A statistical testing framework for evaluating the quality of measurement processes

Abstract

In this paper in which we address the evaluation of measurement process quality, we mainly focus on the evaluation procedure, as far as it is based on the numerical measurement outcomes. We challenge the approach where the "exact" value of the observed quantity is compared to the error interval obtained from the measurements under test and we propose a procedure where reference measurements are used as "gold standard". To this purpose, we designed a specific t-test procedure for this purpose, explained here. We also describe and discuss a numerical simulation experiment demonstrating the behaviour of our procedure.