Avaliando a aceitação da impressão global do suco de caju via modelos de chances proporcionais mistos

Abstract

The proportional odds model is a method that can be used to describe the relationship between ordinal categorical variables. Since this is an extension of generalized linear models, it allows to obtain cumulative probabilities for each of the response categories and, therefore, the marginal probabilities. Thus, this study aims to fit a proportional odds model to verify acceptance of the overall impression of the cashew juice. The possible answers to this variable are ordinal categories for five different brands of cashew juice. A model was adjusted for this experiment, and in order to model the possible dependence between the observations a random effect was added to the proportional odds model.