The effect of dimensionality on results from the triangular method

Publisher Website
Google Scholar
Abstract
Methods to measure differences between complex stimulus sets (such as foods and beverages) are numerous One of the most commonly used procedures in food and beverage sensory research is the triangular method. A comparison of unidimensional and multidimensional normal models for the triangular method using Monte Carlo simulation showed that the expected subject response distribution depends not only on the size of the unidimensional discriminal distance between stimulus sets, but also on the number of dimensions for which the discnminal distance is zero in each case Since the number of dimensions for which these conditions apply are usually unknown in complex systems, the power of the triangular method will be unknown. These findings may have important implications for the interpretation of results from many methods which involve a comparison of distance estimates in a multidimensional space.

This publication has 0 references indexed in Scilit: