Good studies are "controlled." In a controlled study some participants are given the treatment and others—the "control" group—are given either a placebo or another treatment, or, less optimally, left untreated altogether.

"Uncontrolled" studies, in which everyone is given the treatment, are highly unreliable for many reasons. One problem is that it is impossible to know how well participants would have done without any treatment at all. Additionally, individuals who know they are being treated may naturally begin to pay more attention to their disease and take better care of themselves.