Not so different

by Ted on July 1, 2005

Shortly after September 11th, 2001, Andrew Sullivan wrote:[1]

The terrorists have done the rest. The middle part of the country – the great red zone that voted for Bush – is clearly ready for war. The decadent Left in its enclaves on the coasts is not dead – and may well mount what amounts to a fifth column. But by striking at the heart of New York City, the terrorists ensured that at least one deep segment of the country ill-disposed toward a new president is now the most passionate in his defense. Anyone who has ever tried to get one over on a New Yorker knows what I mean. The demons who started this have no idea about the kind of people they have taken on.

I thought of this quote when I came across something on Lifehacker: a map of US military casualties in Iraq by hometown. From the maps, it didn’t immediately appear that the Democratic-leaning coastal states had avoided their share of casualties. But, of course, the coasts have a heavy share of the population.

Have states with a high percentage of Bush voters suffered a larger share of casualties per capita? If Sullivan’s statement had been true, I might expect to see that I could predict the rate of military casualties per capita by looking at Bush’s support in 2000 and 2004.

This is not the case. Only a very small percent of the variance in casualties by state can be explained by looking at the 2000/2004 elections. The relationships do not approach significance. I ran simple regressions of the most recent casualties by state divided by state population, versus Bush’s percentage of the popular vote by state in 2000 and 2004. The correlation between state casualites and Bush’s 2000 state percentages was .295; a regression produced an adjusted r-squared of .07. In 2004, the correlation was .241, with an adjusted r-squared of .04. These are not significant numbers.

I was surprised by some of the results. Eight of the ten states with the greatest casualty rates went for Bush in 2004, as did six of the states with the lowest casalty rates. However, the state with the greatest percentage of losses in Iraq was Vermont. My state of Texas was number 19. (Feel free to write for this data set; hand-coding an HTML table takes forever.)

Now, there are any number of methodological flaws in this approach. Casualty numbers for many states are small; in these states, a few more or less casualties would change their place in the rankings. It’s good enough for a blog post, not good enough for much more. Just thought I’d throw it out there, and reiterate that these blue-vs-red comparisons are hard to justify, more often than not.

[1] I wrote to Sullivan to give him a chance to distance himself from this quote. He wrote back:

of course i stand by it because it is true. when you read the full context, you also see that it occurs in a passage highlighting the patriotism of new yorkers. i meant it to apply to a small number of hard lefties, hence “enclaves”; and i didn’t even mean it to refer to the left as a whole, let alone liberals. i used the words ‘decadent left.’ looking back, i wish i hadn’t written it in ways that could be misinterpreted. but it’s still true. ask ward churchill or michael moore.
if you quote it, i would prefer you link to my posting of the whole article and include the entire paragraph in which the quote appeared.

I have any number of disagreements with this reply. His critics didn’t create the geographical binary between great red zone in the middle vs. decadent leftist coasts. I don’t follow Michael Moore enough to know why he’d be accused of being a fifth columnist. Ward Churchill is certainly a loathsome scumbag, but he lives and works in that great red zone state of Colorado (#37). I sincerely doubt that Sullivan had heard of him when he wrote this. Furthermore, I don’t think that many people read Sullivan’s phrase “decadent Left” as “left-wing extremists”, especially given the vitriolic firehose of his other output at the time. But, there it is.



Uncle Kvetch 07.01.05 at 3:09 pm

I don’t follow Michael Moore enough to know why he’d be accused of being a fifth columnist.



Barry 07.01.05 at 3:13 pm

He’s fat, haven’t you heard?


Ray 07.01.05 at 3:15 pm

‘Casualties’ is probably the wrong figure to look at anyway, because 1) the number of deaths is so small compared to the number who have served, and 2) many of those who were sent to Iraq would have enlisted before 2001 (especially all those National Guards), so they weren’t there as a response to 11/9.
The interesting thing would be differential rates of enlistment in the various states since 2001 (compared to the change in Bush’s vote in 2000 and 2004!), but that would probably be a real pig to calculate.


John Emerson 07.01.05 at 3:17 pm

I think that the most promising angle for analyzing casualties would be looking at the high schools they graduated from. Blue State vs, Red State doesn’t tell you a lot. Oregon is a blue state, but I suspect that most of the Oregon troops come from the red part of the state. If I had the names of the high schools, I’d know.


paul 07.01.05 at 3:20 pm

“I don’t follow Michael Moore enough to know why he’d be accused of being a fifth columnist.”

Judging by the perennial description of him, he’s actually a fifth, sixth and seventh columnist. Or rather first, second, third, fourth, and, unfortunately, fifth columnist.

