The Peace Research Institute in Oslo (PRIO) has just released the latest version of the Armed Conflict Location and Event Dataset (ACLED), which I blogged about last year here. The new peer-reviewed paper on this latest release is available here and you can watch ACLED’s presentation at the 2009 International Conference on Crisis Mapping (ICCM 2009) right here. The unit of analysis for ACLED is “an individual event that occurred at a given location.”
This new version has geo-referenced data for 50 unstable countries from 1997 through to 2010. The real breakthrough here is not just the scope of geographical coverage but more importantly how incredibly up to date the data is. I’m excited about this because it is rare that academic datasets can actually inform policy or operational response in a timely way. Academic datasets are generally outdated.
PRIO’s updated dataset codes the “actions of rebels, governments, and militias within unstable states, specifying the exact location and date of battle events, transfers of military control, headquarter establishment, civilian violence, and rioting.” As the authors note, the dataset’s “disaggregation of civil war and transnational violent events allow for research on local level factors and the dynamics of civil and communal conflict.”
Indeed, “micro-level datasets allow researchers to rigorously test sub-national hypotheses and to generate new causal arguments that cannot be studied with country-year or static conflict-zone data.” The authors identify four distinctive advantages of disaggregating local conflict event-data:
- Data can be aggregated to any desired level for analysis;
- The types of conflict events (e.g. battles or civilian violence) can be analyzed separately or in tandem;
- The actors within a conflict can be grouped or analyzed separately;
- The dynamics of national or regional war clusters can be addressed together.
The academic paper that discusses this new release of ACLED doesn’t go into much geospatial analysis but the dataset will no doubt catalyze many analytical studies in the near future. One preliminary finding, however, shows that using country-level data can lead to biased results when studying conflict dynamics. “The average percentage of area covered by civil war from the data sample is approximately 48%, but the average amount of territory with repeated fighting is considerably smaller at 15%. Further, most conflicts initially start out as very local phenomena.”