Proficiency testing interpretation based on analysis of variance method

Abstract

Proficiency testing is the use of interlaboratory comparisons to assess the ability of a laboratory to perform a particular test or measurement competently and demonstrate the reliability of the results generated. To achieve this goal, the international standard ISO 13528:2022 presents four performance statistics and one of them is the z score. Some researchers have proposed an alternative approach for proficiency testing scheme interpretation based on an analysis of variance method. This short communication presents residual analysis to choose an appropriate analysis of variance model for proficiency testing interpretation when the assigned value is obtained by consensus value among participants. The analysis of variance approach was compared with z score to assess the performance of interlaboratory comparison for determining the amount of NOx present in vehicle emissions. Preliminary results indicate that the conclusions obtained by both methodologies differ for one of the participating laboratories. This occurs due to violation of the homoscedasticity assumption which is not considered when using the z score. On the other hand, residual analysis can provide such information and feasible generalized least squares can deal with heteroscedasticity.