There, that’s it. Like many of the rest of us, he is a first through fourth columnist. But those columns are just not big enough to contain him, so it is accurate (though perhaps incomplete) to characterize him as a fifth columnist.


Jonah 07.01.05 at 3:20 pm

Perhaps looking at the county-by-county election results would bolster your argument. The casualty data has the city, just need a city-to-county lookup.


paul 07.01.05 at 3:21 pm

Barry – you beat me to it, though I think your reasoning in incomplete.


Uncle Kvetch 07.01.05 at 3:24 pm

Ted, in case my first response (and Barry’s helpful addendum) were not sufficient, I’ll let Miss America take over:

Prosecuting Attorney: Tell the court why you think he is a traitor to this country.

Miss America: I think Mr. Mellish is a traitor to this country because his views are different from the views of the president and others of his kind. Differences of opinion should be tolerated, but not when they are too different. Then he becomes a subversive mother.

–Woody Allen, Bananas

What I’d really love to know is what MM said or did before 9/11, such that he was already the kind of person Sullivan had in mind when he levelled his sadly “misinterpreted” accusations of treason. Or maybe Sully was just prescient enough to know that MM would say or do something treasonous at some point, because that’s just the kind of person he is. And that’s the kind of brilliance that explains why Sully’s such a highly regarded star in our political firmament.


abb1 07.01.05 at 4:15 pm

I enjoy hyperbolic metaphors as much as anyone else (or maybe more), but what’s all this crap about the ‘fifth column’? I mean, OK, Spanish fascists in 1936 had four columns marching on Madrid and they expected the fifth column inside the city to rise up, attack the Spanish Republicans and take over. Fine.

So, now, what about this ‘decadent Left’? Were they in 2001 getting ready to, like, suddenly attack the good ol’ boys inside the US and convert ’em all to Islam or something? One has to be a total cretin to believe in something like that.

So, what is it, then? Clearly, it’s nothing but a difference of opinion on how to respond to the 9/11 attack. The decadent Left had its opinion, the fascist Right boys had theirs. So, now, again, how does this amount to a ‘fifth column’?


joel turnipseed 07.01.05 at 4:17 pm

A Minnesota microdata–about what I expected… commentary to follow. (Hope table export from Excel doesn’t suck).

% Victory
PCI as % Mean


Big Stone
DFL (7th)

DFL (8th)

DFL (7th)

DFL (8th)
Lance Corporal

DFL (8th)
Private 1st Class

DFL (8th)
Private 1st Class

DFL (8th)

St. Louis
DFL (8th)

DFL (8th)
Staff Sergeant

GOP (6th)
Staff Sergeant

DFL (4th)

GOP (1st)

GOP (2nd)
Private 1st Class

GOP (2nd)
Petty Officer 3rd Class

GOP (3rd)
Sergeant 1st Class

GOP (3rd)

GOP (3rd)
Staff Sergeant


GOP (1st)
1st Lieutenant

Chief Warrant Officer

GOP (6th)

GOP (6th)
Chief Warrant Officer (CW4)


nic 07.01.05 at 4:17 pm

There’s a really big methological problem lurking in the background, that John Emerson hints at. You can ask the question: “Have states with a high percentage of Bush voters suffered a larger share of casualties per capita?”. But it’s a meaningless question if what you’re trying to get at is: “Have Republicans suffered a larger share of casualties per capita than Democrats?”, because of the ecological fallacy. You can ask questions about geographical distibutions, but questions posed about spatial units sadly don’t map on to questions about individuals in a meaningful way. I admit that this is a problem implicit in Sullivan’s original quote.

“Casualty numbers for many states are small; in these states, a few more or less casualties would change their place in the rankings.”

Is this strictly a problem – as it relates to the question posed? Casualty figures aren’t samples, they’re a population. I’m not sure there is any randomness to worry about. If you were asking whether people in states with a high percentage of Bush voters suffered a larger risk of being a casualty, casualty figures would be a sample of the probability of being wounded, and it’d be relevant. But if you’re asking about total casualties per capita then all the data’s there, chance doesn’t come into it, and the data is what the data is.


joel turnipseed 07.01.05 at 4:34 pm

OK… until we get preview back, I give up. Basically, what you find in the date (when you can view it in table format) is that there is (small data set given) that most of the casualties come from rural areas, the higher ranks come from more wealthy areas (and this corresponds to political affiliation), and, overall, somewhat lower per-capita income for hometowns of all casualties (using county data… my hunch, given what I know of cities listed in stats, is that it’s actually lower than this data states).


