| Mediation has been the most frequent type of third-party intervention in armed conflict since the end of the Cold War, and a substantial body of scholarly work asks whether and how mediation can be effective in bringing disputants to an agreement. Yet research on the subject has fallen short in that the effects of mediation are generally studied separately from its causes. This means that an apparent effect of mediation on conflict settlement may in fact be spurious and perhaps the result of certain states and international organizations sending mediators to particular types of conflicts.This dissertation uses formal, qualitative, and quantitative methods to help fill this gap in the literature. It makes two key contributions. First, I develop a theory of mediation that accounts for the strategic considerations of disputants and third parties during both the initiation and the implementation stages of mediation. I show how disputants and third parties anticipate the likely consequences of mediated negotiations when choosing to engage in such talks, and I derive hypotheses about the likely impact of this strategic process on the sample of mediation events we actually observe. Second, I introduce an original dataset on conflict management in post-1990 wars, which was compiled from roughly 15,000 newspaper and wire reports, and present an empirical analysis that takes account of just this selection effect and problems of causal identification more generally.The dissertation finds reason to be optimistic about mediation overall. Third-party mediation does help disputants reach a settlement, although not all types of mediators are equally effective. This finding also depends crucially on its proper causal identification: We recover a positive effect of mediation only if we account for the fact that the sample of observed mediation efforts is subject to strategic selection. |
