‘Replication games’ test the robustness of social-science studies

Key Points

• The aim is to assess the reproducibility and robustness of social science studies published in top journals.
• Evaluated 110 articles from leading economics and political science journals.
• Conducted computational reproducibility assessments.
• Performed robustness checks on statistically significant results.
• Over 85% of articles were found to be computationally reproducible.
• 72% of statistically significant results remained significant in robustness checks.
• About 25% of articles had non-trivial coding errors.

Abstract