Sequential procedures for triangular and paired comparison tasting tests

Abstract
Summary: The necessary number of tasting tests which must be carried out in the comparison of two samples by triangular tests or paired comparisons is considered in relation to: (1) the risk of stating that the samples are different when they are in fact not distinguishable, (2) the risk of stating the samples are identical when a distinguishable difference really exists, and (3) the degree of ‘difference’ it is desired to distinguish.Sequential procedures are put forward which enable a decision to be reached relatively quickly, with 1 in 20 risks of error, when the two samples either are identical or differ by a defined amount. Separate tables are given for difference testing and quality testing. In cases where the sequential test would be lengthy the procedures terminate the tests after an amount of tasting which only slightly increases the risk of error.