Editore"s Note
Tilting at Windmills

Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

September 8, 2007
By: Kevin Drum

SEASONALITY REVISITED....A few days ago I argued that much of the recent decrease in violence in Iraq was probably due to normal seasonal factors: violence tends to peak in fall and spring, and drop in winter and summer. I showed this seasonality pictorially via a chart of U.S. troop fatalities, one of the few data series that's both consistent and available for the entire course of the war.

But troop fatalities are merely a proxy for violence, and an anonymous fellow who goes by the handle Engram argues that if you look directly at civilian death statistics there's no seasonality at all. Thus, the recent drop in deaths is almost certainly due to the surge.

Now, Engram is pretty seriously invested in a highly distinctive view of the sources of violence in Iraq that I won't get into here (i.e., insurgency vs. terrorism vs. civil war), and is therefore also pretty invested in demonstrating that the surge is succeeding. Still, numbers are numbers. So what does he have?

The answer is on the right: a chart of civilian casualties since March 2005 that's taken from ICCC data and modified slightly to remove some artifacts. Engram then fits a 6th order polynomial through the data to demonstrate that there's no evident seasonality.

Unfortunately, there are several problems with this. The reason I didn't use this data in the first place is because it's available for only two and a half years. That's just not long enough to show seasonality, especially with a noisy data set. Even my original chart, with 4+ years of data, was barely long enough to pick out seasonal differences from the long-term noise of Iraqi violence, and the ICCC data is nearly useless in this regard. It's simply not going to show much seasonality whether it's there or not.

There are other problems too. First, there was an enormous secular increase in violence throughout 2006. This increase is so pronounced that it drowns out any local variability in the data when 2006 makes up nearly half your series.

Second, fitting a polynomial to a small data set, even a 6th order polynomial, is going to smooth the data. If there is any seasonality, it's going to make it harder to see, not easier.

In fact, if you look at the data hidden beneath Engram's curve, you can see glimmers of seasonality that the curve is hiding. The local peaks are all in spring and fall: May and September of 2005, March and November of 2006, and May of 2007. If I wanted to, I could draw a curve through the data that shows off this seasonality — and in fact I did draw just such a curve. But it's not worth putting up. It's just fighting one ad-libbed curve with another ad-libbed curve, and there's not much point in that. There's simply not enough data here to draw a conclusion one way or the other.

The seasonality of Iraqi violence has been well known and much discussed for a long time (especially the fall peak), and the 4+ years of troop fatality data bear that out. After all, why would troop fatalities show seasonal variation unless there were also some seasonal variation in the broader violence levels? In the end, I think the data suggests that (a) violence in Iraq follows a modest seasonal pattern, (b) violence rose dramatically throughout 2006 and then began to drop before the surge began, and (c) fatality levels this summer were down only mildly compared to earlier in the year — and probably not down at all when you take seasonality into account. Bottom line: Violence is as bad as it's ever been, sectarian cleansing is proceeding at a murderous pace, Kirkuk and Basra are both timebombs, refugees are fleeing the country at staggering rates, the Iraqi infrastructure is in ruins, and the surge just doesn't seem to be having much of an effect outside of a small number of handpicked neighborhoods in Baghdad. What's more, it's having no effect on the political situation, which continues to be a sectarian disaster.

I'll end with two related points. First, no matter how many different ways you look at the civilian casualty numbers, what you're going to see is a huge increase in violence during 2006 followed by a slow decrease. That's just all there is. If you want to argue that the surge is responsible, you can do that, but since the decrease began in December of last year that requires some pretty creative torturing of the data. Or maybe the decrease is due to seasonality. Or the Anbar Awakening. Or something else. But no matter what you do, you've basically got one observation (violence rose through November 2006 and then started to decrease) and that's it.

Second, and ironically, the ICCC data gives the lie to the primary pro-surge talking point these days: namely that violence is down 75% from its peak. It's actually down by only a fraction of that amount, and the data is even more discouraging if you pick a more appropriate starting point. In February, just before the surge began, there were 1,529 civilian fatalities. In the past couple of months we've averaged....1,528 fatalities per month.

