Statistical penalty analysis

• Penalty analysis displays the link between JAR variables and a measure of liking.
• This paper enriches the usual penalty analysis.
• The penalty analysis graphic is enriched by confidence intervals.
• Links among the JAR variables is highlighted by multiple correspondence analysis.

Penalty analysis is a very popular method in the food industry. We show how to enrich the analysis by visualizing, in the usual graphic, the uncertainty in penalties by the way of lines representing confidence intervals. It is the occasion to discuss the data on which the penalties are calculated: we show the interest to use simultaneously the data for all the products. Moreover, such analysis implies to consider the links between the JAR variables (just about right): multiple correspondence analysis offers a visualisation of these links. The whole constitutes a complete methodology for the improvement of products.