Premium
A Statistical Agreement‐Based Approach for Difference Testing
Author(s) -
CalleAlonso F.,
Pérez C.J.
Publication year - 2013
Publication title -
journal of sensory studies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.61
H-Index - 53
eISSN - 1745-459X
pISSN - 0887-8250
DOI - 10.1111/joss.12061
Subject(s) - complement (music) , computer science , statistical hypothesis testing , prime (order theory) , data mining , test (biology) , mathematics , artificial intelligence , natural language processing , machine learning , statistics , paleontology , biochemistry , chemistry , combinatorics , complementation , biology , gene , phenotype
Abstract A statistical approach based on measures of agreement is proposed for use in a sensory analysis context. This approach considers the idea of using statistical agreement to provide information on the homogeneity of the raters' responses, so that this information can then be used to discriminate between products. It can also be used to measure the expertise level of raters. Although the prime focus is on difference testing by the triangle test ( ISO 4120:2008), the proposed methodology can also be applied in other contexts such as the paired comparison test ( ISO 5495:2009) or the duo‐trio test ( ISO 10399:2010), among others. The proposed approach is not a substitute for binomial statistical analysis, but rather it can be used as a complement. It is especially useful when few panelists are available and replications are needed. An experiment that evaluates two types of Iberian dry‐cured pork loins through the triangle test is performed to illustrate the applicability of the proposed approach. Practical Applications The proposed methodology is very helpful for users of sensory analysis techniques, and especially for those focused on differentiation tests. It can complement the classical methodologies on this topic. It is especially useful when there are few raters and replications must be performed. It can also be applied to evaluate the accuracy of panelists being trained by an expert.