Research scientists know that data alone do not lie - but data can mislead. That's especially true in clinical trials that rely on subjective endpoints to assess drug efficacy.
Psychiatry studies, for example, which often center around the clinician's assessment of disease progression or severity, have modest success rates. Just 24% of Phase II and 56% of Phase III psychiatric studies demonstrate a statistically significant drug-placebo separation. But is it true that so many medications, having moved so far through the development pathway, do not work, or is something else at play?