A comment on Arechar et al.'s (2023)" Understanding and combatting misinformation across 16 countries on six continents"

In this commentary, we report the results of our computational reproduction of the main claims from Arechar et al. (2023), which examined psychological and social predictors of susceptibility to COVID-19 misinformation across 16 countries. Using the dataset and replication materials provided by the authors, we attempted to reproduce the study’s findings using Stata (based on the authors’ analysis script) and R (by translating and reconstructing the code). To assess computational reproducibility, we ran the original Stata code and independently reimplemented key analyses in R. While the Stata script allowed reproduction of several key statistical outputs, it lacked plotting code for most figures and required code edits for compatibility with Version 18. In contrast, our R implementation successfully reproduced Figures 2, 3, 4A-C, 5A-D, and 6 with visual and numerical results closely resembling those in the original publication. We also compared the pre-analysis plan and the article. We found inconsistencies in the specificity of the research questions, analysis scope, and terminology, making it challenging to assess fidelity to the preregistration. Overall, our results highlight that, while many of the reported statistical findings are robust to reanalysis, the study’s graphical reproducibility could be improved by providing the complete code for all figures, better documentation, and ensuring greater consistency across the replication packages.

I4R Discussion Paper Series No. 277, Institute for Replication (I4R)