And what about the latest talking point, namely that what we really ought to be paying attention to is the single month of August, which was awesome? Aside from the fact that this doesn't really seem to be true (and the military data to the contrary seems to be very heavily massaged), you're in pretty desperate straits when you have to hang your entire case for continuing a failed policy on equivocal data from a single month. It's no way to run a war.

POSTSCRIPT: I know, I know, I know. I keep saying I'm not going to obsess over casualty stats anymore. I feel like Michael Corleone in Godfather III. What can I say?

Kevin Drum 8:11 PM Permalink | Trackbacks | Comments (67)

Bookmark and Share

As long as they can keep people debating stats, it give the "Teh Surge is teh awesome!" claims credibility. And everyone then buys in for another endless FU.

The "war" is unwinnable. We are the cause of a significant amount of the violence. Iraq will never find stability under US occupation.

End. The. Occupation.

Posted by: Gore/Edwards 08 on September 8, 2007 at 8:19 PM | PERMALINK

Packer, posted by Big Media Matt:

The Petraeus-Crocker testimony is the kind of short-lived event on which the Administration has relied to shore up support for the war: the “Mission Accomplished” declaration, the deaths of Uday and Qusay Hussein, Saddam’s capture, the transfer of sovereignty, the three rounds of voting, the Plan for Victory, the death of Abu Musab al-Zarqawi. Every new milestone, however illusory, allows the Administration to avoid thinking ahead, to the years when the mistakes of Iraq will continue to haunt the U.S.

Posted by: Gore/Edwards 08 on September 8, 2007 at 8:26 PM | PERMALINK

If the surge is working, then why was there an increase in civilian fatalities in July and August? And the whole point of the surge was to get the political reconciliation going. The only "going" it's getting is "going to hell in a handbasket."

Remember the saying "You make peace with your enemies, not your friends." Make the Sunni and Shiite insurgents an offer, a cease fire now for an orderly withdrawal. It'll take months to withdraw everybody and that will perhaps give them time to pull together some kind of government.

Won't work? Well, it's not like what we've got is working, eh?

Posted by: tomeck on September 8, 2007 at 8:30 PM | PERMALINK

I have an idea: let's continue the Iraq war for as long as Republicans want to continue fighting it, in any way they want to fight it, and as a liberal I promise to stay quiet. However, they'll have to agree to a couple of conditions: The war should be financed solely from donations from those who want to continue fighting it, and soldiers should be given the option to fight there or stay stateside.

I don't pretend to understand why the the Republicans want this war to last, well, forever. I'm sure there are deep psychological issues I'm not qualified to analyze. However, I think it's pointless for liberals to continue trying to prove to them what a disaster it's been. What does it matter if the surge reduced or didn't reduce casualties slightly? This war was the Stupidest Fucking Idea Ever because it was pointless, unwinnable, and a distraction from fighting the real enemies of democracy and progress. But the Republicans want perpetual wars, and if it's not this one it will be another. So let them own this one, for as long as they like. I grieve for all the people who are yet to die in this horrid misadventure, but I see no way to end it as long as the pundits and the think tanks and the spineless Democrats support the lunatics, either explicitly or implicitly. So make this a Republican war, totally and completely.

Posted by: Kanenas on September 8, 2007 at 8:37 PM | PERMALINK

A 6th order polynomial fit to 30 data points?! Good God. I don't think anything more needs to be said. Aside from mockery, perhaps.

Posted by: RSA on September 8, 2007 at 8:41 PM | PERMALINK

What a crock. Nobody in his right mind fits a six-order polynomial to 30 data points.
You'll get a great fit for the data, but it will do a terrible job at predicting the next data point.
Your Engram buddy has learned how to game the data, just like George.

Posted by: Ronn Zealot on September 8, 2007 at 8:42 PM | PERMALINK

Not disagreeing or agreeing, but I can think of at least one reason why troop violence would show its own seasonality: if the new deployments peaked at certain times of the year. I don't know if this is the case, but it seems like a possibility.

Posted by: MarkedMan on September 8, 2007 at 8:49 PM | PERMALINK

I think you would expect to see a greater seasonality in the troop death numbers than in the civilian numbers, at least for the summer. The argument for the drop in violence in the summer is that it's simply too hot to fight (120 deg). But the Iraqis are likely to be more adapted to the heat than we are. So it would seem plausible that the US military would reduce its activities in the summer and stay on the air conditioned base, making them less vulnerable. The Iraqis would likely reduce their activities less in the summer, since they are more used to the heat and don't have an air conditioned place to wait out the summer.

