Instability and heterogeneity: a new approach needed!

Abstract

The outcome of routine testing for instability and heterogeneity of materials used in proficiency tests tends to be weakly informative because of the unacceptable cost of obtaining enough data to support a suitably powerful statistical test. A better strategy would be to spend considerably more on a large initial validation of a type of material or a procedure for preparing it, and then to monitor all subsequent batches of material of that type by a much cheaper quality control system. The expense of a powerful validation would be quickly recouped by savings in the longer term.