(Received 18 January 2000; accepted 20 April 2000)
The primary goal of an interlaboratory study (ILS), or of a collaborative study as it is also called, is to verify that a measurement method will work over many testing or analytic laboratories. The data produced and the calculations of the precision statement are well established in statistical standards. We suggest some calculations beyond the precision statement to deal with detecting laboratory and data point outliers. These calculations, as suggested by Mandel, were designed to mesh with those of the precision statement but to be applied when the data first arrive back to the study director. We apply these calculations to 11 sets of ILS data from a variety of test methods and show how to read the results for indications of problems with the test method and for locating deviant laboratories.
Paper ID: JTE12109J