Multivariate analysis of Just-About-Right data with optimal scaling approach
Abstract
In consumer research, Just-About-Right (JAR) scales are commonly used to identify sensory attributes perceived to be at their optimal level or, on the contrary, at too high or too low level. For most of the published works, JAR data are combined with overall liking scores to identify a top list of sensory attributes that might be considered for product reformulation, optimization or characterization. The penalty analysis is a popular and relevant way to get these information but each attribute is considered independently. However, considering all the attributes may provide a complementary point of view and component-based multivariate analyses (e.g Principal Component Analysis or Multiple Correspondence Analysis) could be applied to investigate the relationships between attributes. The analysis of JAR scales with multivariate methods has not been addressed much in the sensory community. JAR scales being bipolar, they will be considered as nominal and Principal Component Analysis with Optimal Scaling (PCAOS) is investigated. This method performs a component-based multivariate analysis with an optimal quantification of JAR data. This method is applied to a JAR sensory experiment on French cheeses.