In the master's thesis, we determined the practicality of three selected discrimination tests for their use in sensory laboratories with an emphasis on their sensitivity given the number of assessors available. With 90 untrained assessors, divided into panels of 12, we conducted three discrimination tests: triangle, three-alternative forced-choice (3-AFC) and degree of difference (DOD), which were performed in two replicates with three sessions and longer intermediate breaks in order to provide each panel with a different series of apple nectar based samples each time. In all tests, they had to choose the sample that differed from the other two. Sensory analysis data were statistically processed using the χ2 test. We concluded that the triangle proved to be the most sensitive test, although the 3-AFC has inherently greater statistical power. The DOD also didn't appear to be more sensitive compared to the triangle, as they have similar decision strategies. We also found that the repeat tests did not appear to be more sensitive compared to the original tests. We didn't detect other notable similarities or differences in the sensitivity of the triangle, 3-AFC and DOD. Assessors were equally successful at recognizing all series of modified apple nectar samples as well as differently composed triads of samples in the triangle test. We also didn't prove a connection between the success in choosing the correct answer and the assessors’ confidence in the given answer, or a connection between the success in choosing the correct answer and the type of known differences.
|