With the Pentagon and the White House working to justify a continuing presence in Iraq based, in part, on claims that the surge has made great strides in reducing sectarian killings, the accuracy of such assessments becomes critically important.
Against this backdrop, it is troubling that a review of the Pentagon's quarterly reports indicates that the military's count of sectarian violence has varied widely, evidencing ex post facto revisions to the official statistics.
Brian Katulis, senior fellow at the Center for American Progress, tells me that it's too early to draw any conclusions, "but it is clear that there are some discrepancies in the figures that the Pentagon has provided during the past year on the levels of violence in Iraq."
The discrepancies were first raised by Katulis and reported by Spencer Ackerman at TPMmuckraker. The above graphic was put together by Ilan Goldberg of the National Security Network to demonstrate the variance in the Pentagon's official graphing of sectarian killings between the August 2006, November 2006/March 2007, and June 2007 reports.
The overlay shows that the level of sectarian killings reported between January and July 2006 for the August 2006 report was revised significantly downward for the November 2006 and March 2007 reports. Then the June 2007 report re-revises the earlier period to show an even lower estimate for the first half of 2006 and a higher one for all the time since.
Goldberg explains the abnormalities as best he can:
Abnormality A: Between August and November 2006, DOD started reclassifying “casualties” as “deaths by execution” and suddenly you see a dramatic drop in killings. For example, in March 2006 right after the Samarra Mosque bombings you go from 1,750 “casualties” to 750 “deaths by execution.” Between November 2006 and March 2007 “Deaths by Execution” becomes “Sectarian Murders” but the numbers remain the same.
Abnormality B: Between the March 2007 report and the June 2007 report there was a dramatic change in the number of killings that were reported for the second half of 2006. In both cases the numbers were described as “sectarian murders.” The impact here is that it makes the “pre surge” situation look extraordinarily dire and therefore signals progress thereafter.
Abnormality C: Somehow the reclassification that occurred between the March and June 2007 reports caused the violence numbers in April and May of 2006 to drop dramatically. This was in the months following the Sammara bombings in February 2006 when sectarian violence was escalating.
In a recent estimation, Gen. David Petraeus's told the Australian in an interview this week that there had been a 75% reduction in "religious and ethnic killings" since last year. Petraeus was citing the statistic to support the view that the surge is making progress, but it is unclear what kinds of deaths would be classified specifically as "religious and ethnic killings." For example, the recent attack against Yazidis in Sinjar, which killed an estimated 350 people, could be logically classified as a "religious and ethnic" killing, but it's clearly not.
Katulis argues that the Pentagon needs greater transparency to explain the way it classifies sectarian violence, though also emphasizes that any decline in such killings may not provide a valuable metric for evaluating progress of the surge. Attempts at such neat categorization, while useful for qualitative analysis, may oversimplify the dynamics of Iraq's instability--not only in overlooking nuance for simplified classification, but also in attributing any decline to progress of the surge.
As Slogger reported yesterday, confusion over the way the military is collecting and processing statistics of sectarian violence has sparked a call for greater transparency in methodology.
Phone calls to the Pentagon were unable to track down anyone with a knowledge to explain the methodology used to calculate sectarian violence, but a public affairs officer should be looking into it and getting back to me. Stay tuned...