Beginning in 2017, we implemented a new statistical model. The new model increases the forecasting window to two years, includes both state- and nonstate-led mass killing, and revises some of the data inputs to benefit from newly available sources.
For these reasons, risk and rank statistics from 2014-2016 should not be compared to results from 2017 forward.
From 2014-16, we used three models representing different ideas about the origins of mass atrocities to create the Statistical Risk Assessment. Those assessments were an average of forecasts from those three models.
Key Risk Factors
All Risk Factors
Our statistical model includes the following variables, sourced from publicly available data on a variety of country characteristics. See more details in our data dictionary (PDF)
from the Early Warning Project and the Simon-Skjodt Center for the Prevention of Genocide