Robust estimation in the simple errors-in-variables model

Abstract

Fekri and Ruiz-Gazen [2004. Robust weighted orthogonal regression in the errors-in-variables model. J. Multivar. Anal. 88, 89–108] propose a new class of robust estimators in the errors-in-variables linear model (EIV model) under the assumption that the measurement error variances are equal. In the present paper, we extend this new class of estimators to other usual error variances assumptions for the simple EIV model and calculate their influence functions and their asymptotic distributions in the elliptical case.