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

A large-scale assessment of the reproducibility and robustness of 110 articles published in leading economics and political-science journals finds that more than 85% were computationally reproducible, and 72% of statistically significant results remained significant and in the same direction in robustness checks. About 25% of the evaluated articles contained non‑trivial coding errors. Work to replicate studies in economics and political science highlights the importance of sharing data and code.