New Perspectives on Linear Calibration

In univariate calibration, two standard estimators are usually opposed: the classical estimator and the inverse regression estimator. Controversies have followed the use of both estimators and we consider them from a decision-theoretic perspective, establishing the inadmissibility of the classical estimator and the admissibility of the inverse regression estimator. The latter allowing for a Bayesian interpretation, we also develop a fully noninformative study of the calibration model and derive a reference prior which avoids the inconsistency drawbacks of the inverse regression estimator.