...from my associations with doctors in controlled trials I have learned that the better the statistician understands the doctor/patient relationship and the doctor's very real and unique ethical problem the better he can help to devise a trial that may be less than ideal experimentally but yet likely to be of some, and perhaps considerable, value to medicine. ...still more important, I have learned that though the statistician himself may never see a patient... nevertheless, he cannot sit in an armchair, remote and Olympian, comfortably divesting himself of all ethical responsibility. As a partner in a combined endeavor a full share of that responsibility will always lie with him... Sir Austin Bradford Hill, as quoted in Ethics and Statistics in Randomized Clinical Trials. Royall RM. Statistical Science 1991; 6(1): 52-88
You can find an earlier version of this page on my original website.