CATA questions for sensory product characterization: Raising awareness of biases

Abstract Research into sensory product characterizations by consumers using CATA (check-all-that-apply) methodology is entering the stage where better understanding is required of ways that CATA question implementation influences sensory profiles and product discrimination. With this aim, five studies were conducted. Focusing on strategies that have been suggested as ways to reduce primacy and order bias in CATA data, this research explored the effect of these initiatives on the elicited sensory product profiles and conclusions drawn with regards to sample differences. The use of within-subjects randomization of CATA terms did not significantly affect frequency of use of CATA terms but significantly affected conclusions regarding differences among samples, compared to the evaluation of multiple samples with CATA ballots where terms are presented in the same order. The use of multiple shorter CATA questions defined by sensory modality also yielded results that significantly differed from when single longer CATA questions with terms from multiple sensory modalities were used. CATA question length and the use of single/mixed sensory modalities both appeared to be contributing factors to these differences. ‘Dynamics of sensory perception’, which refers to effects on CATA results arising when CATA terms are assessed in the sensory evaluation process (during/after sample consumption) was confirmed as a source of bias. A key learning from this research was that sensory product characterization and differences among samples are subject to multiple minor biases related to how the question is formulated and that the exact experimental conditions under which CATA data are generated should be reported.