Pivot profile method: What is the influence of the pivot and product space?

The Pivot Profile (PP) was proposed recently by Thuillier, Valentin, Marchal, and Dacremont (2015) as an alternative to classical descriptive methods. Its principle is to describe each product by comparing it to a stable reference (called Pivot). While the method seems promising there is little data available and some issues still need to be examined. This paper proposes to evaluate two of these issues: the effects of the similarity within the product space and of the choice of pivot. We compared the pivot profiles obtained for three different sets of beers, varying in their within-set sensory similarity, using different pivots. We found that PP results are more influenced by the within-set similarity than by the choice of the pivot. We suggest that the PP method is more suitable for restricted product spaces in terms of qensory characteristics, and that the creation of a "central product" as the pivot can be a good option when the type of products allows it. However, further studies need to be conducted to assess PP in terms of assessors' repeatability and consensus, as well as to propose alternative statistical analyses that would take into account the individual PP data