Affordable Access

Publisher Website

Defining and validating assessor compromises about product distances and attribute correlations

Elsevier Science & Technology
DOI: 10.1016/s0922-3487(96)80033-5


Publisher Summary This chapter introduces statistical framework allowing firstly to deal with free-choice profiling (FCP). The basis of this framework is the RV coefficient, which is a generalized correlation coefficient between two sets of variables recorded from the same samples. The chapter discusses techniques that are applied to an interlaboratory sensory analysis of 16 coffees evaluated by 11 different trained panels located in 8 different European countries. The chapter discusses the advantages and limitations of the techniques proposed. The advantages include the ability to take flee-choice profiling into account, RV coefficient for measuring similarity between two sample spaces, permutation tests for validation of the panel homogeneity with almost no computation, β coefficient of individual dimensionality (complexity), and the compromise obtained as a mean of assessors weighted by their individual agreement with the panel. The principal limitations include possible over-powerful tests, no analytical permutation tests with dual STATIS (Structuration des Tableaux A Trois Indices de la Statistique), and that individual weights only depend on agreement with the panel.. The comparison of these two lists makes it clear that the RV-related techniques have a great potential in the sensory field.

There are no comments yet on this publication. Be the first to share your thoughts.


Seen <100 times