ERROS DE ESPECIFICAÇÃO NO MODELO DE REGRESSÃO BETA COM DISPERSÃO VARIÁVEL

Main Article Content

André Antonio de OLIVEIRA
Tatiene Correia de SOUZA
Saul de Azevêdo SOUZA

Abstract

Our goal with this article is to evaluate the effects of misspecifications in the inferences of the beta regression model with varying dispersion. For this, a simulation study was carried out. In these simulations, the response variable was generated with Beta distribution assuming known covariates and link functions, thus the model has been adjusted in the correct and incorrect specification, particularly considering six kinds of specification errors. We evaluate the effects of these errors through rejection rates and coverage rates in relation to one of the average submodel parameters and, in addition,  we also evaluated the relative bias and the mean square error of the estimates mean responses. We verified from the results obtained, that the specification errors involving the linear predictor of the precision regression  structure had a considerable influence on the model inferences. Finally, we performed an application to real data in order to compare the effects of different ways specification on the inferences of Beta regression model with varying dispersion. 

Article Details

How to Cite
OLIVEIRA, A. A. de, SOUZA, T. C. de, & SOUZA, S. de A. (2018). ERROS DE ESPECIFICAÇÃO NO MODELO DE REGRESSÃO BETA COM DISPERSÃO VARIÁVEL. Brazilian Journal of Biometrics, 36(3), 601–624. https://doi.org/10.28951/rbb.v36i3.245
Section
Articles