Personality out of proportion?

The field of personality assessment has evolved the normative practice of centering scores on their means, evaluating associations among measures with Pearson correlations, and using factor analytic methods to reduce redundancy and provide putative explanatory variables. At least some of these explanatory variables, or factors, have become well-known elements in trait theories of personality structure (e.g., the Five-factor model). Hofstee and Ten Berge (2004/this issue) suggest an alternative set of procedures arising from treating the midpoints of bipolar rating scales as true zero points. These procedures lead to a very different view of personality structure in which one factor provides a nearly sufficient summary of personality judgments. I scrutinize the methodological choices implied by these procedures here. This evaluation leads to the conclusion that Hofstee and Ten Berge provide methods and results that cannot serve to replace normative practice and well-known findings but do provide insight into important questions not typically addressed by personality assessors.