1051-1055.
   See equations 10, 11 and 12 for his adjustments from the binomial distribution to the underlying scale. He proves Robertson's approach and infers that it can be expanded to include classifications greater than binomial.

Home Pages: [ Volume One ] - [ Volume Two ] - [ What's new ] - [ Book]

Bruce Walsh. jbwalsh@u.arizona.edu . Comments welcome.