Jake 07.01.05 at 4:35 pm

Like nic, I’m skeptical about this sort of thing. And while I enjoy, as a good liberal, noting the higher teen pregnancy, suicide, murder, welfare etc. rates in red states relative to blue states, I find looking at the gastner-shalizi-newman maps ( is a salutary exercise.

Andrew Sullivan is still a wanker in any case, though.


joel turnipseed 07.01.05 at 4:49 pm

OK, well, I should have thought of this sooner: if you’re interested in nice view of Minnesota dataset, it’s here.


idiot/Savant 07.01.05 at 5:11 pm

hand-coding an HTML table takes forever



Matt Weiner 07.01.05 at 5:31 pm

I believe Paul Krugman actually is a fifth columnist, if you order the columnists alphabetically by last name.


Matt Weiner 07.01.05 at 5:32 pm

Carp, missed one out. It’s Kristof.


Matt 07.01.05 at 6:02 pm

I hold Sullivan’s inability to use the ‘shift’ key against him.


John Emerson 07.01.05 at 7:14 pm

Joel’s URL is garbled but works.

Joel, 8 of the 20 Minnesota casualties come from the 8th district where I grew up — probably the poorest and most rural in the state. Its representative, Colin Peterson, is a conservative Democrat who caucyssess with the Blue dogs.

I think that there are about three overlapping selectors for military participation. First, some families, areas, and ethnic groups highly honor military service and expect it of young men. Second, some go into the military for excitement, adventure, and to get away from home. Third, for some basically functional kids with no immediate prospects, the military seems like a good deal practically speaking. Only for the third of these groups is economic status the major factor.


pog 07.01.05 at 7:16 pm

Have a look at this paper by UC Berkeley’s Ted Miguel. Ted and is co-author argue the casualty rate has an impact on the 2004 (versus 2000) presidential election. Is Berkeley part of the enclave?


don 07.01.05 at 8:56 pm

I which column are George Will, O’Reilly, and Coulter? Patriots all. None are urging young people to sign up for the glorious, mother of all wars, Iraq.


joel turnipseed 07.02.05 at 1:07 am

John –

Thanks for pointing out the bad link (this no preview thing kills me). As for “why kids join up?” … well, I grew up in West Duluth–if yer from that area, you know what a shit-hole it is (I thought the American Indian kids were rich…). I joined the Marines at 17. So, I know a few things about the wildly variant reasons kids join, too (mine was one part “fuck you,” one part “what the hell else?”). Who knew, in 1985, that they’d actually send my ass to war in 1990? Who among the 1700+ dead thought the same thing about 2003-2005 when they joined? I just wanted to complicate the “blue state/red state” picture by showing how within one “blue” state, the stats showed a much more complicated affair.


Kevin Phillips Bong 07.02.05 at 3:37 am

When Sullivan’s column was originally posted in the Sunday Times he wrote: “The decadent left in its enclaves on the coasts is not dead – and may well mount a fifth column.” That version is stil up here. “[W]hat amounts to” is a later addition.


James Wimberley 07.02.05 at 4:49 am

Matt wrote “I hold Sullivan’s inability to use the ‘shift’ key against him.”
Of course Sullivan is shiftless.


James Wimberley 07.02.05 at 4:59 am

Seriously. Tom Wolfe’s brilliant essay “The Last American hero” on the Appalachians and stock car racing ( includes a passage that encapsulates the paradigm of the macho South:
“In the Korean War, there were seventy-eight Medal of Honor winners. Thirty-two of them were from the South, and practically all of the thirty-two were from small towns in or near the Appalachians. The New York metropolitan area, which has more people than all these towns put together, had three Medal of Honor winners, and one of them had just moved to New York from the Appalachian region of West Virginia.”
If I’m getting this right, the Iraq casualties simply don’t reflect this pattern. Is it really a myth, and red blood flows in blue veins?


Barry 07.02.05 at 7:41 am

Ted: “hand-coding an HTML table takes forever”

idiot/savant: “Perl?”

Take the Excel table, copy in to Word, pasting special as rtf, save Word document as html. The various style aspects of the blog page might still mess this up, and I’m not somewhere where I can test this.


Randy Paul 07.02.05 at 12:50 pm

Matt wrote “I hold Sullivan’s inability to use the ‘shift’ key against him.”

Of course Sullivan is shiftless.

Actually, I believe he’s quite shifty.


W. Kiernan 07.03.05 at 8:13 am

The “decadent left” a “fifth column,” Sullivan? It was my first thought when I watched the teevee on the morning of 9/11/2001, that the right-wingers’s Afghanistan scheme had backfired on them harder than anything before in American history. That when people recalled Ronald Reagan declaring that the Islamic fundamentalists in Afghanistan were “the moral equivalents of the Founding Fathers,” and contemplated their Saudi financiers and their intimate relations with the leading families of the American far-right, the right-wing would be properly held responsible and discredited for decades.

