The use of personality trait measurement is increasing in sensory evaluation for linking certain variables (i.e., consumption behavior and product preferences) to particular attributes. For this study, 976 consumers rated agreement on 44 statements from the Big Five Inventory using a 5-point Likert-type scale. Data handling methods for personality segmentation were compared: (1) the original 44 variables versus the five computed personality variables; (2) standardization versus nonstandardization of data; and (3) k-means versus Ward's hierarchical clustering method used with principal component analysis.
Results indicate using the five computed variables in mapping gave higher percentages of explained variability because of the small number of input variables. However, maps created from the 44 individual variables illustrated that participants were distributed throughout and separated visually into groups. Standardization of the data set did not affect mapping or classification. k-means and Ward's clustering methods provided different classification results within the same data set.
Results suggest that when using the Big Five personality traits measurement, the original 44 unstandardized variables and k-means clustering should be used for obtaining consumer segmentation because this captures the greater variability inherent in the 44 variable tests and easily separates consumers into personality groups.