Posted by: fostert on September 8, 2007 at 9:04 PM | PERMALINK

I don't pretend to understand why the the Republicans want this war to last, well, forever.

It's because they are fundamentally incapable of admitting they made a mistake. It starts with Bush and goes all the way down to some lonely fundie Christian in Kansas. They are right, they've always been right, they always will be right.

The facts of the situation don't matter.

Posted by: tomeck on September 8, 2007 at 9:13 PM | PERMALINK

I think the Iraq war has been a great success, regardless of any statistic. The Democrats are more divided than ever, and are unable to coordinate a domestic political strategy against the most unpopular president in history.

THAT, my friends, is a successful war.

Posted by: anonymous on September 8, 2007 at 9:18 PM | PERMALINK

Kevin Drum are being a little disingenuous in a couple respects. First, the fitted polynomial equation was not engram's primary argument against seasonality in civilian casualty data. In fact, he has some year-by-year analyses. Second, it won't do to say that a particular time series (i.e., for civilian deaths) is too short, therefore we will rely on a longer time series (i.e., for military deaths) that measures something else. Come on. The only honest response if you can't demonstrate seasonality in a short time series is agnosticism. This is especially true when anyone with five minutes to spare could invent a half dozen reasons why one series would show seasonal variation and the other would not.

Posted by: y81 on September 8, 2007 at 9:20 PM | PERMALINK

First, there was an enormous secular increase in violence throughout 2006. This increase is so pronounced that it drowns out any local variability in the data when 2006 makes up nearly half your series.

This is exactly right. If there is a seasonality to this data, it could be best understood as represented by the positive or negative value of the second derivative, not the first -- i.e., not changes from an upward to a downward slope, but rather as downward changes to the already clear and definite upward slope.

Posted by: frankly0 on September 8, 2007 at 9:22 PM | PERMALINK

Five interrelated points to make:

1) People "wait" and "pause" when there is a change in the political climate over there to see if the wind direction is going to change. We are at one of those "wait and see" moments now. That's primarily why Moqtada told his followers not to attack us. There isn't enough visibility-when the visibility gets low they back off before resuming any power play. This just happens to be coincident with a slight seasonal lull so it makes it more pronounced now.

2) You have two dynamics going on here: The Samarra bombing set off the spike in sectarian violence; violence against the US is not always directly related to the ebbs and flows of sectarian violence.

3) It isn't the "surge" that's working. What's "working" is the continuing segregation of the Shia and Sunni populations. At some point this should decrease naturally with or without our help. I think we've been "helping" them desegregate.

4) The "lack of seasonality" in THEIR violence has more to do with their political situation or lack thereof. The "seasonality" that you DO see with OUR troops has to do with US not being able to handle the heat-I suggest that we aren't EXPOSED as much as targets.

5) Motivations for violence. If it is hotter than hell and uncomfortable, you might delay setting that IED on the roadway for the Americans for another day, but if some fucker KILLS a member of your family, that's not going to stop you from putting a bullet in him, or use a power drill or whatever.

Posted by: Doc at the Radar Station on September 8, 2007 at 9:30 PM | PERMALINK

Just to clarify what I said, I think that the upward trend mostly applies to 2006 -- it seems more stable in both 2005 and 2007 -- and that the way to understand any seasonality in the data in 2006 might best be with the second derivative.

Of course, this admittedly is a bit hokey, but with small data sets and complex situations, it's the best one can do.

Posted by: frankly0 on September 8, 2007 at 9:32 PM | PERMALINK

Another thing to keep in mind is that violence seems to peak in the fall beginning with Ramadan. Ramadan is based on a lunar calendar so that the month of Ramadan begins 11 days earlier each year. So back in 2003, Ramandan began on October 28. This year it has advanced so that it begins September 13. This will tend to shift the seasonal effect.

Posted by: ben on September 8, 2007 at 9:34 PM | PERMALINK

anonymous does a nice job of making my point. The deaths of tens of thousands Iraqis and thousands of American soldiers matters little when compared to some perceived political success for Republicans.

Thanks man.

