Human rights monitoring consists primarily of receiving information from witnesses and conducting investigations. The resulting information is often stored in databases. However, the statistics generated from databases collected in this way may tell us more about the functioning of the organization doing the monitoring than about the violence being monitored.
Using examples from Guatemala, El Salvador, Kosovo, Colombia, Timor-Leste, and Sierra Leone, Patrick Ball, vice president of the Benetech Human Rights Data Analysis Group, explains how statistics derived from data collected by direct observation can be inconsistent or misleading representations of true patterns of violence. He concludes with examples showing how reliable statistics can be produced using multiple independent databases (and a lot of math) or using a random sample of respondents.
Darius Cuplinskas, director of the Open Society Information Program, moderates the event. Open Society Foundations president Aryeh Neier participates as a respondent.