Measurement and prediction, clinical and statistical.

Clinical vs. statistical prediction is only ? the problem—and the last ? at that. The prior problem, largely neglected, is clinical vs. mechanical measurement—for data can be collected, as well as combined, in either way. Such neglect promotes an incomplete and mismatched dialogue where "clinical" and "statistical" may have different meanings to different persons. Examining clinical and mechanical methods of both measurement and prediction provides a broadened framework that defines the several possible "clinical" and "statistical" methods, and their combinations. Applying this framework to 45 studies shows an apparent superiority for mechanical modes of both data collection and combination, and also suggests that the clinician is more likely to contribute through observation than integration. Grossly uncontrolled differences, however, in clinical training, Ss, criteria, etc., prevent definitive conclusions. To achieve more adequate comparison requires certain specified methodological improvements. (3 p. ref.) (PsycINFO Database Record (c) 2010 APA, all rights reserved)