Blog Entry

Statistical models vs. "expert" psychologists

The superiority of statistical formulas in predicting gives rise to what can be termed a "base rate" psychology. People's behavior and feelings are best predicted by viewing them as members of an aggregate and by determining what variables generally predict for that aggregate and how. That conclusion contradicts experts' claims to be able to analyze an individual's life in great detail and determine what caused what. Unfortunately, it is exactly the individualized-causality type of analysis that is most expected of professional psychologists and other mental health professionals. This expectation arises not only from our intuitive beliefs about the world but from these psychologists' own declarations about their abilities. [...]

Moreover, as we have seen, the inability to predict implies a lack of understanding--not because understanding and prediction are synonymous but because a claim to understanding implies an ability to predict. Evaluating the efficacy of psychotherapy has led us to conclude that professional psychologists are no better psychotherapists than anyone else with minimal training--sometimes than those without any training at all; the professionals are merely more expensive. Moreover, in predicting what people will do, clinicians are worse than statistical formulas, and statistical formulas are a lot less expensive; [...] Why not instead put our efforts into improving the methods we know to be superior by developing better statistical models? That should benefit almost everyone--except, of course, the people who are being highly paid to make inferior predictions.

House of Cards: psychology and psychotherapy built on myth
Robyn Dawes

Posted on:
2008.09.23 -0500

Tags:
texts