Sensory profile of peach nectars. Evaluation of assessors and attributes by generalized Procrustes analysis

Abstract

A trained panel evaluated eight peach nectars by conventional descriptive profile analysis. Two-way analysis of variance showed significant (α ≤ 0.05) differences between samples for the 13 attributes evaluated. For nine of them, interaction between samples and assessors was also significant. The internal structure of the data was analysed by generalized Procrustes analysis. It was found that two attributes (granularity and aftertaste) were not good discriminators between samples and that two assessors (1 and 2) differed considerably from the average response. Principal component analysis was applied to the mean scores of samples, considering all attributes and computing means for all assessors and after removing the data for two attributes and two assessors. Similar sample configurations were obtained in both cases. The two first axes explained more of the total variability in the second case (78.8% vs 69.6%).