Applications of robust estimators of covariance in examination of inter-laboratory study data

Abstract

Selected robust estimators of covariance or correlation are presented. Their use in the identification of anomalous laboratory results in inter-laboratory data is illustrated. It is shown that robust estimators can substantially reduce the impact of outlying values on multivariate confidence regions and consequently lead to sharper identification of anomalies, even where traditional outlier detection may fail to locate anomalous results.