Using reproducibility to test the adequacy of GUM based uncertainty quantification.

摘要

In this paper we presented the basic theory of GUM, and discussed how UQ is conducted via a reproducibility study (one-way ANOVA) We pointed the reader to pertinent quality assurance literature discussing the justification that reproducibility (im)precision as a good expedient for true uncertainty in measurement. While the literature discusses the SS U statistic, we are not aware of the application of the F-test to assess the significance of the SSU and therefore the adequacy of the participant uncertainty. We illustrated the application of this test to a small collaborative trial on U and Pu isotope ratios. The advantage of using the F-test for assessing the correctness of participants uncertainties is that it accounts for the uncertainty in the participants GUM UQ (v_(eff)) as well as the uncertainty in estimating the reproducibility which is limited by the number of participants p.