Comparison of methods for generating sensory vocabulary for further use in CATA studies with consumers

A fundamental basis of the CATA method is a predefined list of vocabularies. However, it has not yet been sufficiently investigated which methods are suited best for generating attributes which are appropriate for a description and differentiation of product samples by untrained consumers and to what extent the CATA results differ due to the different vocabulary lists. Gaining a deeper understanding of whether and how different methods of generating sensory vocabulary will affect the applicability of these attributes in CATA studies (i.e., their suitability for describing and differentiating product samples) can contribute significantly to increasing the quality of CATA studies.
The present study compares three methods for developing sensory vocabulary focusing on taste terms. Three CATAs with 10 standardized chocolate bar samples and untrained consumers (n = 311) were conducted. The vocabulary was taken from a previous napping study with untrained consumers, a qualitative sample description with experts and an existing vocabulary list of chocolate products.
The three determined attribute lists have been cleansed using the Cochran Q test and np chart by eliminating unsuitable terms for description and differentiation of product samples as well as inappropriate subject data.
The analysis indicates that napping with untrained consumers generates the highest proportion of suitable sensory attributes (79.2%), but also that after data cleansing the objects coordinates on the bi-plots indicates a high consenses. To illustrate this, a GPA was calculated with the coordinate data. The result clarified this impression with a tucker´s congruence coefficient of Rc .861.
Our results demonstrate that by focusing on essential attributes that are suitable for the description and differentiation of product samples in CATA studies with consumers, attributes can be generated using different methods. However, they also illustrate that there is potential for more consumer-oriented approaches in the case of ready-made attribute-lists out of literature.