Measuring overall difference from a combination of attribute ratings with the many-facet Rasch model

Abstract

The Total Intensity Measure (TIM) approach offers an innovative solution for quality control by combining ratings of individual sensory characteristics into a single measure using the Many-Facet Rasch Model (MFRM). While the traditional Difference-From-Control (DFC) test is simple and useful for comparing products against a standard, it requires significantly more samples when examining a larger number of products unlike in attribute difference tests. This study aims to determine if the TIM method can serve as an alternative to the DFC method when comparing samples against a control. An untrained panel (n = 67) evaluated three UK commercial brands of Jaffa cakes using attribute difference and DFC tests. Assessors evaluated samples in triplicates according to each test's procedure on two different days. Friedman tests on the DFC scores compared to Rasch-produced measures of the combined attributes both showed significant differences between samples (P < 0.01). Pairwise comparisons with a control (α = 0.01) for the DFC showed only one brand was different from the control, while the TIM showed that both brands were different from the control. Additionally, the Many-Facet Wright map showed the degree to which each attribute contributed to the overall difference. Of the five attributes evaluated, Sweetness and Orange flavor contributed the most followed by Cocoa flavor. Milky flavor and Saltiness did not contribute significantly, highlighting that while all attributes were assessed, only certain ones had a notable impact on the overall product differences. The proposed method is potentially beneficial to sensory analysts in obtaining better diagnostic information to support decisions about product differences.