Aaron's comment about the placebo effect got me thinking. This question shows my ignorance, but do all clinical trials that use a placebo have some other kind of control? They better, right? If not, how do you tell the difference between a true placebo effect (where the mechanism is the expectations of the subject) and a person getting better (off the drug being tested) for some other kind of reason (such as time passing)? Or are all trials that use a placebo also designed so as to have a "no treatment" control, such that there would be three experimental treatments: the drug to be tested, the placebo, and no treatment (that is, neither drug nor placebo)?
See, this is what happens when you have the training of a philosopher but to try to think about scientific studies.
Nevertheless, in conversation, people have often confused a placebo effect where the mechanism is the expectations of the subject with any improvement in a subject's condition that has not caused by the drug to be tested.