So-called randomized controlled trials are directly inspired by the clinical trials conducted to measure the efficiency of a drug. The method consists in randomly distributing the individuals involved into two groups, in order to measure the causal effect of the intervention studied:
– a “treatment” group, which benefits from the scheme; and
– a “control” group, which does not benefit from the trial.
The interest of this approach for understanding causal relationships lies in the fact that the random allocation of individuals to the treatment and control groups guarantees that the two groups share on average the same characteristics. The control group can hence be used as counterfactual for predicting what the results of the “treated” individuals would have been had they not benefited from the intervention.
Randomized controlled trials: a decade of growth
In randomized controlled trials, a group of individuals is randomly directed towards the intervention being evaluated, whereas another group is denied access to it. The random allocation mechanism raises ethical issues, which ought to be discussed prior to the trial. In addition, the observation mechanism is generally specific to the experiment and should be set up along with the evaluation measure. Under those conditions, controlled experiments constitute the most reliable method for evaluating the effects of a public policy. It is advisable to resort to them when these conditions can be met.
The Poverty Action Lab (J-PAL), founded in 2003 at MIT by Abhijit Banerjee, Esther Duflo and Sendhil Mullainathan, bases its reputation on the exclusive use of randomized controlled trials for measuring the effects of social and anti-poverty programmes.
J-PAL Europe, based at the PSE – Paris School of Economics, brings together European researchers working on various projects, to evaluate public policies in France, Europe and other parts of the world. When setting up randomized controlled trials, researchers at IPP may call on the expertise of its partner J-PAL Europe.