Replicates in sensory profiling: Quantification of the impact of moving from two to one assessments. (April 2018)

• Record Type: Journal Article
• Authors: Moser, Mireille; Lepage, Mélissa; Pineau, Nicolas; Fillion, Laurence; Rytz, Andreas
• Is Part Of: Food quality and preference. Volume 65(2018)
• Page Start: 185
• Page End: 190
• Publication Date: 2018-04
• Subjects: Sensory descriptive profile -- Replicates -- Estimates -- Discrimination -- Food industry; Food preferences -- Periodicals; Food -- Quality -- Periodicals; Food industry and trade -- Quality control -- Periodicals; Préférences alimentaires -- Périodiques; Aliments -- Qualité -- Périodiques; Aliments -- Industrie et commerce -- Qualité -- Contrôle -- Périodiques; Food industry and trade -- Quality control; Food preferences; Food -- Quality; Periodicals; 664
• DOI: 10.1016/j.foodqual.2017.12.002
• Languages: English
• ISSNs: 0950-3293

Abstract

Highlights: This work assesses the loss of using 1 assessment rather than 2 in sensory profiling. Results are based on 15 real datasets from food industry. Results show very limited impact on the obtained sensory profile. These results can help optimizing sensory panel resources (e.g. focus on training) Abstract: Sensory descriptive profiling is a sensory method commonly used in the food industry to describe products' sensory characteristics. The usual recommendation is to use a panel of trained panelists, generally around twelve, and to assess the products at least in duplicate (ISO 13299:2016) to get robust average estimates. In practice, in order to optimize the use of the panels, the necessity of duplicates was challenged. This work presents a comparison of results obtained with one vs. two assessments, based on 15 studies featuring at least 8 products each, representing diverse product categories, diverse panels, a total of 309 attributes and 2836 comparisons of product pairs. It is shown that average estimates are very similar for one vs. two assessments: correlations are very high for all attributes that discriminate products. It is also shown that 89.4% of pairwise comparisons lead to the same conclusions for one and two assessments. The remaining 10.6% are misbalanced showing more significant comparisons with 2 assessments than with 1 (8.0%) more often than the reverse (2.6%). As a conclusion, one assessment is considered sufficient in most cases, provided that the sensory panel has been sufficiently well-trained.