Testing for Compliance with a Performance Specification

Statistical data from an interlaboratory trial of a new test method are applied to yield advice on quality-control criteria. It is shown how repeatabilities and reproducibilities of the test can be converted into critical differences for comparison with a specified fastness rating. Adding the right critical difference to the rating tells one what result to lay down as a quality-control target for a defined, high probability of the bulk goods conforming to the specified rating. The difference between ‘two-tailed’ and ‘one-tailed’ limits is explained. The stages in the procedure described include: correlating critical difference with fastness rating; comparing limits for different numbers and types of tests and assessors; suggesting the most useful procedures; deriving the minimum results needed to satisfy different specifications for the particular test; and explaining the related appendix to the test method on precision for quality control. Information is given on ways to alter limits according to particular needs