Kevin Drum on Casualties in Iraq

Reader Nate is trying to engineer a cross-blog debate between myself and Kevin Drum. Drum will probably never even see what I have to say, but far be it from me to refuse this challenge anyway.

Kevin Drum is thinking through casualties in Iraq, but he is starting off with the wrong attitude (at least that's how it seems to me). The attitude that comes through is not that of a truth seeker. Instead, his apparent attitude can be summed up like this: "no one is going to fool me by saying that things have improved in Iraq, no siree." Still, I am glad that someone on the left is actually starting to think about casualty details. The story of Iraq is contained in those details, but almost no one on the left knows it. Perhaps that will change if Drum continues along the path he is on (and if he adjusts his attitude a bit).

A few days ago, Drum presented casualty figures provided by the LA Times that I discussed here. About those figures, he said:

The chart above shows war-related violent deaths for the entire year of 2007 as compiled by the Times. What's remarkable is that not only does it not show any decrease since the beginning of the surge in February, but it doesn't even show a significant dip during summer, traditionally the quietest season in Iraq.

Actually, the numbers do show an average decrease of about 260 deaths per month during the 3 months of the troop surge compared to the previous 5 months. Drum seems to think that this might simply reflect a seasonal decrease in casualties that occurs in the summer months. The truly suspicious mind refuses to be convinced that matters have improved -- even a little -- until those seasonal trends are factored into the equation.

In a follow-up post, he looks at U.S. military casualties (not civilian casualties) and says:

That said, here's a chart that shows the seasonality pretty well. It comes via reader Thomas J., who has graphed U.S. troop deaths per thousand soldiers, a data series that's (a) available for every month since the beginning of the occupation, (b) highly consistent, and (c) a fairly decent proxy for the overall level of violence.

The issue that was under discussion when the seasonality issue arose was civilian casualties (not military casualties). The idea that military casualties -- which do show seasonal trends -- provide a "fairly decent proxy" for civilian casualties is an idea that comes right out of the blue (and is not correct).

He goes on:

The obvious followup would be a similar chart showing the seasonality of civilian casualties. Unfortunately, there's simply no reliable data series for civilian casualties over the course of the war...

Is that so? Obviously, he doesn't read my blog (which is no mark against him -- not many do!). But if he is interested in casualty counts, he really should swing by here every now and again. If he did, he'd learn a lot. For starters, he'd know that reliable casualty figures are available, and they can be found at Iraq Coalition Casualty Count. I have repeatedly validated the casualty trends they identify (most recently here, but also here and here and on many other occasions as well), which is why I have confidence in the numbers they provide.

The LA Times casualty figures that Drum cited show an average reduction of 260 civilian casualties per month during the 3 months of the operational troop surge compared to the 5 previous months. The ICCC figures put the drop at a similar 300 per month. You are not required to regard that decrease as being a significant (especially since casualties were at their lowest in June, not August), but one should be cognizant of what the raw data show before pronouncing on the effectiveness of the troop surge.

It's not crazy to wonder if the reduction in civilian deaths associated with the troop surge reflects nothing more than a seasonal trend (which is what Drum is concerned about). Iraq Coalition Casualty Count has numbers dating back to March of 2005. Thus, we have civilian casualty counts for the summers of 2005, 2006 and 2007. If their numbers are reliable (and I have repeatedly shown that they are), and if they show that civilian casualties tend to decrease in the summer months compared to the preceding months, then one should take heed of Drum's bottom line:

Bottom line: you should be skeptical of any claims about reductions in violence unless they take seasonality into account. So far, though, I haven't seen any credible claims of reduced violence that even mention seasonality, let alone adjust for it. That should tell you something.

I completely agree. If casualties always drop in the summer, and if they drop by an amount that is roughly proportional to the amount that they have decreased during the troop surge, then one would have to conclude that the surge has not had an appreciable effect on casualties.

On the other hand, if civilian casualties (unlike military casualties) typically increase in the summer, and if they have decreased only in the summer of 2007 (i.e., during the troop surge), then one would have to conclude that the effectiveness of the troop surge is even greater than it appears to be (if, that is, you follow Drum's suggestion). Let's inquire into the matter because we have the data, and we've had it for a long time.

ICCC has civilian casualties going back to March of 2005. The question of interest is how civilian casualties during June, July and August compare to the preceding months. If you are a novice at analyzing casualty data (as Drum clearly is), then don't try this yourself until you read this post. Here is the story for 2005, going back as far as the data allow:


Hmm. Casualties were 33% higher in the summer compared to the previous 3 months. Here is the story for 2006:


Hmm. Once again, casualties were higher in the summer months compared to the previous 5 months. Thus, if anything, casualties increase in June, July and August. Back then, it was not uncommon to encounter the suggestion that the rising casualties partly reflect the frustrations associated high temperatures coupled with no electricity for air conditioning.

