McNeil, B. J., Pauker, S.G., Sox, H.C., & Tversky, A. On the elicitation of preferences for alternative therapies. New England Journal of Medicine, 1982, 306, 1259-1262.

---

They asked various groups (patients, students, and physicians) to choose between two lung cancer therapies. They varied the following factors:
* whether the treatments were identified or not
* data framed in life expectancy vs cumulative probability
* probability of dying vs living
* population of "choosers"

According to efficacy data, surgery offers better long-term survival odds than radiation therapy for lung cancer.

Results
People who received life-expectancy data chose radiation therapy less frequently than those who received cumulative probability data.

Radiation was chosen more often when it was not identified. Identification favored surgery.

Surgery was less attractive in the mortality fram and the survival frame. This was true for all subjects (even those with statistical training).

Radiation therapy was least popular with students, more so with patients, and highest with physicians.

Note that the physicians were subject to the same data perception bias as the patients and students.