A. Nilsen, O. Tomic, T. Nęs
Matforsk, Norwegian Food Research Institute
Simple graphical techniques can be utilised to investigate the performance of individual assessors and to detect individual differences among assessors in a sensory panel. By visualising different type of information in a set of various plots the panel leader can extract relevant information without the need of extensive knowledge in statistics. The joint information content gained from the plots can then be used as a basis for selective training in order to improve performance of single assessors and the panel as a whole.
Six sensory panels, both industrial sensory panels that usually perform quality control and trained laboratory panel, took part in an inter collaborative test evaluating six fruit jelly samples using sensory descriptive analysis. Six attributes were used to describe the samples with repeated measurements. The composition of the fruit jelly samples was based on an experimental design with varying levels of sugar, acid and stabiliser.
Various plots can be generated with PanelCheck to visualise panel performance. To compare results from the inter collaborative test the following plots were utilised: Tucker-1 common scores and correlation loadings plots, correlation plots and line-plots.
PanelCheck is a free open source software-tool for visualising data from sensory panels. New versions are published frequently with latest methods for panel performance.
Download from website: 
www.matforsk.no/panelcheck
