Premium
Derivation of Recommended Values: An Opinion *
Author(s) -
KANE Jean S.
Publication year - 1991
Publication title -
geostandards newsletter
Language(s) - English
Resource type - Journals
eISSN - 1751-908X
pISSN - 0150-5505
DOI - 10.1111/j.1751-908x.1991.tb00091.x
Subject(s) - nist , certification , base (topology) , principal (computer security) , statistics , certified reference materials , quality (philosophy) , computer science , work (physics) , database , operations research , mathematics , engineering , political science , mechanical engineering , mathematical analysis , philosophy , epistemology , detection limit , natural language processing , law , operating system
Two principal approaches are being used currently to develop reference samples used in geochemistry. The first is typified by work at The National Institute of Standards and Technology (NIST) to develop standard reference materials (SRMs) whose elemental concentrations are certified. All data is collected under stringent quality control (QC). Agreement between average results for replicated analyses (6≤n<20) obtained by two or more analysts using either a single, definitive method or independent methods of analysis is required for certification; a large number of laboratory/method average results is not. The United States Geological Survey (USGS), International Working Group (IWG), and others base recommended concentrations on larger sets of compiled data, which may or may not have been collected under similar conditions of QC. The derived recommended concentrations are updated periodically, on the assumption that recommended concentration quality improves as the size of the data base increases. That assumption is examined in this paper, and the general lack of statistically significant difference between original and updated concentration is demonstrated.