Recommendations for the conduct and interpretation of co-operative trials

Abstract

In a co-operative trial, six or more laboratories analyse a series of specimens by any method that is appropriate. The results are collated and examined by a nested analysis of variance procedure, with a view to making a realistic quantification of random and systematic errors in the system. The performance of the laboratories as a whole and individually can be assessed. The information can be used to initiate remedial action in laboratories producing discrepant results, or to select laboratories for particular tasks. Suggestions for the conduct of these trials is given in respect of the choice, preparation and analysis of the specimens, the recording and reporting of data, and the statistical evaluation of the data.