A comparison between nine laboratories performing triangle tests

Fifteen groups of participants in nine laboratories performed triangle tests with two pairs of soft drinks. Groups differed in practice level with triangle tests: eight groups of 60 consumers who were not used to triangle test, three groups of qualified assessors who have already performed a few triangle tests, and four groups of trained assessors with a more extensive practice of triangle tests; qualified and trained groups included 9 or 18 assessors. The soft drinks were made from syrups at two levels of dilution in order to achieve about 55% of correct responses to test for difference and about 40% of correct responses to test for similarity. Participants performed three replicated tests with each pair of drinks, except the groups of 9 assessors who performed six replicated tests. When testing for difference, large inter-groups differences were observed. One consumer group and one trained group from two different laboratories failed by far to reach the critical number of correct responses leading to demonstrate a significant difference between products. For similarity test, all consumer groups demonstrated a significant similarity whereas two qualified groups and one trained group did not. This is explained by a slightly higher level of performance for qualified and trained assessors compared to consumers.

► Fifteen groups of consumers or selected assessors performed triangle tests with two pairs of products. ► Data from one pair were tested for difference, the other one for similarity. ► When testing for difference, large inter-groups variations were observed. ► Two groups failed to demonstrate a significant difference. ► A significant similarity was demonstrated for all consumer groups but not for three selected groups.