The bias of the unbiased estimator: a study of the iterative application of the BLUE method

TL;DR

This paper investigates how iterative application of the BLUE method can reduce bias in combined measurements when uncertainties are estimated from data, especially with relative uncertainties.

Contribution

It provides a detailed analysis of the bias introduced by using estimated uncertainties in BLUE and demonstrates how iteration can mitigate this bias.

Findings

Iterative BLUE reduces bias in combined estimates.

Bias depends on the nature of uncertainty estimates and their correlation.

The study covers a wide range of uncertainty scenarios.

Abstract

The best linear unbiased estimator (BLUE) is a popular statistical method adopted to combine multiple measurements of the same observable taking into account individual uncertainties and their correlation. The method is unbiased by construction if the true uncertainties and their correlation are known, but it may exhibit a bias if uncertainty estimates are used in place of the true ones, in particular if those estimated uncertainties depend on measured values. This is the case for instance when contributions to the total uncertainty are known as relative uncertainties. In those cases, an iterative application of the BLUE method may reduce the bias of the combined measurement. The impact of the iterative approach compared to the standard BLUE application is studied for a wide range of possible values of uncertainties and their correlation in the case of the combination of two measurements.