Randomized Controlled Trials
Randomized controlled trials are the gold standard in impact evaluations. The “treatment” is administered randomly to participants, without regard to how it might affect the outcome of interest for each. For large samples with random assignment, the characteristics of the treatment and control groups will be balanced, i.e., the treatment and control groups will be virtually the same. As such, the difference in outcomes for the treatment and control groups provide an unbiased estimate of the treatment effect.
How it works
RCTs generally require an extensive experimental design and implementation strategy. Given a successful experiment (in which the randomization is effective, and take-up is perfect, and there are no treatment spillovers to the control group, etc.) impact evaluations are straightforward. They are estimated by the simple difference in outcomes between the treatment and control group. For cases in which the experiment was not implemented perfectly (which is a common occurrence), a host of quasi-experimental methodologies can be used to overcome data issues and estimate the treatment effect.
When to use
- Whenever it is financially feasible and logistically possible
- Example dataset:
- A dataset containing health outcomes for individuals, where medical treatment is assigned randomly to patients
- How to measure the treatment effect:
- The simple difference in health outcomes between patients who received the treatment and those who did not
Further explanation on Randomized Controlled Trials
