BAYESIAN ANALYSIS OF A CALIBRATION MODEL

Ignacio Lira, Dieter Grientschnig

Abstract

A Bayesian analysis of a calibration model was presented in <i>Metrologia</i>, <b>43</b> (2006) S167-S177, wherein two approaches were considered to obtain the probability density function associated with the measurand. In one of them, Bayes' theorem was applied directly to an input quantity for which measurement data were available. In the other approach, that same input quantity was expressed in terms of the measurand and the other input quantities. Since the forms of the likelihood function used in each approach were not the same, different prior functions were needed. In this paper we show that both approaches produce the same final results if the prior function to be used in the second approach is derived from that applicable to the first approach. By following this procedure, both prior functions are assured to encode the same initial information.

Keywords

Bayesian analysis, prior functions

Download

IMEKO-WC-2009-TC21-053.pdf

DOI

-

IMEKO TC

TC21 - Mathematical Tools for Measurements

Back to the proceedings