Get access

Which Side Are You On? Bias, Credibility, and Mediation


  • Earlier versions of this article were presented at the Research Frontiers in International Relations Conference, La Jolla, CA, at the PIPES Workshop, University of Chicago, at the 2000 meetings of the Midwest Political Science Association, Chicago, at the University of California, Davis, at the DVPW in Halle, Germany, and at the Olin Institute National Security Seminar Series, and I thank the participants for their comments. I would also like to thank Shaun Bowler for the conversation that started the whole project, and Jasen Castillo, Jörg Faust, Page Fortna, Erik Gartzke, Barbara Koremenos, Lisa Martin, Robert Powell, Holger Schmidt, Naunihal Singh, Monica Toft, Benjamin Valentino, and Barbara Walter for helpful feedback along the way.

Andrew Kydd is Assistant Professor of Government, Harvard University, 1033 Massachusetts Ave. 318B, Cambridge, MA 02138 (


Mediators are often thought to be more effective if they are unbiased or have no preferences over the issue in dispute. This article presents a game theoretic model of mediation drawing on the theory of “cheap talk” which highlights a contrary logic. Conflict arises in bargaining games because of uncertainty about the resolve of the parties. A mediator can reduce the likelihood of conflict by providing information on this score. For a mediator to be effective, however, the parties must believe that the mediator is telling the truth, especially if the mediator counsels one side to make a concession because their opponent has high resolve and will fight. An unbiased mediator who is simply interested in minimizing the probability of conflict will have a strong incentive to make such statements even if they are not true, hence the parties will not find the mediator credible. Only mediators who are effectively “on your side” will be believed if they counsel restraint.