Advances in representation and analysis of mono and multi-intake Temporal Dominance of Sensations data

Mara Virginia Galmarini (1, 2) , Michel Visalli (1) , Pascal Schlich (1)

1 CSGA - Centre des Sciences du Goût et de l'Alimentation [Dijon]
2 National Scientific and Technical Research Council

Using the concept of dominance, other than intensity, Temporal Dominance of Sensations (TDS) has become one of the most used temporal descriptive techniques providing information on sequentiality and duration time of dominant sensations. TDS is now evolving towards the evaluation of successive product intakes (multi-bite, multi-sip). This gives richer information, but it is challenging in terms of data analysis and representation, which should be user friendly but faithful to product characterization. In the present work we propose: (i) a new graphical tool for representing the sequentiality of (multi-intake) TDS data and (ii) a new inferential approach to data analysis based on duration of dominance. TDS data are generally presented as TDS curves. However, when plotting several successive intakes, visual inspection and curve comparison among intakes (and products) becomes a cumbersome task. To better show sequentiality of dominant attributes, we propose the TDS band-plot by descriptor This representation shows, by attribute, whether they are dominant or not along consumption (at panel level) allowing the follow-up of each descriptor and improving visual comparison. Nonetheless, visual assessment is not enough to determine significant differences. Using the total duration of dominance for each attribute, three-way ANOVA's including subject, product and intake as factors, together with their two-way interactions, with subject as random effect and intake as a repeated measure, were performed. This reveals if a certain attribute is dominant for a significantly longer or shorter period of time in some products, as well as its changes over intakes; interaction between product and intake being highly important. Further, differences can be summarized over attributes by a MANOVA and represented by a Canonical Variate Analysis (CVA), allowing the comparison of different intakes of a same product, or different products, or altogether. These advances will be illustrated by data from wine and cheese multi-intake TDS done by consumers.

Mots clés: manova; anova; cva; tds curves

Domaines: Alimentation et Nutrition

Mara Virginia Galmarini, Michel Visalli, Pascal Schlich. Advances in representation and analysis of mono and multi-intake Temporal Dominance of Sensations data. 11. Pangborn sensory science symposium, Aug 2015, Gothenburg, Sweden. 1 p. ⟨hal-01512158⟩