The Washington PostDemocracy Dies in Darkness

How should we count the war dead in Syria?

Photo by Manu Brabo (AP)

In February, John McCain denounced on the Senate floor what he called an “apocalyptic disaster” in Syria, citing “more than 130,000 people dead” there. The number he cited came from just one of the jumble of NGOs tabulating casualties in Syria.

Counting the dead during war has always been an inexact science. There are obvious methodological hurdles verifying the data in a conflict setting like Syria, with poor security and few impartial witnesses. There are also potential biases among the organizations doing the counting, which are frequently human rights groups loosely aligned with the opposition. Recent civil wars have occurred in places that have not seen a census in decades. And in situations where territory changes hands and citizens find it difficult to predict whether the government or rebels will ultimately prevail, many refuse to report casualties for fear of repercussions. Nor does it help that the U.S. government and UN have basically thrown in the towel on publicizing such data.

Micah Zenko and Amelia Wolf of the Council on Foreign Relations recently blogged that the actual number of people killed in Syria is vastly different from what is portrayed in the media. They suggest the level of civilian fatalities is in fact lower than what is being reported, and that pro-regime forces are dying in greater numbers than civilians. Their claims are not merely speculative, but based on data released by the Syrian Observatory on Human Rights (SOHR).

At first glance, the data indeed paint a puzzling depiction of violence. How can it be safer to be an unarmed civilian caught in the middle than a trained soldier behind fortified walls? After all, according to SOHR, the number of civilians killed (51,212) is lower than pro-regime forces, including both soldiers and militia members (57,511).

Herein lies the problem with counting the dead in Syria’s civil war: different methods to identifying casualties, different standards, and different agendas can lead to starkly contrasting interpretations of the violence on the ground.

Consider just the conflicting enumerations of fatalities in Syria. SOHR claims to be able to account for 99% of all violence in Syria and provides video or photographic evidence in 70% of its cases. No death gets recorded unless there is an accompanying name. Its 150,000-plus-casualty figure is the one most often cited by the international media and NGO community, even though the number could be well above 220,000, SOHR’s director Rami Abdel-Rahman told Lebanon’s Daily Star.

By contrast, the Violations Documentation Center in Syria (VDC) relies on a loose network of a few dozen reporters and activists located abroad and on the ground in Syria. Its data, which is periodically audited, is vastly different from the SOHR’s and presents a much more one-sided picture of the violence. VDC claims to be able to document 60% of the instances of violence comprising its data. (Moreover, in light of the escalating number of foreign extremist fighters being killed on the rebel side, it is reportedly going to be doing away with the culturally loaded term “martyr”).

A few patterns jump out: First, VDC numbers are much lower than SOHR’s. Given that the SOHR does not collaborate or share its documentation with the other groups, which it accuses of being financed by Gulf donors and therefore lacking credibility, it is difficult to verify its data or explain the discrepancy.

Second, the number of pro-regime fighters killed is far lower than SOHR’s estimates (57,511 in SOHR vs. 10,755 in VDC), while the tally of pro-rebel fighters is roughly the same (24,275 in SOHR vs. 20,365 in VDC). Finally, the number of civilians killed is more lopsided in VDC (57,690) than in SOHR (51,212), though probably more accurate, based on what we know about violence trends in Syria. But the SOHR’s figure excludes rebel fighters. If we add the 18,000 civilians unaccounted for but believed to be either in prison or killed, then the number soars to nearly 70,000 dead. Again, that is not including 24,275 rebel fighters, the 2,286 dissident fighters, and the 11,220 foreign fighters. Presumably, then, the total dead on the rebel side eclipses 100,000.

Zenko and Wolf also challenge an assumption that 90 percent of those killed in civil wars tend to be civilians. But this claim does not correspond to the empirical regularity that civilians are often much more targeted than combatants in civil wars. In the case of Syria, it is erroneous to suggest that pro-regime fighters are dying in greater numbers than civilians. Violence is not only targeting the frontlines. The use of car bombs, airstrikes, and barrel bombs often target civilian locations, and they fall disproportionately in pro-rebel strongholds near major population centers like Aleppo and Homs.

Again, based on qualitative evidence from the conflict, and from what we know of past civil wars, the distinction between a civilian and a rebel is often murky at best, which makes tabulating civilian casualties so controversial. Recall the confusion in counting the war dead during the height of the Iraq War. The common practice among policymakers was to rely on independent monitors like IraqBodyCount and compare its numbers with those of the Iraqi government and United Nations, which relied on figures from hospitals, morgues and municipalities. A controversial 2006 Lancet study that relied on cluster sampling estimated that over 600,000 Iraqis had been killed in postwar violence.

The U.S. government and UN still reportedly keep data on civilian casualties in Syria, presumably which is audited and verified regularly, but does not release it to the public or to the academic community. This means we are reliant on third-party sources like SOHR and VDC for our data and documentation. Oxford Research Group is reportedly working on a project to document and name every war-related casualty, not just in Syria but across the globe. Moreover, the Human Rights Data Analysis Group has shown in settings ranging from Guatemala to Kosovo that multiple systems estimation, a statistical technique that exploits overlap between multiple lists of fatalities, can generate more comprehensive estimates of total deaths than simply enumerating the known dead.

Zenko and Wolf’s larger point that we should be suspicious of the casualty reports, given the hostility between the groups collecting the information, is a valid one. The danger is that somebody who knows little about the complexities of Syria – say, an American lawmaker voting on whether to authorize the provision of aid or arms there – might presume based on the SOHR data that the violence in Syria is one-sided, but inflicted primarily against the regime. Yes, the number of pro-regime casualties is probably higher than what outside media report. But we need sanguine analysis of what existing data actually show us about the war, and to strive to advance credible and systematic ways of counting the dead in Syria, not just for academic purposes but also for the sake of transitional justice. If we cannot accurately establish the patterns of violence in Syria, we cannot hope to craft smart policies, whether to advance U.S. strategic interests in a narrow sense, or to stem the bloody tide of this wrenching civil war.

Laia Balcells is an assistant professor of political science at Duke University.  Lionel Beehner is a PhD Candidate in political science at Yale University. Jonah Schulhofer-Wohl is an assistant professor of politics at the University of Virginia.

Correction: “The original post misstated the basis of multiple systems estimation. Multiple systems estimation can be used whether data samples are random or non-random. Its key features are that it uses multiple lists and exploits the overlap between them to model the process by which deaths are included in the lists, therefore accounting for unreported deaths.”