In 2014, Innovations for Poverty Action (IPA) launched the research transparency initiative to promote sharing data and code, as well as to register research studies. We believe that high-quality research and transparency go hand in hand.

With our initiative, we are joining the growing movement which includes funders, journals, and other research groups. We are collaborating with other groups including Berkeley Initiative for Transparency in the Social Sciences (BITSS), Institution for Social and Policy Studies at Yale University (ISPS), and the Abdul Latif Jameel Poverty Action Lab (J-PAL).

As a part of our initiative, IPA has adopted organization-wide guidelines and services for projects that we implement and/or fund:

Registering studies with the AEA Registry.

The purpose of registration is to provide a public record of all studies, which helps to combat publication bias (i.e., the skew to seeing positive rather than null results). All of IPA’s randomized trials should be registered on the AEA site. The required AEA trial registration is pretty basic, but one can optionally include a detailed pre-analysis plan.

Sharing data and code. 

We have established a data repository through Dataverse, in order to provide a stable location for publicly sharing data from IPA studies.

Our goals:

  • Providing guidance for IPA-affiliated researchers on data-sharing:
    We have adopted data publication guidelines outlining our recommendations and requirements. Our aim is to share data from IPA projects with a timeframe of 3 years following endline data collection or first publication (whichever comes first).
  • Centralizing access to important datasets, for the purposes of re-analysis and re-use.
    In addition to sharing data from IPA studies, we provide central access to data from randomized controlled trials in the social sciences by linking data from J-PAL and other groups into the repository.
  • Curating data to make it useable:
    For our collaborating researchers, we offer IPA research support for:
  1. Data curation: For all studies that we upload to our repository, we check the data to make sure variables are labeled, ensure that the code runs and produces tables/figures, confirm that proper documentation and metadata are included, and double-check that no personally identifiable information (PII) is released.
  2. Code checks: We offer to run the code and check to see that it reproduces the tables in the article before publication. This service is particularly for IPA researchers who are submitting their data and code in accordance with a journal data-sharing requirement. (See for example AER’s requirement).

Resources & training in reproducible research

Blog posts about research transparency: