Digital Echoes - Understanding Mass Violence with Data and Statistics

Explore the complexities of analyzing mass violence data in this keynote address by Patrick Ball, Director of Research at the Human Rights Data Analysis Group. Delve into the challenges of interpreting statistical patterns in human rights data collection, examining how biases in raw data can skew our understanding of real-world phenomena. Learn about innovative estimation techniques used to correct for heterogeneous underreporting through case studies spanning Chadian prisons, Iraq, US police homicides, the Syrian conflict, and the Guatemalan genocide. Gain insights into the application of these analytical methods in public debates and criminal trials for genocide and war crimes, while considering the ethical implications and potential pitfalls of using machine learning models in predictive policing and drug crime analysis.

Intro
Prisoners of War
Mortality
Hidden Graves
Predictive Policing
Drug Crimes
Bias
Feedback Loop
Why are ML models wrong
Cost of being wrong
Edgar Garcia case