Adjustment of generalized linear models to evaluate the control of insect pests

In this study, the statistical method of generalized linear models was applied in an experiment with mortality proportion data, in a residual contact bioassay using maize weevil, Sitophilus zeamais (Coleoptera: Curculionidae), as a bioindicator. These data are frequently overdispersed, thatis, they are characterized by having a greater variation than that assumed by the binomial model. One of the advantages of properly modeling overdispersion is to avoid incorrect inferences. Therefore, the objective was to adjust a probability distribution to the data, using the logistic link function, whose adjusted model was the quasi-binomial logistic and, in turn, the quality of the adjustment was evaluated through the half-Normal plot, usingthe R software. Thus, the model used allowed the separation of the most effective treatments, bypassing the overdispersion present in the data. Consequently, it can be used in studies aimed at the selection of alternative compounds with insecticidal properties against maize weevil or in the evaluation of the effectiveness of commercial products.