Definition and assessment of sampling and analytical accuracy

Two independent definitions for quantifying measurement accuracy and two li miting schemes for their assessment are examined in this paper. Gauss' mean square error MSE is compared to the symmetric-range accuracy A, describing the range of measurements about a measurand. Both measures of accuracy acc ount for systematic error (bias) and imprecision so as to quantify the clos eness of estimates to the actual values being measured. Remarkably, it is f ound that the accuracy functions are closely equivalent for most method app lications. Furthermore, details are presented on how to compute confidence limits on measurement accuracy so as to account for error in method evaluat ion. The confidence limits are qualitatively different in the case that the method undergoes extensive initial evaluation in comparison to a continual re evaluation at each method application. To this end the statistical theo ries of tolerance as well as more familiar types of confidence intervals ar e applied. Published by Elsevier Science Ltd on behalf of British Occupatio nal Hygiene Society. All rights reserved.