Weekly links Friday the 13, reproducibility edition

• This week DIME Analytics hosted an event on making research reproducible joint with 3ie, BITSS (CEGA), IPA, and J-PAL. Ted Miguel’s awesome keynote brought this new book to my attention (I even won one, thanks Ted and Garret!): “Transparent and Reproducible Social Science Research” by Garret Christensen, Jeremy Freese, and Ted Miguel (Pub. University of California Press). While the book exposes the problem of non-reproducible research, it focuses on the solutions. This is a great read for anyone wants to understand why registering trials and metaänalyis are good for science, difference between pre-registration and pre-analysis, reporting standards and, last but not least, why and how to create reproducible workflows.
• For more tools of the trade, our DIME Analytics also just put out a new book, Data for Development Impact, and they want your feedback!
• Since we are talking pre-registration and pre-analysis, let’s mention a new guide to power calcs for panel data: Burlig, Preonas and Woerman present the “serial-correlation-robust” power calculations, that extends previous versions by allowing for non-constant serial correlation. They show that ex-post standard error adjustments (clustering) are not the solution to get properly powered experiments – you need to get your power calcs right. They also release a Stata command to help you do it (pcpanel).
• A newly published J-PAL blog argues that RCTs can help reduce violence and conflict, featuring tons of fabulous papers across a wide array of interventions.
• “How many dollar bills do we leave on the sidewalk from inefficiently low internal migration?” ICYMI, watch Leah Boustan’s great keynote at the 6th Urbanization And Poverty Reduction Conference, “Moving to Opportunity in the Developing World”, in which she provides insights from her work and the literature at large, pulling together evidence from a wide range of contexts and methods.
• IPA and J-PAL have just announced the creation of a joint dataverse: Datahub for Field Experiments in Economics and Public Policy, thus making available over 140 datasets. If you are looking for data from impact evaluation studies, you can also consult the World Bank’s micro data catalogue for impact evaluation where 149 curated datasets are currently posted.
• Some additional reproducibility links from DIME Analytics:
  • A Ted talk (haha) on reproducibility at NBER: 
  • Seven Reasons Why: A User’s Guide to Transparency and Reproducibility 
  • AEA checklist 
  • IPA guidelines and JPAL’s guidelines for Data Publication
  • 3ie research transparency policy
  • IPA resources on research transparency

And, that’s all folks!