An alternative estimator in inverse linear regression

Abstract

The calibration problem is an important problem from a practical view point. In recent years it has attracted the attention of several research investigators (see references) specially after Krutchkoff (1967) suggested the inverse method of calibration and claimed its superiority to the classical one. However, as will be reported in Section 1, the discussion on the merits of these two methods in the statistical literature can be characterized by disagreement and confusion. For this reason the authors suggest in this paper an alternative estimator and expound its properties. It is shown that the alternative estimator is uniformly better than the classical as well as the inverse method of calibration.