Mining a database of sensory descriptive analyses to understand panellist performances
N. Pineau, D. Brajon, S. Cordelle, P. Schlich
Centre Européen des Sciences du Goût, Dijon, France
The raw data from numbers of sensory descriptive analyses are currently being collected by the authors and stored into a database. The data providers are food and flavour industries, technical institutes, public research institutes and universities from several countries. They send their datasets through the internet and receive statistical results on the panellists’ performances presented in a set of html pages (see poster of Schlich et al. in the same conference). In return for this free statistical analysis, the data providers agree to let their data contribute to the base. The goal of this data collection is to document the practices of sensory analysis, to conduct meta-analysis of these data to better understand panellist performances (age, gender, level of training, type of attributes and products …) and finally to offer a benchmark for testing and comparing sensometric techniques.
The number of studies already gathered is about 50 (January 2005) and is expected to be above one hundred by the time of the conference. As an example of meta-analysis, the figure below investigates the effect of age and gender on the ability to discriminate among products. One can observe a decrease with age of this ability and no clear gender effect yet. By the time of the conference, the larger number of datasets available will increase the power of such analysis and will likely show more significant effects.
We greatly acknowledge ESN (European Sensory Network) to provide travel grant to N. Pineau