ABSTRACT
Performance testing (PT) is a special kind of collaborative test or round robin conducted to evaluate the performance of a laboratory for reasons. Advance estimates of precision should be based on no less than 7 and preferably at least 15 degrees of freedom. While a laboratory should know both its short-term and long-term standard deviations, the latter is the one of concern in performance testing. Before the PT is conducted, there should be some expectation of what would be classed as good performance. The critical parameters of each test method should be known and tolerances evaluated for them to the extent possible before a PT is undertaken. In the absence of the "true value" for a test material, a consensus of results is sometimes used to evaluate laboratory bias. Method-inherent bias requires research for its correction. Elimination of application-related bias requires improved quality assurance practices.
