Predicting the Likelihood Of armed Conflict

A former British army officer once said, “History is littered with the Wars which everybody knew would never happen.”

To better understand the likelihood of conflict, Ohio State University political science Professor Bear Braumoeller, Ph.D., and doctoral student Austin Carson recently used Ohio Supercomputer Center (OSC) resources to employ several statistical techniques relatively new to the quantitative study of international politics. Specifically, the project analyzed a handful of very large-sized datasets regarding the statistical correlates of war (350,000+ observations, 10-20 variables per observation) using a series of estimation models in the statistical program R.

“The OSC computing resources were instrumental in helping us compute the level of statistical certainty of our inferences,” said Braumoeller. “The supercomputing protocol allowed us to make these inferences with a high level of accuracy; such calculations were crucial for inference.”

Braumoeller and Carson found that statistical literature on conflict studies has generated strong and consistent findings on the relationship of political irrelevance (large distance between two countries) and democratic regime type on conflict. However, they found that scant attention was paid to whether these factors directly influence the likelihood of disputes or indirectly by modifying the influence of other variables.

Braumoeller’s project determined that the literature to date had misunderstood the important theoretical question of how these variables influence peace and war. Greater distance between states and greater political liberalism make other war-related variables less influential and dramatically reduce the probability of conflict.

Project lead: Bear F. Braumoeller, The Ohio State University
Research title: Political irrelevance, democracy, and the limits of militarized conflict
Funding source: The Ohio State University