Detection of influential observations in the determination of the weighted-mean KCRV

Since the signing of the Mutual Recognition Arrangement, national metrology institutes (NMIs) have carried out many key comparisons in a wide range of metrological areas to establish the equivalence of their measurement standards. The determination of a key comparison reference value (KCRV) and its associated uncertainty is the central task in the evaluation of key comparison data. One of the most popular ways to estimate the KCRV is to use a weighted mean of the NMI's reported values, with weights inversely proportional to the variances of the reported value. One potential problem with the use of the weighted mean is its reliance on weights, which may vary greatly across NMIs. Consequently, some of the NMIs can be influential in the determination of the weighted-mean KCRV. Thus it is of interest to identify the influential NMIs based on some simple and well-defined criteria. In this paper, we present several easy-to-use criteria for detecting influential data in the calculation of the weighted-mean KCRV.