A COMPARISON OF METHODS FOR EVALUATING THE PERFORMANCE OF A TRAINED SENSORY PANEL<sup>1</sup>
Сравнение методов оценки работы обученной сенсорной панели
2001-12-01
SCID: 54.1/ce89hett
Discuss with AI
GRAPES programcluster analysisconsonance analysisprincipal component analysissensory panel performance
Figures from the paper
Abstract (AI)
ABSTRACT Cluster analysis, consonance analysis, principal component analysis (PCA) and the GRAPES program (Schlich 1994) were compared for the evaluation of panel performance. Ten judges evaluated 25 Merlot wines for 24 color, aroma and flavor attributes. Cluster analysis grouped similar judges. PCA identified judges according to their attribute use. Consonance analysis determined a numerical index for attribute agreement and the GRAPES program compared judges in their use of the scale, reliability, discrimination and disagreement. Three of the four techniques provided a graphical representation of similarities and differences between judges. Methodologies were best used in conjunction with one another. Ultimately the application of these tools will serve to improve the quality of sensory evaluations.
Key Findings
1
Cluster analysis grouped similar judges, while PCA identified differences in how judges used attributes.
2
Cluster analysis, consonance analysis, PCA, and GRAPES were compared for evaluating trained sensory-panel performance.
3
Consonance analysis quantified attribute agreement, and GRAPES evaluated scale use, reliability, discrimination, and disagreement.
4
The study analyzed ten judges assessing 25 Merlot wines across 24 color, aroma, and flavor attributes.
5
Three methods provided graphical representations of judge similarities and differences; the methodologies were most effective when used together.
Research Object
Performance of a trained sensory panel evaluating 25 Merlot wines across color, aroma, and flavor attributes
Research Subject
Comparative evaluation of judges' similarity, attribute use, agreement, scale use, reliability, discrimination, and disagreement using four panel-performance analysis methods
Publication Details
Publication Date
2001-12-01
Journal
Publisher
ISSN
Access Type
Author Information
Download PDF
Subscribe to digest