ABSTRACT
J. A. Hoekstra emphasizes the distinction between natural mortality, and control mortality, which may well not equate to natural mortality. The presence of natural mortality can prevent proportions responding dropping to zero, as dose levels are reduced. Natural immunity can have the effect of preventing proportions reaching unity, as dose levels are increased. A variety of Bayesian approaches have been proposed for the analysis of quantal assay data. The statistical analysis is one of errors-in-variables for binary regression, and has been considered by R. J. Carroll et al. and D. Burr. An early paper was by K. M. Patwary and K. D. C. Haley, in which it was assumed that the dose received had a Poisson distribution. Wadley’s paper was followed by that of Anscombe, who provided a model based on the negative binomial distribution, which may provide a better description of data which may be over-dispersed relative to the Poisson distribution.
