Assessing Petraeus: Less than meets the eye

Move On’s over-the-top characterization of General Petraeus’ credibility notwithstanding, there is something very fishy about Petreaus’ claim that sectarian violence has markedly decreased. Petraeus’ own slides show a very minor drop in sectarian violence between February and August: from about 1900 sectarian deaths in February to about 1700 in August.

The real sign of progress comes in the graph on attacks against US forces. But that has more to do with the Anbar awakening (pre-surge) and Sadr replenishing his militia than any major advance against enemies of the central Iraqi government. For an independent corroboration of the more important civilian statistics, McClatchy reports, based on consultations with health ministry officials, military sources and local sources throughout Iraq, 1,011 civilians killed in August, compared with 984 in February, the first month of the surge.

In other words, no reduction at all. But an anonymous official at the Interior Ministry says the number was much higher, going from 1,980 civilian dead in July to 2,890 in August. Petraeus’ numbers show a slight decline (morbidly explained by many officials by the extirpation of the Sunni population from Baghdad), official Interior Ministry numbers show an almost steady rate between February and August. And the unofficial report, which likely includes a broader characterization of “sectarian” murders shows a marked increase (half of which is probably accounted for in the Yezidi massacre).

So what’s happening in Iraq and is there a genuine increase in civilian security? Unidentified sectarian deaths in Baghdad have, indeed, dropped some since February. Unidentified bodies found dead in Baghdad dropped from 596 in February to 428 in August. The February number was actually a major dropoff from December 2006, which saw 1,030 bodies found across the capital city. But the overall picture shows civilian deaths either on the rise or, if slightly declining, explained more by displacement than by increased security.

In a follow-up article today on Petreaus’ testimony, McClatchy points out some of the holes in the success narrative. First and foremost, Petraeus’ maps of Baghdad fail to take into account the sectarian cleansing going on in the last six months. As various sources have pointed out, including Newsweek and the New York Times, Shi’ite militias have taken advantage of the surge to wipe out Sunni neighborhoods in Baghdad, driving hundreds of thousands into exile in neighboring countries or more homogeneously Sunni areas of the country. Baghdad went from a 65 percent Sunni city as recently as December 2006 to a 75 percent Shi’ite city.

Take a look at the green-blue-brown background color in Slide #5. It’s the same for the whole period, even though the make-up of those neighborhoods has radically changed.

The increased calm in Ramadi and Fallujah has been offset by large-scale militia terrorism in Baghdad. Why would Petraeus use the same sectarian color-coding for Baghdad in August as he does for February, knowing full well that the sectarian makeup across the capital city has changed dramatically? Had he accurately portrayed the changes in neighborhood make-up, Congress would have seen more graphically the costs of displacement. That’s no sign of increased security.

The most obvious explanation for discrepancies in the amount of violence is the differing definitions used for ethno/sectarian violence. The crude measure of “bullet in back of the head equals sectarian vs. bullet in the front equals criminal” has already gotten widespread attention. Petraeus’ own slide lists “ethno-sectarian” violence, leaving it up in the air how they define ethno-sectarian. A Heritage Foundation memo issued by Kirk Johnson, the statistician for Ryan Crocker, notes that sectarian violence misses“between 35 percent and 65 percent of civilian casualties on a month-to-month basis.” Johnson claims that overall violence has, in fact dropped around 20 percent (though he doesn’t provide raw numbers.

In the second McClatchy article Johnson is quoted as saying that violence has fallen back to May 2006 levels). But the larger admission is that violence takes on so many forms other than Shi’ite vs. Sunni. Perhaps even more important, Johnson notes that the military reports should be seen as a “lower-bound estimate because they tend to miss smaller incidents.”

Alas, this may explain the differing numbers in the end.

I don’t believe General Petraeus is a “bald faced liar.” But I have no doubt that Petraeus, like any general, is using statistics in a way to make his case as best he can.

His numbers, according to Crocker’s own statistician, miss much of the violence because of the reporting mechanism used by the military. He shows a decrease (albeit slight) in sectarian violence because he can identify statistics to prove that assertion.

But there are good reasons to believe that those statistics hide more than they reveal about the true extent of violence and overall despair in Iraq after the surge began.

The end result is that violence has probably dissipated in some parts of the country – notably Anbar – and in selected Baghdad neighborhoods. It has almost certainly increased in other neighborhoods untouched by the troop build-up, and in the North. Perhaps most sadly, even where the violence has, in fact dropped, it may be more a factor of either successful homogenization by Shi’ite militiamen (no more Sunnis left to kill) or the complete imprisonment of the Sunni population behind gigantic walls (cited in the McClatchy article as psychologically catastrophic) and not any genuine sectarian reconciliation or drawdown in militia activity. It’s no wonder Sunnis responded to a recent poll saying that security has gotten much worse since the surge began.

In the fundamental area of civilian security the surge has failed. Displaced civilians cannot return home. And civilians trapped inside their neighborhoods cannot work or visit family elsewhere in the capital.

Let’s not be fooled. Petraeus’ slides do nothing to counter this gloomy portrait.