Decision rules based on the estimate of a proportion of measurements

Frédéric Bertrand (1) , Myriam Maumy (1)

1 IRMA - Institut de Recherche Mathématique Avancée

The purpose of validation of analytical methods, for each laboratory, is to demonstrate that the method is suited for its intended purpose. Validation is not only required by regulatory authorities but it is also the ultimate phase before the routine use of the method. Analytical method validation must bring confidence for the laboratories in the results that will be produced since they are used to make critical decision. From a theoretical point of view, this work provide second-order accurate one-sided of the confidence intervals whose lower limit will be used in the decision rule. Even if Cornish-Fisher expansions are asymptotic ones, they often work well with moderate sample sizes. This should result in an improvement of the correctness of the decision rule and reduce both consumer and laboratory risks.

Mots clés: Validation; Estimate of a proportion of measurements; Quality control; Edgeworth expansions

Domaines: Statistiques [math.ST]; Théorie [stat.TH]; Méthodologie [stat.ME]; Applications [stat.AP]

Frédéric Bertrand, Myriam Maumy. Decision rules based on the estimate of a proportion of measurements. 11th Conference on Chemometrics in Analytical Chemistry, Jul 2008, Montpellier, France. pp.245 - 249. ⟨hal-00287748⟩