Product discrimination ability of temporal sensory evaluation methods used with consumers

The objective of this study was to determine which temporal sensory evaluation methods (TDS, TCATA, AEF-D, AEF-A), variables (citations, durations, times of citation) and statistical analyses are the most discriminative when measurements are collected from consumers. This study is based on data collected on model stimuli delivered by a gustometer (controlled temporal differences) and on four categories of commercial food products (crisps, guacamoles, chocolates, ice teas) presenting different levels of sensory complexity. With controlled stimuli, analyses of citations and durations both revealed the same differences, suggesting that product discrimination is mainly brought by presence/absence of sensory attributes. Compared to TCATA, TDS highlighted slightly more significant differences (including expected and unexpected differences). Short differences in durations were not captured. AEF-A was less discriminative, but presented no validity nor repeatability issues. With commercial food products, analyses of citations were more discriminative than analyses of durations and times of citations. However, the combined analysis of citations and time-related variables resulted in a small gain of discrimination. Applicability-based methods (TCATA and AEF-A) were overall more discriminative compared to dominance-b ased methods (TDS and AEF-D), but had more repeatability issue. These results suggest that when data are collected from consumers, in most situations analysing durations and times of citations would not bring much more information compared to a time-independent analysis of citations. Performances of TCATA and TDS are almost similar when used with simple stimuli presenting clear temporal patterns. With complex food products, TCATA is the most discriminative, but also the less reliable, and AEF-A presents the better compromise between discrimination and repeatability. This study showed that there is probably no ultimate method, and that every gain in temporal resolution or discrimination is assumably at the cost of a loss of reliability.