Research Transparency & Reproducibility in the Social Sciences
by UC Berkeley/Berkeley Initiative for Transparency in the Social Sciences (BITSS)
in partnership with the United Nations University-MERIT (UNU-MERIT)
With the many replication controversies (academic research misconducts) that happened over the last decade on social sciences, there has been a growing interest toward finding mechanisms as well creating institutions to strengthen integrity in social sciences research. For this aim, BITSS (Berkeley Initiative for Transparency in Social Sciences) was established in 2012 to enhance the quality of social science research and evidence used for policy-making by raising awareness and identifying strategies and fostering adoption of transparency methods.
LSE Africa Summit Workshop
The aim of this workshop is to train the next generation of social science researchers to tools and practices to enhance the transparency, reproducibility and openness of their research.
It intends to cover the following aspects: Conceptual and emerging issues in the practice and ethics of research; Overview of issues that make research unreliable; Theory and Implementation of Pre-analysis Plans; Transparent data management, data sharing & statistical analysis using STATA and R.
Session 1: Emerging Issues in the Practice of Empirical Social Science
This half-hour session will discuss conceptual issues in the practice and ethics of research, as well as recent case studies of misconduct. What is a reliable, a transparent or a reproducible research?
Session 2: Overview of Issues that make Research Unreliable
This 40 minute session will discuss a range of issues that can make research unreliable, such as false-positives, P-hacking, P-curve and power analysis, publication bias, and the files drawer problem.
Session 3: Theory and Implementation of Pre-analysis Plans
The third session (30 minutes) will offer a walk-through of a Pre-Analysis Plan, as well as examples of Pre-Analysis Plans written under two development programmes in Sierra Leona (CDD project by Casey et al.) and in Kenya (WASH project by Garett et al.).
Session 4: Transparent data management, statistical analysis, data sharing & data registration:
the TIER1 (Teaching Integrity in Empirical Research) protocol workflow
The final session will last about 1 hour and will aim to give an overview of concrete principles in research transparency, such as file structure, consistent variable labeling, code commenting, readme files, data de-identification and codebooks in STATA.
Added to these principles will be hands-on practical exercises with data and code using the TIER 2.0 protocol model, as well as a presentation of Open Science Framework, Github and dataverse for collaborative workflow management; STATA Markdoc and R Markdown for creating dynamic documents with STATA and R.