ENBIS: European Network for Business and Industrial Statistics

Approaches to comparing different designs of key comparison

23 September 2009, 10:05 – 10:25

Abstract

Submitted by

Peter Harris

Authors

Maurice Cox, Peter Harris and Clare Matthews

Affiliation

National Physical Laboratory, Hampton Road, Teddington, TW11 0LW, UK

Abstract

Interlaboratory comparisons (ILCs) are an important activity in monitoring and assessing the capabilities of laboratories in performing specific measurement tasks. The Bureau International des Poids et Mesures (BIPM), as part of its role in supervising the activities of national metrology institutes (NMIs), has published a set of protocols for ILCs involving NMIs. The Mutual Recognition Arrangement (MRA) provides for the mutual recognition of national measurement standards and calibrations and measurement certificates issued by NMIs. The MRA calls upon the results of a set of key comparisons that lead to a quantitative measure of the degree of equivalence of national measurement standards. There can be appreciable costs for the NMIs participating in a key comparison, for example, in terms of the costs of acquiring, measuring and transporting artefacts. The costs are to be balanced against the quality of information provided by an analysis of the key comparison data, in terms of (a) the uncertainties associated with the estimates of degrees of equivalence, and (b) the ability of the analysis to identify NMIs that have provided results that are inconsistent with a model of the key comparison. An approach is described for comparing different designs of key comparison in terms of measures of their cost and quality of information provided. The approach is illustrated using a simulated example.

Return to programme