On the detection limit

Abstract

In order to define the detection limit the two-step model is used, based on a false detection probability P10 (error of the first kind) to which there is a corresponding detection level as well as a proper detection probability P11. Given two probabilities P10 and P11 one may associate them with the detection level yk as well as with the detection limit ȳd, respectively [formula ommitted], two values of the entropy Hk, respectively Hd. In order to estimate the detection limit two procedures are given: the procedure based on the calibration function ȳ = f(c) obtained for [formula ommitted] c < cd and the frequentometrical procedure based on the calibration function P11= f(c), where [formula ommitted]. In both cases a confidence interval is associated with the detection limit. Finally, there is a description of the manner in which three tests are applied in order to check the hypothesis concerning the detection of a component in the sample under investigation: a parametrical (Student), a non-parametrical (Wilcoxon) and a sequential (Wald) one. © 1975, Walter de Gruyter. All rights reserved.