The uncounted dead: statist bias and civilian targeting in conflict data

Quantitative research on violence against civilians relies extensively on “off-the-shelf” data, such as the widely cited Uppsala Conflict Data Program's (UCDP) One-Sided Violence dataset. We show that, due to data collection and coding protocols that privilege government narratives of violence, such data often reproduces statist biases pervasive in the international system. These dynamics are particularly visible when civilian deaths result from airstrikes, shelling, and other forms of long-range bombardment. Such capabilities are disproportionately possessed by states, yet conservative coding practices, combined with government control over information and access restrictions, dictate that UCDP consistently codes civilian deaths at the hands of governments as “battle-related” or incidental rather than deliberate targeting. We analyze patterns in the UCDP data release, version 23.1, using evidence from Sri Lanka and Ethiopia to illustrate these patterns.