Posted by: tomeck on September 8, 2007 at 9:43 PM | PERMALINK

Well, America's favorite general is starting out the new week with wiggle that amonts to a big-assed lie.

In a letter to his troops, he writes:

One of the justifications for the surge, after all, was that it would help create the space for Iraqi leaders to tackle the tough questions and agree on key pieces of "national reconciliation" legislation. Emp mine.


I thought it was the only reason.

Posted by: Keith G on September 8, 2007 at 9:45 PM | PERMALINK

Well, the Republicans torture everything else.

So why wouldn't they torture statistics?

Posted by: lambert strether on September 8, 2007 at 9:45 PM | PERMALINK

Why not try a spatio-temporal analysis? Does the data exist to look at correlations between violence-however one chooses to measure it-and troop numbers as a function of location and time?

Posted by: rege on September 8, 2007 at 9:46 PM | PERMALINK

Dead Iraqis don't matter to U.S. citizens. They don't matter to Bush, nor the military, nor the electorate. They are the proverbial trees falling in an uninhabited forest. Do they make a sound as they die? Who cares? Certainly no one in this nation. It also is of no importance if they're sick, homeless, destitute or rotting in some godforsaken gulag Cheney has constructed. All peoples of the Middle East and all Muslims are the new American Indian. Remember "The only good Indian is a dead Indian"? Fast forward 125 years and plug in Arabs for Indians. They are the new American object of bloodlust. Bin Laden and Bush have teamed up to warp this nation beyond any hope of repair or forgiveness. We're all a bunch of sick, twisted souls on a slow ride to hell.

Posted by: steve duncan on September 8, 2007 at 10:02 PM | PERMALINK

That's the ticket to ending this madness.


Fucking unbelievable.

Posted by: gregor on September 8, 2007 at 10:03 PM | PERMALINK

Hmm . . . 6th order polynomial . . . why?
I wonder . . . what's the highest order that Excel's trend-line analysis lets you fit . . . a 6th order, right?

If I were looking for seasonality, I'd have fit a sine function to it.

Posted by: Barry on September 8, 2007 at 10:04 PM | PERMALINK

If I were questioning Petraeus, I'd refer to the conflict as the Iraqi Civil War. He is leading American participation in that Iraqi Civil War, and he should answer questions about that, and nothing else.

May as well define the situation as it stands, at least.

Posted by: blatherskite on September 8, 2007 at 10:23 PM | PERMALINK

Mcnamara has to be aghast. Even if the stats were totally encouraging, there is no way the factions don't go at each other big time as soon as we are gone(if ever), unless a sunni regime every bit as repressive and devious as sadaam's is installed. The world's last superpower is committing diplomatic, economic, cultural, and military suicide.

Posted by: Michael7843853 G-O/F in 08 on September 8, 2007 at 10:35 PM | PERMALINK

Ah, Kevin.

There is no seasonality to the civilian deaths, as anyone in his right mind would conclude after the most quick glace at the data. Also, the polynomial (6 degree) shows the trending down of the violence and filters out any seasonal noise.

Zogby polls show that Americans feel we can still win. None of the independant reports on Iraq coming out have said the suge is failing, despite what the liberal MSM says about it. Get your heads out of the echo thing.

Posted by: egbert on September 8, 2007 at 10:41 PM | PERMALINK

Kevin, at what point do you start rolling the little steel balls around in your hand?

Posted by: harry on September 8, 2007 at 10:43 PM | PERMALINK

I'm stunned at the stupidity of this curve fitting. As if there is some underlying process that allows us to create a function, d(t), for predicting deaths. As if the coefficients of the fitted polynomial have any meaning. As if 30 data points possess such overwhelming complexity that the truth can be revealed only through the great power of a least-squares fit.

Posted by: Dr. Drang on September 8, 2007 at 10:48 PM | PERMALINK

And the highest power in a polynomial is its degree, not its order.

Posted by: Dr. Drang on September 8, 2007 at 10:51 PM | PERMALINK

Second, fitting a polynomial to a small data set, even a 6th order polynomial, is going to smooth the data. If there is any seasonality, it's going to make it harder to see, not easier.

As many others have pointed out, this is the key stupidity. Nothing else needs to be said.

