Analysing dose-mortality data when doses are subject to error

In microbial control of insect pests, the dose of pathogen that a subject actually receives varies because of variation in individual intakes. Variation in intake of a given concentration is equivalent to varying the concentration for a given intake. The main effect of such variation is to flatten the dose-response curve. If the dose-variation is known, a correction for this effect can be made; new estimates for the slope and intercept of the probit line are given. However, the correction increases the variance of the estimates. The theory is illustrated with an example in which the diamond-backed month, Plutella xylostella, was fed varying doses of the specific granulosis virus PxGV; intake assays using similar larvae showed the coefficient of variation of intake volume to be about 40%. Simulation studies showed that correction may be practically important only for the estimation of extreme percentiles, e.g. LD₉₅s.