The use of Monte Carlo simulation to evaluate uncertainties of key comparison reference values and degrees of equivalence

摘要

BIPM Director's Advisory Group on Uncertainties has proposed two procedures for the evaluation of key comparison data in the case that a set of specific conditions regarding the key comparison can be assumed to apply. The first procedure is based on the use of the weighted mean, with its associated uncertainty, and consistency checks regarding its applicability, founded on classical statistics. Should the checks fail, action to remedy the situation is suggested and, if the remedy is inappropriate, a second procedure can be applied instead. The second procedure is based on the use of the median as a more robust estimator, with its uncertainty evaluated by the principle of propagation of distributions. Monte Carlo simulation provides an efficient implementation of the principle of propagation of distributions. The application of Monte Carlo simulation for evaluating the uncertainties associated with key comparison reference values and degrees of equivalence is described, and illustrated using key comparison data.