Statistical uncertainty analysis of key comparison CCEM-K2

The details of a statistical uncertainty analysis applied to key comparison CCEM-K2 are reported. The analysis presented here provides an approach for addressing known correlations which have been of concern in reporting key comparison results. Uncertainties for each participating National Metrology Institute (NMI), the key comparison reference value (KCRV), and the pairs of NMIs are determined by fully considering the covariances arising from interrelations among the results of participants because of the time dependence of the values of the transport standards. Uncertainties of the pilot laboratory are treated differently than those of other NMIs due to the predicted values of the transport standards for all NMIs being based on pilot laboratory data. The approach separates the Type A and Type B uncertainty components providing the benefit that data from multiple artifacts (three in this instance) reduces the computed Type A uncertainties.