Libby Abbott/J-PAL

The outcomes of evidence-based policy making are only as good as the evidence used. Impact evaluation methods, which are scientifically rigorous, provide highly useful evidence for policy making. But as with all scientific research, the only way to validate the findings of impact evaluation is to make sure that if we conduct the analysis over again, we yield the same results - that is, to replicate them. A well-known example of replication is the Roodman and Morduch replication of Pitt and Khandker on microfinance. But it is to be hoped and expected that most replications confirm the validity of study findings, which is very important for increasing the credibility of evidence for policy making.

3ie is pleased to announce its Replication Program, which aims to strengthen, and sometimes question, the credibility of evidence produced by the most influential, innovative, and controversial impact evaluations.

The 3ie Replication Program will fund internal replications of important impact evaluations, where internal replication means the reanalysis of the original data. The reanalysis seeks to validate the precise findings of the study and can often extend to testing alternate specifications to rule out competing theories of change or using alternate datasets from the same population to confirm the robustness of the findings.