SIMILARITY TESTS USING FORCED-CHOICE METHODS IN TERMS OF THURSTONIAN DISCRIMINAL DISTANCE, d′

Abstract

ABSTRACT This paper discusses similarity tests using forced-choice methods in terms of Thurstonian discriminal distance, d′, a measure of pure sensory difference or similarity. Critical values are derived for the similarity tests using the 2-alternative forced choice (AFC), the duo-trio, the 3-AFC and the triangle methods in terms of d′. Power analysis is conducted for the similarity tests using the four forced-choice methods. Three approaches are used to estimate sample sizes for the tests. The approaches are based, respectively, on exact binomial distribution, normal approximation with a continuity correction and Monte Carlo simulation. S-Plus and R codes were developed and provided for calculation and estimation of the critical values, testing powers and sample sizes for the tests. PRACTICAL APPLICATIONS Similarity testing has wide application perspective in sensory and consumer research. It should be used when similarity rather than difference is of interest. The similarity tests discussed in the paper can be used to infer if a difference, in terms of d′, between, e.g., a prototype and a control, is smaller than a similarity limit, i.e., a specified allowed difference.