Although the panel performance and the panellist performance seems to be linked, at least from an interpretation point of view (one could expect that the panel is discriminating if panellists are discriminated), the analyses involved different data (the panellist performance is based on subset of data, while the panel performance uses all the data at once). Hence, it can happen that assessors are not so well discriminating while once combined, the overall results show good discrimination.

The panel and panelist performance do not rely on the same statistical analysis. Panel performance is based on ANOVA, and the agreement is based on the product:assessor interaction.The panelist performance is taking only the panelist data, and the agreement at an individual level is based on the correlation between that assessor scores and the average across panels.These 2 stats are completely different, have completely different implications/assumptions, so cannot be compared one to one. Also, in some situations, the lack of disagreement can be simply due to a different use of the scale, meaning that the assessors are still “ranking” the samples in the same way, BUT are emphasizing more or less certain differences.