When looking for winter/spring/summer/fall seasonality over a 30 month period, one would expect to have to model 10 min/maxima. A 6th degree polynomial will model at most 5 min/maxima.

Uh, yeah, I think that just might smooth out the data just a tad.

Gawd, I hate wingnuts.

Posted by: Disputo on September 8, 2007 at 11:54 PM | PERMALINK

I'm really skeptical of the idea, represented by comments like y81's (9:20), that there would be a big difference in seasonality between Iraqi civilian deaths and U.S. deaths. Long story short, I just don't think the racial differences between the Iraqis/middle easterners and the U.S. army, as they relate to ability to withstand heat, can account for enough of a difference to matter, especially when you consider who the Americans and the Iraqi insurgents are: the U.S. military is not a bunch of American civilians taking a break from stormy home-lives in a distant middle eastern land for 6 months, and ready to sit out a hot summer indoors. It's implausible and funny to me, the idea that an American military commander would stop running certain kinds of ops because it was hot out in the summer, while ordinary Iraqis were running around like everything was normal. The U.S. military is running military missions and they have experts on the human body to tell them what they have to do to stand up to the heat. You can do a lot just by covering your head and having water with you. It simply doesn't make sense that it would be so inhospitable in the middle of Iraqis cities throughout the summer, that Americans would often not be able to go outdoors, but Iraqis would be basically fine going out. These are human beings, not camels.

Posted by: Swan on September 8, 2007 at 11:59 PM | PERMALINK

Oh, and one other thing -- irrespective of seasonality, if you use a odd-th degree polynomial to model a data series that begins low and then increases, the resulting polynomial will always show a decrease at the end of the series, which is of course all the wingnut was trying to prove.

Did I mention recently how much I hate lying dumbass wingnuts?

Posted by: Disputo on September 9, 2007 at 12:04 AM | PERMALINK

sorry, that should have been "even-th degree polynomial" which generates an odd number of min/maxima.

Posted by: Disputo on September 9, 2007 at 12:08 AM | PERMALINK

What I'm saying, in other words, is if there's a seasonal effect on violence, it should be the seasonal difference in Iraqi activity, not American activity, that accounts for it, as Iraqis can set their own agendas and decide not to go out when they're uncomfortable. Also, since Iraqis blend in more in their own country, insurgents have more of an incentive to operate during the day, and U.S. forces have more of an incentive (relative to the insurgents) to rely on night for surprise, since they can't rely on blending in with the surroundings during the day. Therefore, the hottest part of the day, daytime, no matter the season, shouldn't discourage Iraqi fighters' ops the way it discourages American ops, and Americans have incentives to operate when it's a lot cooler out (during night) which makes the summer environment a lot different. American forces could be just as active during the summer and just as comfortable by shifting ops to night-time ops, while Iraqis would remain more suspicious/conspicuous if they were creeping around at night.

Posted by: Swan on September 9, 2007 at 12:14 AM | PERMALINK

Ah, Egbert.

Seasonal noise?

Posted by: DFH on September 9, 2007 at 12:49 AM | PERMALINK

Swan raises a good point about night operations. I would add that the US military has a tremendous advantage at night because they have night vision goggles and the insurgents don't. They also have an additional advantage because of curfew laws. Obviously, anyone out after the curfew is suspect, so there won't be many insurgents out. So, for the US, night operations should be much safer than daytime operations. Interestingly, this would cause a seasonal variation. If the US troops shift to more night operations in the summer, then you would expect fewer casualties because such operations are inherently safer.

Posted by: fostert on September 9, 2007 at 1:07 AM | PERMALINK

I read about Generals who say violence in Iraq will increase around Ramadan.

As well can the decrease of violence can it be attributed to Sadrs recent stand down?

Posted by: Ya Know.... on September 9, 2007 at 1:22 AM | PERMALINK

A '6th order polynomial'? That's a horrible way to test for seasonal effects - pretty much guaranteed to smooth over any seasonal signal. Suggest to the guy that he find a stats package that does at least basic time series analysis, and someone who can show him how to use it.

Posted by: gjhick on September 9, 2007 at 1:35 AM | PERMALINK

Garbage in. Given the probable underreporting of civilian deaths, flaws in the data collection are going to swamp any info you can pull out of this chart. Two hundred civilian deaths in March of 2005? The real figure might be three times that, or five times more. Heck-a-rooni, even the ICCC page you link to says right at the top that the numbers are wrong.

