Proficiency Testing in Analytical Chemistry

Abstract

Proficiency testing is described in sufficient detail for the chemometrician to appreciate how the requirements of analytical chemists have determined the shape of the practice. Aspects of proficiency testing treated in more detail are specifically those requiring a close look at the application of appropriate statistical methods. It is noteworthy that some statistical methods required are well beyond those found in typical degree or masters courses in analytical science. The frequency distributions found in the results of proficiency tests often deviate from the normal and may require inter alia robust methods of estimation, kernel density displays, and mixture modeling to describe them adequately. When a test material is distributed to participants, the portions received have to have the same composition to within narrow limits or comparison of results is meaningless. This calls for tests for instability and heterogeneity, but there are pitfalls for the unwary in these tests, and again obvious methods may not be applicable in their simple form or may have low statistical power.