A NEW APPROACH TO A COMPLETE ROBUSTNESS TEST OF EXPERIMENTAL NOMINAL CONDITIONS OF CHEMICAL-TESTING PROCEDURES FOR INTERNAL ANALYTICAL QUALITY ASSESSMENT

The resistance presented by a particular analytical method or procedur e to the generation of biased results when subjected to slight modific ations in nominal experimental conditions has traditionally been known as ruggedness (from an interlaboratory trial) or robustness (from an intralaboratory study). This paper develops a protocol for a complete study of robustness, using a simple statistical technique which does n ot require the performance of ANOVA. The method simply consists of the establishing of a standard calibration and the creation of experiment al screening designs (saturated fractioned or Plackett-Burman factoria ls [R.L. Plackett, J.P. Burman, Biometrika 33 (1946) 305-325]). The va rious stages of the test are discussed and the proposed model is appli ed to the real data referring to a spectrofluorimetric method of deter mination of Lithium. (C) 1998 Elsevier Science B.V. All rights reserve d.