An alternative method for assessing the agreement between test results

Abstract

The paper proposes a new way of assessing the agreement between measurement results during test quality assurance procedures in the laboratory. The decision-making rule is based on the measurement uncertainty. The probability, with which mathematical expectations of the measured data lie within the uncertainty of the indicator measurements, is proposed as a quantitative indicator. Such a quantitative indicator of the impact of methodological errors is proposed as the ratio of the difference between mathematical expectations of the measurement results obtained in different series of measurements to the average uncertainty of their determination, taking into account the applied decision-making rule. This indicator is based on the assumption that the measurement results are normally distributed. To simplify the process of calculating quality assurance indicators and reduce the risk of making mistakes in calculations, the “Agreement” module was created on the basis of a virtual test laboratory. The input data for the calculation of the agreement in the module are: measuring equipment, measurement results themselves (at least three for each tester), the coverage coefficient, full names of the people performing the measurements, and the coefficient of the decision-making rule.