Directed Sorting Using Trained Assessors for Categorization of Honeybush Tea: Comparison of Correspondence Analysis and DISTATIS Combined With Barycentric Text Projection

Abstract

The validity of sorting as rapid method for sensory characterization of food has been demonstrated, however, research on its effectiveness to profile complex products, especially when using trained assessors, is limited. In this study, directed sorting based on aroma (sorting^AROMA), palate (sorting^PALATE) and both aroma and palate attributes (sorting^GLOBAL) were conducted by trained assessors to categorize honeybush tea prepared from five Cyclopia species. Descriptive sensory analysis provided a detailed sensory profile of the same samples for comparison. Sorting^AROMA and sorting^PALATE resulted in similar sample configurations. Sorting^GLOBAL was less effective at grouping samples per species. Despite the use of trained assessors, three replicate sessions were needed to ensure consistent results, particularly for sorting^AROMA. Both DISTATIS, combined with Barycentric Text Projection (BTP), and Correspondence Analysis (CA) produced similar correspondences between attributes and sample groupings. The bootstrap ratios of DISTATIS-BTP confirmed significant sample-attribute associations, thereby providing additional information.