Posted by: mcdruid on September 9, 2007 at 2:03 AM | PERMALINK

Barry and gjhick ave mentioned this, but is important that people understand this. If you are trying to fit cyclical data to a polynomial, you really won't get a good fit. The reason is that a polynomial function creates a fixed maximum number of inflection points. When you are looking for cyclical patterns, fitting to a cyclical function is more appropriate. This is a situation where Fourier analysis (combination of sine functions) makes the most sense.

Posted by: fostert on September 9, 2007 at 2:25 AM | PERMALINK


Heres a bombing 8 hrs ago, 15 Shiites killed. Ramadan starts thursday. Register site. Talks about possible Ramadan increase in violence.

Posted by: Ya Know.... on September 9, 2007 at 2:28 AM | PERMALINK

fostert wrote:

So, for the US, night operations should be much safer than daytime operations. Interestingly, this would cause a seasonal variation. If the US troops shift to more night operations in the summer, then you would expect fewer casualties because such operations are inherently safer

You're totally missing my point. There's no reason why they would only do nighttime ops in summer if nighttime ops were better for them. They'd rely on them more all year round. Night isn't better in the summer than it is in the winter, or anything. They'd do just as many night ops all year round. The use of night ops by the U.S. explains why it's the change in Iraqi activity, not the change in U.S. activity (despite Americans' air conditioners) that effects any seasonality and why heat effects the Iraqis, not the Americans as far as seasonality.

If not operations are inherently safer, there's no reason the U.S. would use them in the summer, more than non-summer. There's nothing about non-summer that keep the U.S. from running night-ops.

Posted by: Swan on September 9, 2007 at 2:48 AM | PERMALINK

Fitting polynomials is not the way to go to analyse seasonality in time series. If anyone's interested in using Excel to carry out this sort of data analysis, I recommend Michael R. Middleton's "Data Analysis Using Excel", which includes step-by-step instructions for techniques such as classical time series decomposition.

Posted by: derek on September 9, 2007 at 5:28 AM | PERMALINK

What everyone else is saying about the polynomial fit. Use a function that has no periodicity, and of course your fit won't display it either.

That said, the function clearly has a non-periodic part, namely, its general trend upward. If you want to answer the seasonality question, subtract (or maybe divide) off the trend and try fit the remainder to a year-periodic Fourier series. Success would be getting a better fit with fewer parameters.

Posted by: matt on September 9, 2007 at 5:58 AM | PERMALINK

Somehow, I don't imagine that Engram will change his mind based on the technical rebuttals offered by the stats warriors in this comments thread. But I'm impressed!

The other problem for Engram's hypothesis is that of causality. Assuming an association between the surge and fewer civilian casualties, is the cause the surge? Alternative hypotheses would be 1) that fewer targets remain to be assassinated because the Shi'as have taken over Baghdad and the Sunni have fled or been killed ; or that 2) Iraqis have taken advantage of the 6-month surge to build their strength, pending the September report that might lead to the withdrawal of US troops. If 1) then the surge has done nothing, and 2) then whether the US stays or goes, we will see an increase in violence, and the surge has done nothing.

Posted by: PTate in FR on September 9, 2007 at 6:26 AM | PERMALINK


Nice graph, but you could make all your Excel graphs look sharper by right-clicking on the graph area and getting rid of the gray background.


Posted by: Steve Sailer on September 9, 2007 at 7:33 AM | PERMALINK

Yeah, civilian casualties are down - Huzzah - Kind of hard to kill more of those fleeing into Syria, when no one will sell and train Iraqis the F-Series jets to reprise the Highway of Death scenes coming out of Kuwait in Gulf I.

2.2 million displaced civilians in Iraq - So many trying to cross the border into Syria, that the Syrian government is starting to impose visas. Lots of new ethnically cleansed neighborhoods in Iraq - BUT, The Surge in Bush's Pants is Working.

Posted by: thethirdPaul on September 9, 2007 at 9:07 AM | PERMALINK

Nice summary Kevin, particulary the Bottom Line down around paragraph 9. Keep going with this stats subject if you want, but I think you've settled the argument. Congress would do well to read this post. Sad to say they'll probably get less useful information from the General's report. He's under too much pressure to postpone the inevitable beyond Bush's term.

