STATISTICAL VALIDATION OF TRAINING OF HIGH PERFORMANCE JUDGES FOR DESCRIPTIVE ANALYSIS OF DARK CHOCOLATE

For application QDA®, it is essential that the group of judges is pre-selected and subsequently selected by statistical proof: discrimination of attributes between samples, repeatability of their own results and agreement between team members (values of sample pF < 0.5, repeat pF > 0.05 and agreement between participants). The objective of the work was to demonstrate the importance of pre-selection and rigorous training of assessors to obtain a high-performance team to carry out QDA. Participants determined and defined each of the descriptor terms so that everyone determined the same methodological language and sensory memory. In the study, 23 advisors were pre-selected, by triangular tests with Wald sequential analysis, who participated in determining the descriptor terms, definition and their respective references. After pre-selection and training the assessors analyzed seven samples of dark chocolate with different plant extracts in four replications. The samples were presented coded, in complete blocks balanced in a sequential monadic manner. The results were analyzed by two-factor ANOVA (samples and repetition and interaction between sample and panelist) in relation to each of the descriptor terms, using the SAS v.9.4 software. It was proven that the 23 pre-selected evaluators were prepared to participate in the QDA, ensuring good efficiency in obtaining results. The training results were statistically validated, and all tasters who presented sample pF values < 0.50 (were able to differentiate the samples in relation to the descriptor terms that showed differences) and repetition pF > 0.05 (showed repeatability of their own results in relation to each descriptor term) and agreement between group members, were approved to determine the descriptive sensory profile. The sample pF value < 0.50 was important for selecting advisors with good performance as mentioned by authors (Damasio and Costell, 1991) in research with products in which the texture is of high importance and complexity.