162 / 274
2
INTRODUCTION
Proficiency testing (‘PT’) is an economical approach to a collaborative study which has the
specific principal goal of measuring a participating collaborator result with respect to the mass of
the other collaborator results. This differs in aim from a randomized, controlled collaborative
study that is designed specifically to measure repeatability, reproducibility and bias. PT studies
are generally performed for a nominal (middle) concentration of analyte in a particular matrix.
Designed collaborative studies typically span the gamut of practical concentration levels and use
challenging matrices. Participants in PT studies may use nominally the same method, but
typically there is no direct control over the exact protocol used. In designed collaborative studies,
the precise protocol is specified. In PT studies, replication may or may not be present, and may
vary among participants, sometimes without disclosure.
Traditionally, ‘robust’ statistical methodology has been used to analyze PT data. In TR322 and
TR323, the use of such statistics for estimating reproducibility was deprecated.
Here guidelines are given for valid use of data and ‘robust’ statistical estimates derived from PT
studies for repeatability and reproducibility. (See TR323 for more discussion.)
The choice of performing or not performing a designed collaborative study is that of the method
developer. The principal premise assumed here is that of ‘caveat developer’: Statistical estimates
are to be designed to be conservative with respect to method approval.
GENERAL GUIDELINES
1. Results must be reported as pertaining only to the specific matrix and concentration
involved.
2. The combined set of estimates across all studies will be considered adequate only if the
gamut of low to high concentrations for each matrix are studied.
3. All statistical estimates must be reported with 95% confidence intervals. These intervals
are important to making the quality of the data visible to reviewers.
110