A COMPARISON OF TWO INTER-LABORATORY MEASUREMENT COMPARISON METHODS

摘要

This paper provides a description and an example of a statistical method of inter-comparison, which utilizes F and t significance tests to compare the variability and systematic measurement differences (precision and bias), respectively. A Chi-square (Χ{sup}2) test is additionally used to assess the significance of measurement data with respect to a reference data set. A fabricated example set of laboratory comparison data is evaluated. The evaluation results are compared to the results obtained by a method based on ISO 5725 "Accuracy (Trueness and Precision) of Measurement Methods and Results" [1]. In addition to utilizing repeatability, the conventional index of variability, the ISO method utilizes the statistical evaluation of reproducibility which, simplistically stated, treats systematic differences among measurement values as random quantities. Even when the differences are truly systematic in nature, it is statistically valid to evaluate the differences by this method, as long as the evaluation is statistically justified. A comparison is made between the two methods of evaluation. In addition, a Microsoft Excel program tool is described which is used to perform the calculations.