NIST Home > PML > Radiation Physics Division > Radioactivity Group > International Comparison of Lu-177 continued

Since that time, several more NMIs have standardized this radionuclide. In order to establish a link between primary standards of ß-emitters in the NMIs and the SIR, as well as to provide a means for laboratories to substantiate CMC claims for ß-γ nuclides, a CCRI(II) Key Comparison of A total of 12 laboratories took part in the comparison. One of the results was from a laboratory that was neither the national metrology institute (NMI) nor the designated institute (DI) for their country, so this was combined with that of the DI in their country to give a total of 11 values to go into the calculation of the Key Comparison Value (CRV). The analysis of the set of results indicated that it was statistically inconsistent. Since the various tests that were applied to the data could not identify any single laboratory as being an outlier, an approach was taken that would allow all of the results to be used in the calculation of the CRV. To do this, it was assumed that the uncertainty in each laboratory’s result contains both within-laboratory and (generally unobserved) between-laboratory sources of variability [2]. Applying the Maximum Likelihood approach of Vangel and Rukhin [3] to the estimation of the consensus mean, a proposed CRV of 3.288(4) MBq∙g-1 was calculated, where the quoted uncertainty corresponds to a standard (k=1) uncertainty interval that considers both within- and between- laboratory effects. Figure A shows the activity concentration values for the One of the important outcomes of this comparison was the fact the previously-observed 1.4 % difference between NIST and PTB appears to no longer be present.
References [2] M. Crowder, “Interlaboratory comparisons:round robins with random effects”. J. Royal Statist. Soc. Series C, 41(2), 409-425. 1992. [3] M. G. Vangel and A. L. Rukhin, “Maximum likelihood analysis for heteroscedastic one-way random effects ANOVA in interlaboratory studies”. Biometrics, 55, 129-136 (1999). |