Assessing Insecticide Bioassay Data with Extra-Binomial Variation

Development and use of residual plots from the analysis of binary response data are described. Examples of residual plots demonstrating various departures from probit binomial models are presented. One type of departure that can be detected by residual plots is the existence of extra-binomial variability. A procedure is described for fitting binary data with extra-binomial variability and calculating the appropriate statistics for testing the hypotheses of parallelism and equality. An illustrative example is given to demonstrate the use of the procedures and the macros that were developed to be used within the GUM statistical package.