Statistical Critiques of the Democratic Peace: Caveat Emptor


  • I am deeply grateful to Giacomo Chiozza, David Freedman, and Bruce Russett for their time, insight, and encouragement. Thanks also to Devin Caughey, Chris Chambers-Ju, Sara Chatfield, Joanna Dafoe, Ron Hassner, John Henderson, Charlotte Jandér, Nina Kelsey, Matthew Kroenig, Rengyee Lee, Michael Mousseau, John Oneal, Ben Oppenheim, Ivo Plšek, George Willcoxon, Rick Wilson for editorial suggestions, and the anonymous reviewers. All errors are my own. Replication files and links to data are available at Additional analyses are reported in this article's Supporting Information (SI).

Allan Dafoe is a Ph.D. candidate in the Travers Department of Political Science, 210 Barrows Hall, University of California, Berkeley, Berkeley, CA 94720 (


The “democratic peace”—the inference that democracies rarely fight each other—is one of the most important and empirically robust findings in international relations (IR). This article surveys the statistical challenges to the democratic peace and critically analyzes a prominent recent critique (Gartzke 2007). Gartzke's claim that capitalist dynamics explain away the democratic peace relies on results problematically driven by (1) the censoring from the sample of observations containing certain communist countries or occurring before 1966, (2) the inclusion of regional controls, and (3) a misspecification of temporal controls. Analysis of these issues contributes to broader methodological debates and reveals novel characteristics of the democratic peace. Gartzke and other critics have contributed valuably to the study of IR; however, the democratic peace remains one of the most robust empirical associations in IR.