Valid inferences from invalid tests?

I read with interest the remarks of Reuben J. Silver, which were published in the November 2001 issue of the American Psychologist on the occasion of his receiving the Award for Distinguished Contributions to Applied Psychology as a Professional Practice. Silver took the opportunity to delineate his approach to the practice of professional psychology. He described a perspective that combines an appreciation for actuarial prediction and empirical data with a sensitivity to the many instances when a practitioner must act without the benefit of empirical support. In this regard, Silver imparted a wonderful example of the integration of science and practice. However, I must disagree with one aspect of what Silver conveyed regarding the use of psychological tests. Silver (2001) emphasized the importance of accurate diagnosis and the use of empirically validated psychological tests to help achieve accurate diagnosis. He then stated,