Self-consistent reference value estimation
- Much has been published over the last years on finding the best (consensus) reference value (RV) for a given set of laboratory data attained in inter- or key comparisons (the latter with a specific view to the CIPM MRA). However, the discussion is ongoing. Except cases where real reference values are available from other sources, the quality of the (consensus value is of major importance in proficiency testing, method Validation, or reference material certification. A distribution-free, self-consistent RV estimation approach has been developed which is based on the single criterion of participant's compatibility, at the cost of adjusted uncertainties. The final result is a fully compatible data set with minimum Overall variance, and adjusted uncertainties for those values reported which had to be made compatible with the (key comparison) reference value. Any outlier discussion becomes obsolete, including considerations concerning different degrees of robustness of the variousMuch has been published over the last years on finding the best (consensus) reference value (RV) for a given set of laboratory data attained in inter- or key comparisons (the latter with a specific view to the CIPM MRA). However, the discussion is ongoing. Except cases where real reference values are available from other sources, the quality of the (consensus value is of major importance in proficiency testing, method Validation, or reference material certification. A distribution-free, self-consistent RV estimation approach has been developed which is based on the single criterion of participant's compatibility, at the cost of adjusted uncertainties. The final result is a fully compatible data set with minimum Overall variance, and adjusted uncertainties for those values reported which had to be made compatible with the (key comparison) reference value. Any outlier discussion becomes obsolete, including considerations concerning different degrees of robustness of the various location and .dispersion estimators available. The penalty for "badly" performing laboratories is a large attributed uncertainty which forces them to either improve the measurement or adjust (i.e. reduce) their performance Claims. The evaluation algorithm is described. An example from a CCQM comparison is given, and consequences for the performance Claims of the participants discussed.…
Autor*innen: | Wolfram Bremser |
---|---|
Persönliche Herausgeber*innen: | F. Pavese, M. Bär, J.-R. Filtz, A.B. Forbes, L. Pendrill, K. Shirono |
Dokumenttyp: | Beitrag zu einem Sammelband |
Veröffentlichungsform: | Verlagsliteratur |
Sprache: | Englisch |
Titel des übergeordneten Werkes (Englisch): | Advanced mathematical and computational tools in metrology and testing IX |
Jahr der Erstveröffentlichung: | 2012 |
Verlag: | World Scientific Publ. |
Jahrgang/Band: | 84 |
Erste Seite: | 58 |
Letzte Seite: | 65 |
Freie Schlagwörter: | Dispersion parameter; Key comparison; Location parameter; Reference value |
ISBN: | 978-981-4397-94-0 |
Bemerkung: | Serientitel: Series on advances in mathematics for applied sciences – Series title: Series on advances in mathematics for applied sciences |
Verfügbarkeit des Dokuments: | Datei im Netzwerk der BAM verfügbar ("Closed Access") |
Datum der Freischaltung: | 19.02.2016 |
Referierte Publikation: | Nein |