To my disgust, not only had the air-headed American public completely forgotten how the CIA had set up those God-lovers in Afghanistan as a weapon against leftism in the first place, and the American news media strenuously refrained from reminding anyone about this fact, but also right-wing politicians and propagandists, eager to point the blame elsewhere, started linking us atheistic leftists with Saudi-led and ISI-funded Wahhabi fanatics within 48 hours. The American public totally bought into that absurd lie, too. It’s very depressing; with such bold, conscience-free propagandists, and such stupid, stupid voters, how can American democracy possibly survive?

The filthy lie that I have ever been ideological kin to or sympathetic toward that swine bin Laden infuriates me to this day. I know it was only words, but just thinking about it even now has me grinding my teeth and clenching my fist. Someday I would indeed like to meet this Andrew Sullivan at arm’s length.


Redwolf 07.03.05 at 11:20 pm

Why exactly do we argue with Sullivan? His voice exists only in our sound chamber. Let him hear back from his readers in the religious right.


jet 07.04.05 at 5:03 pm

Anyone know of a website that will allow you to put in a city and state and get back a county? I have a complete database of all the US killed in Iraq and their hometowns and states, but I need to get their counties in order to find out if they came from predominently Republican or Democrat counties.


jet 07.04.05 at 5:06 pm

Never mind, the US Cenusus Bureau to the rescue.


Ray 07.05.05 at 7:42 am

What kind of conclusion do you expect to be able to draw from your results? If you want to work out how Democrats and Republicans responded to 11/9, how likely members of each group were to sign up to defend their country, then you need a different set of figures. You need to check the change in enlistment rates in the various counties after 11/9.
The low ratio of soldiers killed to soldiers serving means you’ll get a lot of random noise in your figures anyway. (Injury figures would be better)
And there’s probably a positive feedback loop between ‘voting Republican’ and ‘serving in the army’, which would give you a high correlation between red states and casualty figures _in any year or conflict_, which you’d probably want to disentangle from the Iraq figures.


jet 07.05.05 at 9:11 am

I said my database was a list of those killed, but it is actually a full list of all casualties (updated to about 1 week ago). It seems a little dishonorable to try to see if Republicans tend more towards combat and Democrats away from combat, or vice versa. And once I get the data, I’m not sure what it would mean. To make sense of it, I’d need some way to tell what percentage of the military was Republican and Democrat, which I’m not sure it is possible to get accurate enough numbers to make up for the casualty numbers being such a small sample. So while it seems pointless I’ll have created an interesting data set that might have some future value. And it is one more list where people look at the names of the dead making sure they are not forgotten.


Ray 07.05.05 at 9:26 am

Fair enough.


jlw 07.05.05 at 10:33 am

Let me just add that I think this whole exercise–starting of course with the idiotic Red/Blue articles written by David Brooks shortly after the 2000 election–is reduction taking to its most excretible extreme.

We are where we live? It’s not even that precise, this b.s. It’s more along the lines of We are what at least a slim majority of where we live vote for. Did the godless heathens of Iowa and the steely eyed patriots of New Hampshire really exchange identities between 2000 and 2004? Do the Hassidim of Williamsburgh count twoard the decadent left?

And when does the mark get made? Am I the Bircher/Klansman rural area I grew up in? Or the small-town pricy private college I attended (on scholarship)? Or the blue state indigo borough where I live now? If I perform some great act of valor, who gets the credit? If a turn traitor, which party gets the blame?

No offense, but there were likely 1,700 reasons why those killed in action joined to serve in the military. And reducing them to a dot on the map or the political affiliation of 55 percent of the people in the place where they spend some of their childhoods is ghastly.

As is the attempts by people like Sullivan to slice the nation into Republicans and traitors.


jet 07.05.05 at 12:45 pm

I’m completely with you on that (including the part about Sullivan being over the top). I knew that the data I was putting together would be useless for describing the motives, lives, or politics of those who were harmed or killed in the military. But it might somehow be relevant to the communities from which those soldiers belonged.

I’m sure this thread will be long gone by the time I get the data together since I’m being nice to the census bureau and pacing my requests for each county. I didn’t look but there are probably a couple thousand unique city/state combos for all the casulties which means a couple thousand hits to the .gov site (at the medium pace of 5 seconds per hit). I’ll check the archive of this when I’m done to see if anyone cared about the results.

Comments on this entry are closed.