73G-2 |
Combining information from several descriptive sensory tests to assess judge performance. |
M. GUILLET and N. Rodrigue. Statex, 533, Ontario E Suite 430, Montreal, QC H2L 4A6, Canada A variety of tools exist to evaluate the performance of a panel of judges during a sensory test. These tools have in common to address the situation of a specific sensory session (possibly with several repetitions). However, panels trained for descriptive testing often work on different projects over time. Moreover, those projects may involve similar products and thus the use of the same sensory attributes. Trained judges used for quality control purposes where different lots of the same product are evaluated exemplify this situation. In this context, a simultaneous analysis of data arising from different tests provides a more accurate evaluation of the performance of judges at a given point as well as their evolution over time compared with the overall consensus. For this purpose, a simple statistical methodology can be applied, that allows computing the position of each judge relative to the consensus for each test, taking into account the results of the other tests he/she took part in. Furthermore, the developed model works even if each judge did not perform all tests. Results are presented in the form of control charts with acceptance regions. Each judge is plotted on the graph. Such charts can be drawn for each relevant attribute to identify judges with an unacceptable drift from the consensus. An example in quality control of food in a cafeteria allows determining for each judge at what time additional training is required to recalibrate him/her. Within a company, the same trained panel is often used for a variety of tests. Therefore, being able to combine information from several tests to infer the performance of each judge ensures more accurate results. Furthermore, the graphical presentation of the results provides quick decision-making tools for deciding whether to include a judge in a panel or give him additional training.
Session 73G, Sensory Evaluation
|