Commentary on Meehl’s theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology

Abstract

Meehl points out that in “soft” psychology we do not have quantitatively measured variables that allow generalization outside the context of study to other contexts involving these same variables (perhaps combined with others to yield prediction and control). But as illustrated by this own work comparing actuarial versus clinical combination of variables to predict important human outcomes, we do have consistency across qualitatively diverse contexts. Here, actuarial combination is almost always superior. That result might not be what we would wish to have (e.g., a very good prediction of who will succeed in graduate school or on parole—as opposed to knowing that an actuarial prediction is superior to a clinical one), but it is what we have. It is not nothing, in fact far from nothing. Lacking a statistical theory (or even definition) of “qualitative diversity” and how to access it, we often rely on null hypothesis testing—not to be taken literally, but to access a consistent direction of results across qualitatively diverse contexts.