Principles of Principal Component Analysis

With increasing frequency consumer studies are supplementing demographic and price variables with responses to an extended set of Likert-scale questions to elicit information on consumer motivations and attitudes. Principal compo­nent analysis (PCA) is a statistical tool that reduces a large number of variables to a smaller set of "components" that describe as much as possible of the variation in the original variables. Attitudinal responses can then be represented by component scores in statistical models. This paper reviews fundamental principles of PCA and concludes with a proposal for collaborative efforts to standardize attitudinal questions and PCA of responses across studies.