No News Is Good News: Mark and Recapture for Event Data When Reporting Probabilities Are Less Than One

We discuss a common, but often ignored, problem in event data: underreporting bias. When collecting data, it is often not the case that source materials capture all events of interest, leading to an undercount of the true number of events. To address this issue, we propose a common method first used to estimate the size of animal populations when a complete census is not feasible: mark and recapture. By taking multiple sources into consideration, one can estimate the rate of missing data across sources and come up with an estimate of the true number of events. To demonstrate the utility of the approach, we compare Associated Press and Agence France Press reports on conflict events, as contained in the Social Conflict in Africa Database. We show that these sources capture approximately 76% of all events in Africa but that the nondetection rate declines dramatically when considering more significant events. We also show through regression analysis that deadly events, events of a larger magnitude, and events with government repression, among others, are significant predictors of overlapping reporting. Ultimately, the approach can be used to correct for undercounting in event data and to assess the quality of sources used.