Predicting War Just Got Better by Thinking About Who Has Power

The Problem With Traditional Power Measures

Military power rankings used in international relations are surprisingly ineffective at predicting conflict outcomes. The standard capability ratio is no better than random guesses.

Introducing DOE Scores: A Better Proxy for Conflict Prediction

Using machine learning, researchers developed the Dispute Outcome Expectations score as a superior proxy using the same data sources.

A Surprising Finding About Power Dynamics

Contrary to previous assumptions, conflict probability peaks when states with fewer benefits hold power preponderance. This new DOE approach improves prediction accuracy across numerous dyadic studies.