Performance indices in descriptive sensory analysis – A complimentary screening tool for assessor and panel performance

Abstract

Many statistical methods exist for evaluation of different aspects of assessor and panel performance. In order to gain a realistic and exhaustive overview over each individual’s performance in different areas a large number of statistical results or plots need to be considered. Such a process often can be time consuming, cumbersome and may lead to biased conclusions. The proposed performance indices framework aims to act as an effective and practical complementary screening tool for panel leaders to help them quickly detect off-performances by assessors. The framework provides performance indices in the three following areas: agreement, repeatability and discrimination. Performance indices for agreement and repeatability are based on computations of either RV or RV2 coefficients, while the discrimination index is based on results from one- and two-way ANOVA. The performance indices can be easily presented in tables or graphs. Results show that they effectively detect underperforming assessors, and in combination with influence plots, provide a useful first overall impression in a rapid manner. Detailed performance issues can then be studied further in more detail with established statistical methods for performance evaluation.