Guest post by Soazic Elise Wang Sonne (World Bank)
Despite efforts by African governments to significantly raise public spending on scientific research, the continent, which is home to 14% of the world’s population, contributes to less than 1% of published research outputs (David Dunne, 2017). While this can be partly attributed to insufficient financial and human resources for research, it might also be explained by a lack of perceived integrity in African researchers’ manuscripts, which can impede publication in top peer-reviewed journals. Africa cannot afford to lag behind in research quality and can get ahead by taking advantage of the numerous benefits offered by adopting research openness and reproducibility practices.
My initial engagement with the Berkeley Initiative for Transparency in the Social Sciences (BITSS) began in early 2015 when I attended a Research Transparency and Reproducibility Training (RT2). I became a BITSS Catalyst to help change research practices in Europe and Sub-Saharan Africa, and I spent this past year as a BITSS visiting scholar at UC Berkeley. This experience has been most rewarding. I had the opportunity to build my technical skills by attending and leading workshops at meetings of the Working Group for African Political Economy and at BITSS Annual Meetings. I believe training the next generation of African social scientists on research transparency and reproducibility is an important way to improve research quality and address low publication rates in Sub-Saharan Africa. To date, with the generous support of BITSS and other open science organizations (such as SPARC, OpenCon, and Open Knowledge International), I have trained more than 150 African researchers via workshops organized in South Africa, Cameroon, the UK, the US, and the Netherlands.
Participants at my trainings have been keen to learn about scientific misconduct, likely because they recognize the common existence of the “File Drawer Problem,” p-hacking, and selective reporting in both their own and their colleagues’ research routines. In addition, participants have been very excited to learn about solutions to these problems, including the key features of a reproducible workflow (such as the TIER protocol) as well as pre-analysis plans, master do-files, and dynamic documents with R and Stata.
At these trainings, participants reflect on how research transparency norms and practices can be better entrenched in African academic higher institutions. The following are key insights that have emerged from our discussions:
- Transparency is important and deserves to be fully integrated as a distinct module in quantitative tracks of undergraduate research curriculum. This is broadly expressed by workshop participants from Sub-Saharan Africa.
- Transparency is difficult to implement given the predominance of the “hidden” culture in Africa’s academia, wherein researchers are not encouraged to share data, code, or other study materials for fear of being scooped or plagiarized.
- There is a need to use open source software such as R, rather than Stata. This is because most African students working toward quantitative social science degrees use an abbreviated version of Stata, which is limited and therefore offers less room to build a reproducible workflow.
- Last but not least, most of the participants think that a top-down approach is needed to sensitize more senior researchers to transparent and reproducible science. In fact, the majority still use deeply ingrained and outdated research practices without a clear understanding of the benefits of openness and reproducibility. Students and junior researchers have little room to try new things or advise on how research practices should be done. Therefore, training senior researchers on why transparency matters and how to practice it could catalyze the institutionalization of reproducibility.
Later this month, I am heading to Abdou Moumouni University in Niamey, Niger to train young social scientists on research transparency in impact evaluation, in partnership with the UC Berkeley OASIS initiative. This training will not only focus on teaching reproducibility methods to students, but also improving their writing skills. I expect this training will help improve participants’ research practices, as well as support OASIS in preparation of the second phase of their “Room to Grow” impact evaluation — part of their work to empower adolescent girls in Niamey.
Building the capacity of African researchers to conduct reproducible and transparent research will help ensure findings used by policy makers to make key decisions are based on credible evidence. It will also strengthen trust and partnerships between policy makers and the academic community.