The gold standard study design for establishing whether a treatment actually works — participants are randomly assigned to treatment or control, and the results are compared.Randomization minimizes selection bias and allows causal inference — the gold standard distinction from observational studies, which can only show association. Double-blind RCTs (neither participants nor assessors know group assignment) further reduce measurement bias. Sample size must be adequate to detect the expected effect — underpowered RCTs produce unreliable positive and negative results. For most compounds on this site, the absence of an RCT is the most important single fact about the evidence record.