these examples, by having the observer report the value of the decision variable and letting a decision maker compare it with a positivity criterion for a binary decision.
The concept of the optimal positivity criterion has not permeated several important diagnostic fields and is effectively implemented in only a few of them. Within psychology, it has been applied to aptitude testing for personnel selection ( Cronbach & Gleser, 1965). It has made practical inroads in weather forecasting for commercial interests ( Miller & Thompson, 1975), and there may be a few other examples of routine and appropriate use. However, the concept has not enlightened HIV testing or materials testing of aircraft to an appropriate extent, despite the high national visibility and priority of these diagnostic problems. In HIV testing, we usually know quite precisely where the criterion is set, but are not assured that it is in the best place; in materials testing, we have almost no idea where a given inspector will set a criterion. This is not the place to attempt to detail the cost of inappropriate or unknown criterion settings to individuals and society; we can easily surmise that it is large.
Why does this situation exist? The optimality concept is not difficult, and it is not esoteric; doctors use it deliberately, if not quantitatively, when they make decisions about therapy, and airline and aircraft-manufacturing executives do so as well in their decisions. In fact, the idea is intuitively familiar to almost everyone; acting on the basis of probabilities and benefits and costs is something people do in their everyday lives. To be sure, the concept is mathematical as it would be used effectively in diagnosis, but as shown earlier, to first order it is captured neatly by a single graph and a simple equation. As mentioned earlier, it is often very difficult to estimate the prior probabilities of diagnostic alternatives and, especially, the benefits and costs of decision outcomes, and individuals may start out, at least, with widely varying estimates. However, the question that remains is why, with the computational tools available, these variables are treated only tacitly or ignored.
Can matters be improved? It is apparent that communications via conferences, journal articles, and books, although adequate to give the optimality concept scientific vetting, are not sufficient to influence currently serious practical arenas. I believe that the idea will be applied broadly in such arenas only when increasingly aware psychologists--whether primarily versed or interested in decision analysis, diagnosis, social structures, or societal problems--find ways to "give [this] psychology away." Where they have access, they might assist educational or advisory organizations in taking an initiative, or help national societies and government agencies in scientific, engineering, and health fields come officially to appreciate what is at stake. My personal impression is that most regulatory and certifying agencies are not directly approachable from outside the gov