Heteroscedastic regression model

Sergio Augusto RODRIGUES[1]

Carlos Alberto Ribeiro DINIZ[2]

§    ABSTRACT: One of the basic assumptions in the regression theory is that the errors should have a constant variance. When the violation of this assumption occurs, an option is to transform the data or to reestimate the variance structure of the data. This work introduces the linear heteroscedastic regression model as well as the estimate and detection of heteroscedasticity.

§    KEYWORDS: Heteroscedasticity; estimation.

[1]Faculdade de Educação São Luis - FESL, CEP 14870-370, Jaboticabal, SP, Brasil. E-mail:  sergio@saoluis.br

[2]Departamento de Estatística, Universidade Federal de São Carlos - UFSCar, Caixa Postal 676, CEP 13565-905, São Carlos, SP, Brasil. E-mail: dcad@power.ufscar.br