Comparison between sensory profiles provided by trained and untrained panels:: Methodology and results

Obtaining sensory profiles classically implies a long training for the subjects. For obvious economic reasons, it is often necessary to reduce this time of training. But what can one await from sensory evaluations provided by an untrained panel? The answer given brought here is based on a large set of data: a same range of six chocolates was evaluated on one hand by six traditionally trained panels and, on the other hand, by a panel having undergone only one training session. It is thus possible to locate this panel inside the variability of the trained panels. In this study, the training has appreciably improved the repeatability of the subjects but, due to its two times higher number of subjects, the untrained panel has a discriminating power comparable with the one of the traditional panel. In the comparison of the profiles provided by these panels, multiple factor analysis provides graphs illustrating, within a single framework, convergences and divergences between panels, from the point of view of products evaluation and from the point of view of descriptors use.