Feb 1, 2017 |
This one-day workshop at the World Bank on Wednesday, February 1, 2017 aimed to improve understanding of challenges to the transparency and reproducibility of social science research, and to teach the latest techniques for developing a reproducible workflow, focusing on version control and the GitHub platform.
The social science research community is increasingly emphasizing research transparency and reproducibility as essential components of high quality research, particularly research that is intended to inform program and policy decisions such as impact evaluations. There are currently several suggested best practices in research transparency, including: pre-registering studies (such as with ClinicalTrials.gov or the AEA registry), developing and publishing pre-analysis plans (PAPs), following standard reporting guidelines, publicly sharing data and underlying code, and other practices that enable not only reproducible research, but also reproducible results.
The World Bank is a leader in producing research on international development programs and policies and is making great strides to make the data underlying this research public through tools like the Microdata Catalog and its Impact Evaluation repository.
This workshop targeted World Bank researchers who produce academic-level research, policy-focused work, and flagship reports, as well as statisticians and data producers who produce published series or support the aforementioned researchers.
- Master understanding of challenges to reproducibility of social science research including: publication bias, researcher bias and motivated reasoning, reproducibility, and data privacy issues.
- Master and apply techniques for reproducible workflow, specifically, learning how to transparently manage study materials and data/code using version control with Git and Github.
Slides available online at https://github.com/BITSS/WorldBankFeb2017