Now let's look at 2007:


Gee, for the first time, civilian casualties have decreased in the summer months. And because they ordinarily increase, it seems reasonable to believe that Drum will soon claim that the troop surge is even more successful than it appears to be (the opposite of what he surely would have claimed had the data shown the opposite trend).

In truth, the civilian casualties show no apparent seasonality, so no seasonal adjustments in either direction are really warranted. However, the figures do show more global trends, and you really should take note of them. Here is the civilian casualty story for Iraq dating back to early 2005:


I fit a 6th-order polynomial through the data, and it seems to capture the actual trends. If you have not seen this chart before, you should study it hard because it contains secrets you need to know.

As you can see, casualties were on the rise early in 2005 as the insurgency got into full swing. From May of 05 to February of 06, casualties were stable. It was effectively a stalemate, except that democracy really was taking hold in Iraq and the advantage was with the freedom-promoting neocons despite the violence. But then something happened. Something significant. It caused the trend line to start accelerating upward in February or March of 2006. If you don't know what happened and who was behind it and what their main objective was, then you are not very interested in the details (and you obviously don't read my blog!). I'll tell you happened in a moment, but let's first follow the trend line a bit farther.

The next apparent trend is more recent, it is associated with the months during which the troop surge became operational, and it suggests that civilian casualties are going the other way (i.e., down). Obviously, this is the result of the troop surge, not a seasonal trend. Kevin Drum clearly wants to make it go away by adjusting for seasonality, but that's not going to happen. If one wanted to make an adjustment in this case, it would make the troop surge out to be more successful than the raw numbers suggest.

Seek truth. It is much more interesting than seeking that which supports your pre-formed opinions. If you seek the truth, you'll naturally inquire into what happened in February of 2006 to derail progress in Iraq. On February 22 of 2006, al Qaeda bombed the Golden Mosque in Samarra, not as part of a civil war but in a very deliberate effort to provoke a civil war. More specifically, the goal was to get the Shiite militias sufficiently enraged to start killing Sunnis. Civil war was the goal. We know this from a letter written by Zarqawi (then head of al Qaeda in Iraq) that was intercepted in 2004:

As for the Shi'a, we will undertake suicide operations and use car bombs to harm them...

So I say again, the only solution is to strike the religious, military, and other cadres of the Shi'a so that they revolt against the Sunnis. Some people will say, that this will be a reckless and irresponsible action that will bring the Islamic nation to a battle for which the Islamic nation is unprepared. Souls will perish and blood will be spilled.

This is, however, exactly what we want...

And it is exactly what they got. Al Qaeda audaciously bombed the revered Golden Mosque and brought the Shiite militias into the battle (exactly as planned):

Iraq Cites Arrest of a Top Local Insurgent
Officials Call Detainee No. 2 in Al-Qaeda Group

By Amit R. Paley
Washington Post Staff Writer
Monday, September 4, 2006; Page A01

BAGHDAD, Sept. 3 -- U.S. and Iraqi forces have captured a top al-Qaeda leader who ordered the bombing of a Shiite Muslim shrine in Samarra that triggered a wave of ferocious sectarian killings, Iraqi officials said Sunday.
...
In an attempt to thrust Iraq into a full-scale civil war, Saeidi supervised Haitham al-Badri, an operative under his command, in carrying out the Feb. 22 bombing of a revered golden-domed Shiite shrine in Samarra, officials said. That attack sparked brutal reprisal killings by both Shiites and Sunnis that have left thousands of people dead.

That's the day that everything changed in Iraq. Literally on that day. Let's zoom in on casualties from the month of February 2006 alone and take a look at casualties per day before and after that historic bombing:


As you can see, Zarqawi's evil plan worked like a charm. This attack was not carried out by Baathist insurgents fighting in a civil war. Instead, it was carried out by al Qaeda terrorists seeking to incite a civil war in order to destabilize the government and render American forces irrelevant. It was the day that al Qaeda declared war in Iraq. They are still fighting that war, and you are still trying to avoid that reality by calling them "insurgents" (even when they indiscriminately and intentionally slaughter hundreds of innocent civilians). Calling them insurgents makes you feel like it's all a big ol' civil war, but you need to let go of that cherished point of view (because it's wrong). These people are terrorists, and you should not be afraid to say so.

Analyze casualties in Iraq. Seek truth. Let your mind go where the evidence takes it. Blame everything on Bush if it makes it feel you better, but wake up to the fact that we are in a real war against al Qaeda in Iraq.