On the choice of mathematical models for the estimation of lethal gene equivalents in man.

A range of mathematical models and error distributions was used to examine the validity of linear regression methods for the calculation of lethal gene equivalents. Because of the restricted span of inbreeding coefficient F values available in human studies and the limited number of data points, equivalent results were obtained with all combinations tested. It was concluded that linear regressions should be employed only for the detection of significant inbreeding effects in man and that their application to the estimation of lethal gene equivalents was not warranted.