Democratic Peace Holds After Correcting Data and Imputation Errors

Why This Question Matters: The claim that democracies rarely fight one another—the "democratic peace"—is a cornerstone of research on the causes of war. Allan Dafoe, John R. Oneal, and Bruce Russett revisit recent critiques that claim alternative explanations or weakening effects, and they set out criteria for evaluating such challenges to a well-established empirical regularity.

How the Authors Reassess the Debate: The authors apply a structured set of evaluation criteria and perform detailed reanalyses of two recent papers that challenge the democratic peace. They focus on (1) Daniel Mousseau's 2013 claim that contract-intensive economies (proxied by life-insurance expenditures) explain the democratic peace and (2) Gartzke and Weisiger's 2013 claim that the democratic peace has declined as the share of democracies in the system rose. The methods include data correction, replication of original models, identification of coding and specification errors, and large-scale robustness checks.

What They Did (Methods):

• Diagnosed measurement and coding problems, including a miscoded dependent variable and a misleading dyadic democracy specification in Mousseau's analysis.
• Documented an excessive reliance on imputation in Mousseau's models, where more than 90% of the central independent-variable values were imputed.
• Corrected data errors and misspecifications in Gartzke & Weisiger's tests that failed to directly address the question they raised.
• Estimated 144 alternative model specifications that build on Mousseau's approach while controlling for life-insurance expenditures to assess robustness.

Key Findings:

• The supposed contraction of the democratic peace by contract-intensive-economy arguments does not withstand scrutiny: after correcting coding and imputation issues and running extensive specifications, the authors find substantial, robust support for the democratic peace even when controlling for life-insurance measures.
• Contrary to the claim that the democratic peace has weakened as democracy became more common, corrected tests and data show that the peacefulness of jointly democratic dyads actually increased as the proportion of democracies in the international system grew after 1816.

Policy and Research Implications: These reanalyses reinforce the democratic peace as a robust empirical pattern and illustrate the importance of careful coding, transparent imputation, and specification testing when challenging established findings. The paper provides a practical template—criteria and extensive robustness checks—for critically evaluating empirical challenges in international-relations research.