Sobre modelos de covariância com erros elípticos: uma abordagem bayesiana.

Abstract

In this work the Model of Covariance with error in their variables will be studied, where these errors have elliptical distributions, under a Bayesian perspective. In order to accomplish this we will use “a priori” information of the not informative type, as proposed forJeffrey(1961),and we will make inferences on the parameters of the studied model. It will be showed that for any model of covariance with elliptical error in their variables, combined with “a priori” information of the not informative type, the results will lead to the same analyses obtained through the posteriori analyses that correspond to the normal model of covariance with errors in their variables.