Why are physicians so reluctant to sustain a type 2 error (pronouncing a sick person well) that they will tolerate many type 1 errors (labeling a healthy person sick)?

Look at the reward system. Punishments for a type 2 error are serious: guilt, embarrassment, and the threat of lawsuit and scandal. But a type 1 error is sometimes seen as sound clinical practice, indicating a healthy approach to medicine.

Type 1 errors also are likely to generate increased income and a stream of steady customers.

Rewards for research and publications, on the other hand, and punishments for failure to accomplish these, are commonly administered by universities. Plus, publication-heavy résumés usually will be received at other universities, whereas teaching credentials, harder to document and qualify, are much less transferable. Consequently, it’s rational for university teachers to concentrate on research, even to the detriment of teaching and at the expense of their students.

1) Spend $11 million for anti-pollution equipment to keep from poisoning fish in the river adjacent to the plant; or

2) Do nothing, in violation of the law, and assume a one in ten chance of being caught, with a resultant $1 million fine plus the necessity of buying the equipment.

Under this not-unrealistic set of choices, it requires no linear program to determine that XYZ Corporation can maximize its probabilities by flouting the law. Add the fact that XYZ’s president is probably rewarded (by creditors, stockholders, and so on) according to criteria totally unrelated to the number of fish poisoned, and his probable course of action becomes clear.

Back to the workplace, where we typically have crisp measures of performance for productivity and fuzzy ones for safety. Rewards are high for productivity (getting to keep our job) and low for safe performance (some incentive program we created). Thus we reward productivity to the exclusion of safety.(1)

(1) Kerr, S., “On the Folly of Reading A, While Hoping for B,” Academy of Management Journal.