Representing Assessments: Reply to McGrath and Ozer

Abstract

Hofstee and Ten Berge (2004/this issue) present procedures based on an absolute conception of scales in personality assessment as distinct from the dominant interval-scale interpretation. On application, these procedures resulted in a contraction of a 5-dimensional structure into essentially 1 personality dimension. McGrath (2004/this issue) and Ozer (2004/this issue) comment on the various aspects of these procedures: the transformation of data onto a bipolar proportional scale, the adoption of the raw scores product average as an index of association, and raw scores principal component analysis. In reply to these comments, we emphasize that the central ingredient in our procedures is the interpretation of the midpoint on a bipolar scale as a threshold. We provide further arguments for that interpretation and demonstrate the robustness of the simplified structure under that central assumption. We acknowledge the comment that our conception does not fit in comparative contexts capitalizing on individual differences but argue that other contexts involving thresholds are relevant to the study of personality. We also acknowledge that item pools should be sufficiently homogeneous for scales to be meaningful.