Controlled trial
Testing a treatment by comparing groups that differ in one thing only, ideally with neither patients nor researchers knowing who got what.
To find out whether a treatment works, you have to compare. A controlled trial splits people into groups that are as alike as possible and treated the same way in every respect except one: the treatment being tested. If the groups then fare differently, that single difference is the likely cause. Without such a comparison it is easy to credit a treatment for a recovery that would have happened anyway.
An early example came in 1747, when the naval surgeon James Lind gave different remedies to sailors with scurvy who were living on the same ship and diet. Those given oranges and lemons recovered fastest.
Expectations are a hidden trap. The observer-expectancy effect is the way a researcher's beliefs can unconsciously steer results. The classic case is Clever Hans, a horse in Germany in the early 1900s that seemed to answer arithmetic questions by tapping a hoof. The psychologist Oskar Pfungst found that Hans answered correctly 89 per cent of the time when his questioner knew the answer, but only 6 per cent of the time when they did not. The horse was reading tiny, unconscious changes in the questioner's posture and face.
Modern trials guard against such effects. Patients are assigned to groups at random, one group often receives a placebo, a dummy treatment, and in a double-blind trial neither the patients nor the researchers assessing them know who is in which group until the end.
Coming up in future stories
- The Horse That Could Do Maths (Almost) 1904 · Berlin, Germany
- The Letter That Gave MSG Its Bad Name 4 Apr 1968 · New England Journal of Medicine, USA