Walking on solid ground: a replication study on Piso Firme’s impact

Abstract

About 3ie 3ie is an international grant-making NGO promoting evidence-informed development policies and programmes. We are the global leader in funding and producing high-quality evidence of what works, how, why and at what cost. We believe that better and policyrelevant evidence will make development more effective and improve people's lives. 3ie Replication Paper Series The 3ie Replication Paper Series is designed to be a publication and dissemination outlet for internal replication studies of development impact evaluations. Internal replication studies are those that reanalyse the data from an original paper in order to validate the results. The series seeks to publish replication studies with findings that reinforce an original paper, as well as those that challenge the results of an original paper. To be eligible for submission, a replication study needs to be of a paper in 3ie's online Impact Evaluation Repository and needs to include a pure replication. 3ie invites formal replies from the original authors. These are published on the 3ie website together with the replication study. The 3ie Replication Programme also includes grant-making windows to fund replication studies of papers identified on the candidate studies list. Requests for proposals are issued one to two times a year. The candidate studies list includes published studies that are considered influential, innovative or counterintuitive. The list is periodically updated based on 3ie staff input and outside suggestions. The aim of the 3ie Replication Programme is to improve the quality of evidence from development impact evaluations for use in policymaking and programme design.

Which tests not witch hunts: a diagnostic approach for conducting replication research

“…In the Basurto et al (2015) replication study, the researchers note that the development of the HHH2009 dataset used by Cattaneo et al (2009) included two different approaches to handle missing values. The first approach was to impute values using Dummy Adjustment Imputation, which is declared in the supporting documentation.…”