On the development of quality assurance

• Long-term changes in apparatus cause uncertainty about interlaboratory comparisons.
• Outliers should not be rejected in any type of chemical analysis.
• The methodology of quality assurance needs revision.
• We identify irregularities in certificates of certain certified reference materials.
• Consensus science is an appropriate methodology for analytical chemistry.

Contemporary research in quality assurance indicates that large uncertainties observed in interlaboratory comparisons to a large extent originate from a lack of competence of laboratory staff. This explanation is challenged by the present article for which six technologies and multiple series of experiments were investigated with respect to uncertainty of measurement and treatment of outliers. It was found that long-term precision was poor in comparison to short-term precision. The ratio of predicted uncertainty to observed uncertainty was determined as significantly above 1, and it was suggested that a correction factor is needed for the predicted uncertainty. This indicates that statistical control could be obtained only by treating many independent series of experiments and using pooled calibrations in the method validation. Retention of outliers in calculation of contents and calculation of uncertainties of certified reference materials (CRMs) gave results that differed significantly from those in the certificates of the CRMs.