Robustness in Empirically Based Binomial Decision Rules for Integrated Pest Management

Abstract

An empirical linear relationship between log [−log (P_T)) and log (m) has been shown to exist for samples from a wide variety of arthropod species; m is the mean and P_T is defined as the probability of getting no more than a predetermined number (T) of individuals in a sample unit. Previously presented expressions for the variance of the predicted values of m have often omitted terms that should be present. One of these terms is shown to, be relatively large in relation to the whole, and the consequences of omitting the term are described. The operating characteristic curves for decision making (e.g., to spray with pesticide or not) based on the relationship are shown to be sensitive to such a term. Ways of minimizing its effect are described: choosing T so that the linear fit is best and choosing higher values of T. Such remedies must be balanced against the costs of using a high value of T.