Understanding the Bayesian approach to metrology

You have an element to measure/analyze: this element can be a standard block, a standard solution, a reference material, a standard source of a radionuclide... or an unknown sample (taken from a batch, a series, a production run...).

You are interested in a measurable property of this element. This property of interest could be "defined at a level of detail sufficient to be reasonably represented by an inherently unique true value" (ISO/IEC Guide 98-4: 2013).

Before measuring this property of interest, you already have "a priori" knowledge of the expected measurement value, either through previous measurements of the same measurand (a periodically measured standard...), or through previous measurements of similar elements (previous batches produced under the same production conditions with a controlled process...), or through the results of calculation codes or numerical simulations...

This fact sheet simply shows how the Bayesian approach can take this "a priori" knowledge into account when estimating the true value of the measurand and the uncertainty associated with this estimate. It also shows the advantages of taking this "a priori" information into account, and the limits of this exercise.