Number of lethal equivalents in human populations: how good are the previous estimates?

Lee et al. (1996) recently developed a method for interval estimation of the number of lethal equivalents by using a hierarchical structure of likelihood functions. This hierarchical model consists of two multinomial trials: one of the sampling process of the parents from the population of interest, and the other for the survival of the offspring of the families obtained by mating the parents. The method, initially developed for selfing and full-sib mating, is extended here to include more general mating systems as well as mixtures of mating systems. We applied it to human data sets for which confidence intervals were previously not available. Our point estimates were close to previous ones, and the standard deviations were generally quite small. Thus, even if debate over the meaning of the concept of lethal equivalents has not been entirely resolved, our results showed that the previous estimates are at least statistically meaningful.