Experimental quantification of measurement uncertainty and other verification criteria for analytical test methods

Abstract

If the measurement produces a numerical result (or the stated result is based on a numerical result), then it is necessary to evaluate the uncertainty of these numerical results. If the test methodology does not provide for a rigorous, metrologically and statistically reliable estimate of measurement uncertainty, the test laboratory should attempt to reasonably assess the uncertainty of the measurement results. This is applicable in the case of test methods both rational and empirical. As well as in the laboratory activities, verification of test methods plays an important role in order to guarantee the competence of the laboratory by checking for the criteria of Kohren, Grabbs and others. In cases where test results are not numerically expressed or based on numerical data (e.g., fit/unsuitable or positive/negative, or based on visual or sensory perceptions or other forms of quality analysis), estimates of uncertainty or other variability of results are not required. Nevertheless, laboratories are advised to have an idea of the variability of the results, if possible. The importance of the uncertainty of qualitative test results is undeniable, as is the fact that the necessary statistical methodology (procedure) exists for the calculations. However, due to the complexity of the issue and the inconsistency in the approaches, it is not currently necessary for laboratories to assess the uncertainty of qualitative test results. However, this issue needs to be considered. The article discusses estimates of uncertainty of analytical measurements taking into account the requirements of regulatory documents. The criteria for verification of measurement methodology based on experimental works are also considered and analyzed.