Compusense Research

We investigated results from a study in which 100 consumers evaluated 11 products under blind and informed evaluation conditions. Each sample evaluation included a check-all-that-apply (CATA) question accompanied by a list of 34 emotion terms with instructions to check all terms that described the sample under evaluation. To investigate how the evaluation condition affected consumer product characterizations, CATA responses were organized into a four-dimensional array (assessors, products, terms, conditions). Responses were coded 1 if checked and 0 otherwise. Data were then analyzed following the “one citation, one vote” principle. We used permutation tests based on the L1-norm testing framework to investigate whether the two conditions differed (1) overall, (2) in product citation percentages for term t, (3) in product p citation percentages for term t, (4) in the difference in citation rates for each pair of products overall, and (5) in the difference in citation rates for term t for each pair of products. Results showed the conditions differed on 12 of the 34 terms, but did not differ globally. Elementwise tests showed the informed condition differed from the blind condition in various ways; for example, the informed condition had significantly higher Approval citation rates for several products than the blind condition. Pairwise tests revealed differences overall and on particular terms. Multivariate results visualized following an L1-norm principal component analysis provided further insights. The investigation shows the potential of the proposed approach for investigating CATA data in this type of study.

Chaya, C., Greenacre, M.J., & Castura, J.C. (2026). Investigating differences in check-all-that-apply product characterizations under blind vs informed conditions following the “one citation, one vote” principle. Sensometrics 2026. 20-22 May. Valencia, Spain. (Oral Presentation).