Construct validity and content analysis.

We clarify points of agreement and disagreement with Schnurr, Rosenberg, and Oxman, as put forth in their article appearing this issue. Speech samples elicited under different conditions may or may not be interchangeable. This is a matter of empirical research, but it is also dependent upon the meaningfulness of the content categories. used in the analysis. We prefer a theory-driven, construct validational approach that employs well validated coding schemes to assess psychologically meaningful concepts.