Posted by: dennisS on September 9, 2007 at 9:25 AM | PERMALINK

It isn't the "surge" that's working. What's "working" is the continuing segregation of the Shia and Sunni populations.
This is exactly right, and the rest of Doc at the Radar Station's list is on target. I'd only add that there lots of possible explanations for a decrease in civilian deaths which may or may not be seasonal, but it really doesn't matter. A few sometimes overlooked:

- The perpetrators of violence usually have some end in mind. Once a neighborhood is purged and cleansed, at the very least the local factions have achieved their goals.

- In addition to the millions of people who've been displaced, relocated or have left the country altogether, if some of the higher estimates of civilian casualties are correct, killers may simply be running out of people to kill in certain areas.

- The people who remain start to find ways to live that reduce their risk of getting killed.

And of course, there's always fudging the stats in the first place with new definitions of what constitutes victims of sectarian violence (a bullet in the back of the head), vs. just plain old violence (a bullet in the front). As we've seen, with this administration one can never dismiss mendacity, especially since they never bothered to count civilian casualties before the "surge", now did they?

Posted by: R. Porrofatto on September 9, 2007 at 9:48 AM | PERMALINK

Once you kill everyone or drive everyone away, your casualty rates are going to fall. So, they've got that going for them.

Posted by: Jeffrey Davis on September 9, 2007 at 10:25 AM | PERMALINK

Death rates of Jews living in major German cities, from 43 to 45 were remarkably low. Just goes to show how citizens can forget their differences and bond.

Posted by: thethirdPaul on September 9, 2007 at 10:31 AM | PERMALINK

you could make all your Excel graphs look sharper by right-clicking on the graph area and getting rid of the gray background.

Agreed, there are many ways to improve the dreadful Excel defaults, but that's the first and simplest: click on the gray and hit "delete".

Also, don't use jpeg images to display graphics: it's a compression technique designed for photographs with many colors and gradients, and the algorithm's attempt to match the sharp edges and uniform colors of a graph will result in ugly smudges around text, lines and markers. GIF or PNG images will look better and usually be more compact as well: JPG offers no advantage for this kind of image.

Posted by: derek on September 9, 2007 at 10:56 AM | PERMALINK

Anybody want to try fitting an odd-degree polynomial to this data? I haven't tried it, but I'm guessing it would accentuate the positive concavity of the last few months.

Posted by: reino on September 9, 2007 at 11:06 AM | PERMALINK
I know, I know, I know. I keep saying I'm not going to obsess over casualty stats anymore.

No, Kevin, feel free to obsess a bit. You analysis here is compelling and there is always the chance that there will be a few readers who take a look who have not previously heard anything that disputes the noise coming from the Bush administration.

Posted by: little ole jim from red country on September 9, 2007 at 11:25 AM | PERMALINK

Did you compare only with the dates? did you not use temperature? did you not compare to weather data?

because the point here is that it is very, very hot.

Posted by: Hubris Sonic on September 9, 2007 at 11:30 AM | PERMALINK

Here's another question to highlight the stupidity of this model: what does it look like outside the bounds of May '05 to August '07? It's concave downward, which means that at some point it's predicting negative numbers of civilian casualties. Will dead Iraqis come back to life as zombies or vampires?

Posted by: RSA on September 9, 2007 at 12:03 PM | PERMALINK

"It's just fighting one ad-libbed curve with another ad-libbed curve, and there's not much point in that."

What a great reason for leaving the first ad-libbed curve up there where idiots who have to read the whole article to see why it's so braindead will see it without having to think about whether the smooth curve in the data actually makes any sense.

If you'd put the second graph up, then the pictures would have conveyed in mere seconds what your five hundred words of turgid prose— however much fun it was to write— requires minutes of precious time to consume.

Kevin, when are you going to figure out how this goddamn game is played?

Posted by: s9 on September 9, 2007 at 12:45 PM | PERMALINK


this is a first rate analysis, a pleasure to read.

