Abstract
The intervention trial of greatest benefit to patients satisfies three objectives: validity (its results are true), generalizability (its results are widely applicable), and efficiency (the trial is affordable and resources are left over for patient care and for other health research). The first objective, validity, has become a nonnegotiable demand; hence the ascendancy of the randomized trial.The quest for validity requires protection from a false conclusion of efficacy when none exists (the Type I error, the risk of which is described by alpha, the P value). An important review published in the Journal two years ago1 dramatized the high . . .