Experimental designs in public policy evaluation: uses and abuses

The present essay discusses the strengths and, above all, limits of using experimental and quasi-experimental methods in evaluating public programs. It begins with a brief presentation classical design, its requirements related concepts such as internal, external, counterfactual validity. Next, it addresses modalities evaluation wich relax experiment. Two critical sections point out ethical, political, operational limitations designs program subsequently, political-institutional motivations for resilience this approach spite well-known recurrent robustness problems.