The stability of 57 consensus values in a proficiency test material re-issued blind after an interval of 18 years

Abstract

In 2017 a silicate rock test material was re-issued in the geoanalytical proficiency test GeoPT after an interval of 18 years. The second distribution was participant-blind. Consensus values (for each analyte for which measurements were submitted by a sufficient number of participants) were compared and tested for significant difference in location. Such an experiment can provide information about the stability of both the test material and—the main focus of the present paper—the assigned values, which are consensus values as estimated location parameters. The validity of a consensus as an assigned value in proficiency testing has often been questioned in metrological thinking although there is seldom a practicable alternative. Over the whole suit of analytes, there was no indication of change in the location estimates. This is a remarkable finding, given the incidental changes that had occurred between the two rounds of the proficiency test, especially (a) changes in the analytical technology used, (b) the turnover in participant laboratories, and (c) the large increase in emphasis on data quality in the intervening period.