Posted by: orionATL on September 9, 2007 at 12:47 PM | PERMALINK

"If not operations are inherently safer, there's no reason the U.S. would use them in the summer, more than non-summer. There's nothing about non-summer that keep the U.S. from running night-ops."

I think we should keep in mind something about the safety of operations: safe operations are usually less effective. Patrolling empty streets may be safe, but it isn't likely to capture or kill many insurgents. Obviously, we could make the operations even safer by conducting them in the US rather than Iraq, but what would be the point? In general, you want to conduct operations when you are most likely to encounter the enemy, which would be daytime. But in the summer, you will likely conduct shorter operations during the daytime to reduce the risks of heat related medical problems. And you will shift as many operations to the early morning and evening as possible. In the cooler months, you will conduct more daytime operations because they are more effective and you can do them without your troops getting heat stroke.

Posted by: fostert on September 9, 2007 at 12:59 PM | PERMALINK

As Mr. Engram noted in a response here the 6th order polynomial was never the "reasoning" for his assertion that there was no seaonaility, it was a way to show the trend. Since folks seem to have frantically latched onto the 6th polynomical, Mr Engram has provide a response on his blog.

If the readers are actually interested in the analysis, you'll see that comparing Mr. Drums analysis to Mr Engrams is insulting the word analysis.

Posted by: dude on September 9, 2007 at 2:02 PM | PERMALINK

The first problem with the graph is that it uses ICCC figures. The ICCC website starts out with the following disclaimer:

"This is not a complete list, nor can we verify these totals. This is simply a compilation of deaths reported by news agencies. Actual totals for Iraqi deaths are much higher than the numbers recorded on this site."

So all this calculation starts from admittedly incomplete data. Garbage in - garbage out.

The second problem is that it doesn't take into account the ebb and flow of all guerilla wars. As Mao Zedong defined the primary tactics of guerilla war:

""When the enemy advances, withdraw; when he stops, harass; when he tires, strike; when he retreats, pursue."

Posted by: Blue Sun on September 9, 2007 at 4:54 PM | PERMALINK

Blue sun...If you are really interested in accuracy and not in just trying to tear down the messenger spend some time on Mr. Engrams site. He has spent many,many hours validating his data. He has validated it against just about every tabulation of data that I've seen.

In this case you are incorrect. The data is correct and in fact is the most validated I've seen anywhere.

Posted by: dude1394 on September 9, 2007 at 6:41 PM | PERMALINK

On validation of his model, engram has done no such thing. His analysis is completely ad hoc: "Here, I have drawn a 3rd-order polynomial through the data because it fits significantly better than a 2nd-order polynomial (but a 4th-order function does not improve the fit further, so I stopped at 3)." Please. That's not data analysis.

Posted by: RSA on September 9, 2007 at 8:47 PM | PERMALINK

Are you questioning his data or his analysis? The above poster was questioning his data. The main analysis by engram is here. Which easily shows no seaonality. It's incumbent upon Mr. Drum to prove seasonality I would think, which he cannot and does not. As far as 3rd, 4th, 50 polynomial, that's not what Mr. Engram is using to make his case.


Posted by: dude1394 on September 9, 2007 at 8:53 PM | PERMALINK

As far as 3rd, 4th, 50 polynomial, that's not what Mr. Engram is using to make his case.

Actually, I'm agnostic about any seasonal effect. What I do object to is apparent incompetence in data analysis in general. Why should a serious person show a meaningless model for any reason?

Posted by: RSA on September 9, 2007 at 9:34 PM | PERMALINK

i need to actually read the post through. jeez

Posted by: Hubris Sonic on September 9, 2007 at 10:33 PM | PERMALINK

So when did the surge start, in March or July. I was constantly told all spring that you couldn't measure the surge because the troops hadn't rotated in. Now Kevin shows me a graph that some states shows lower casualties due to the surge, except for the fact that it show the drop in the spring, not the summer. Which is it?

I don't understand the cake and eat it too crowd...

Posted by: justmy2 on September 10, 2007 at 12:31 AM | PERMALINK

Hmm. Concave up at the end. Casualties increasing with t^6! This Iraq problem will sort itself out right quick.

Posted by: rs on September 10, 2007 at 8:20 AM | PERMALINK



Read Jonathan Rowe remembrance and articles
Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

Advertise in WM

buy from Amazon and
support the Monthly