Statistics, Schmatistics!

Two Brookings scholars, Ken Pollack and Michael O’Hanlon, seem to have revived the battle over success/the lack thereof in Iraq. O’Hanlon and Pollack, both supporters of the Iraq invasion, took it to a whole new level last week following their return from a trip to Iraq in which they argued in a NY Times op-ed, “We are finally getting somewhere in Iraq.” By now various observers have pointed out that the O’Hanlon-Pollack commentary seemed to be all but entirely based on anecdotal observations and conversations with decidedly optimistic figures in Iraq. But let’s focus on Michael O’Hanlon. Michael O’Hanlon is the author/editor of the Iraq Index, a nearly real-time statistical roundup of the violent goings-on in Iraq and an indispensable resource to those who are trying to judge the progress of Bush’s war. What might be the most appalling part of this is that O’Hanlon seemed to be rendering judgment without even considering his own Iraq Index.

So how about we take a look at the Iraq Index ourselves?

First, I’ll start by admitting one small bit of progress. Multiple Fatality Bombings in Iraq have declined a bit in the last 3 months:

But if you look at the number killed in Multiple Fatality Bombings, things don’t look so good:

It’s important not to forget the numbers of wounded — not good either:

The number of daily attacks by insurgents and militias remained near record levels through April 2007 (no data after April):

The trend is clearly upward for American troop deaths too, not least because of the troop surge:

Our Combat Support Hospitals and other medical facilities save the lives of an amazingly high percentage of soldiers who come through their doors, but the number of seriously wounded is still appalling.

Also trending upward:

Now let’s take a look at Iraqi civilian deaths overall. According to the July 30 Iraq Index, deaths fell a bit from the record levels set in 2006. On the one hand, maybe there’s progress in that we’re not above 3500 anymore; on the other hand, the numbers fell in December, before the troop surge started. The numbers have remained fairly flat since the surge, even trending upward slightly:

But now look at O’Hanlon’s “corrected” figures from the current Iraq Index just 2 days later:

Compare the 2007 months on the 2 graphs.

Huh?

Now how did that happen? In the July 30 Iraq Index — prior to the data “correction” — O’Hanlon explained:

A more thorough accounting will follow in the coming days, but in short, civilian fatality levels in Iraq now seem to have declined substantially more than previous Pentagon reports or data had indicated. In particular, the monthly civilian fatality rate from sectarian violence appears about one-third lower than in the pre-surge months. That is still far too high, and remains comparable to violence levels of the 2004-2005 period, but it nonetheless reflects progress. [Emphasis added]**

Looking at the current Iraq Index, sometimes the data do get revised, yes. But 7 months after the fact? And how did the numbers for January and February get revised UPWARD by 1000 each month? O’Hanlon said the Pentagon underestimated the DECLINE in deaths post-surge, but he didn’t say anything about underestimating any INCREASE in deaths pre-surge. And I guess it’s just purely coincidence that we now see a long, steady decline in Iraqi civilian deaths during the troop surge months, apparently demonstrating that the surge is making real progress. How were they so wrong before, though? January was not exceptional according to other measures of violence. Also, at no point during 2006 did we ever see an increase in deaths from one month to the next at a level even approaching what we saw in the “corrected” December 2006-January 2007 data. The only times civilian deaths increased by 1000 or more in a single month were during 3 spikes in April and October 2004 (the two Battles of Fallujah) and August 2005 (the Sadr uprising), when average monthly death tolls were considerably lower than today.

To sum up, there is evidence that American troops are stopping civilian deaths from hitting new highs. That’s not a surprise when our troops are putting more of themselves in harm’s way, though. On the other hand, bombing deaths, injuries and overall attacks are still at near-record levels. Nothing that suggests there’s any containment going on. In the end, this doesn’t look like progress. More like “nongress.”

So why can’t Michael O’Hanlon do a presentation like this sometime?

**For the sake of clarity, we should define “pre-surge” and “post-surge”: Some would argue that the surge began in January, when Bush authorized these deployments. But the surge really got underway a couple of months later (US troop levels actually fell from 140,000 in December 2006 to the 132-135K range in January and February). Starting in March, deployment levels began to increase (142,000 in March and so on up to 160,000 today).

UPDATE: As of August 28, the Iraq Index Archive no longer contains archived editions for June and July (I referred readers to the July 30 Iraq Index for a comparison of O’Hanlon’s July vs. August data). You can still find civilian deaths data for 2007 on page 12 of the May 31 Iraq Index, though.