Simulation of the accuracy and coverage in experiments including mixtures with binomial and poisson responses studied using generalized linear model and ordinary least squares aproaches

Abstract

This article provides an analysis of experimental designs simplex centroidmixtures with binomial and Poisson responses using a GeneralizedLinear Model. Throughout a simulation process, in which a model wasassumed known, it was compared accuracy and coverage of the confidenceintervals around the expected average obtained under a modelbased on normal theory with that provided by a Generalized LinearModel. Simulations were carried out using specialized software, andrandom perturbation to the true average expected in each experimentalpoint was introduced repeatedly. The new response vectors were analyzedin each of the two models. Accuracy and coverage were reportedin each case depending on the magnitude of the induced disturbanceand were used as a measure of comparison of estimation uncertaintyand to evaluate the performance of both models. In an experimentaldesign with mixtures including no normal reponse, a Generalized LinearModel produced better results in terms of accuracy and coverage thanan Ordinary Least Squares based on a transformation.