Home Statistical Dictionary Randomized Controlled Trial
Randomized Controlled Trial
A study design that randomly assigns participants to a treatment group and a control group, so that any systematic difference in outcomes between the groups can be causally attributed to the treatment rather than to pre-existing differences between participants.
In Plain English
A randomized controlled trial (RCT) is widely regarded as the gold standard for establishing that something actually causes an outcome, rather than merely being associated with it. The key move is randomization: by flipping a coin (conceptually) to decide who gets the treatment and who doesn't, you guarantee that, on average, the treatment and control groups start out balanced on every characteristic that might matter, known or unknown, measured or unmeasured. That balance is what lets you attribute any difference in outcomes afterward specifically to the treatment itself, rather than to some pre-existing difference between the kinds of people who happened to end up in each group, the fundamental problem that plagues purely observational comparisons.
Definition
A randomized controlled trial (RCT) randomly assigns study participants to two or more groups, most simply a treatment group receiving the intervention under study and a control group that does not (or receives a standard-of-care or placebo comparator), and compares outcomes between the groups. Random assignment is what gives RCTs their distinctive causal-inference strength, it ensures that treatment and control groups are, in expectation, balanced not only on measured confounding variables but also on unmeasured ones, a guarantee purely observational designs (case-control, cohort studies) cannot provide, since observational designs can never fully rule out confounding from variables the researcher failed to measure or even think to consider. RCTs are frequently combined with blinding, concealing group assignment from participants (single-blind) and/or from those assessing outcomes (double-blind), to further guard against placebo effects and assessment bias; despite being the strongest available design for causal inference, RCTs are not always feasible or ethical (some exposures cannot be randomly assigned, such as smoking or a harmful environmental exposure), which is precisely the gap that observational study designs like cohort and case-control studies are built to address.
Properties
- Randomization is what fundamentally distinguishes RCTs from observational designs like cohort and case-control studies, in expectation it balances every characteristic between groups, not just the ones a researcher thought to measure and control for, which is exactly why RCTs are considered the strongest available design for establishing causation rather than mere association.
- Intent-to-treat analysis, analyzing participants according to their originally randomized group assignment regardless of whether they actually completed or adhered to that assignment, is the standard analytical approach for RCTs, since it preserves the unbiased comparison that randomization was specifically designed to guarantee, deviating from it (analyzing only compliant participants) can reintroduce the very selection bias randomization was meant to eliminate.
- RCTs are not always feasible, ethical, or practical, exposures that cannot be ethically randomized (like smoking, environmental hazards, or many social and economic conditions) or interventions that are simply too costly or slow to test via a full RCT are exactly the situations where observational designs, cohort and case-control studies, remain the primary available tool for causal investigation, with all their added vulnerability to confounding.
At a Glance