November 04, 2004

Purple Haze: The Real Vote Map

Via Crooked Timber: the real vote map:

Crooked Timber: Red Counties, Blue Counties and Occupied Counties : Robert Vanderbei’s “Purple America” Vote Data Map:

Bigger version at Robert's website.

Posted by DeLong at November 4, 2004 12:54 PM | TrackBack

Posted by: mlhm5 at November 4, 2004 01:00 PM


Posted by: mlhm5 at November 4, 2004 01:01 PM

Yet another reason to take down the electoral college?

Posted by: Jean-Philippe Stijns at November 4, 2004 01:17 PM

Where are the stupid voters? Guess...

Posted by: M at November 4, 2004 01:18 PM

Where are the stupid voters? Guess...

Posted by: M at November 4, 2004 01:20 PM

Where are the stupid voters? Guess...

Posted by: M at November 4, 2004 01:21 PM

I wish Democrats would communicate better. Is there
no-one who has read Tufte?

The power of the visual imagery hides the lie in this

Why cant a Democrat produce and circulate a true map?
Code this with color-intensity based on population?
Much better than that, based on dots, one for every
10000 votes? The key is to do it without distorting
the shape of the map (which does,
thereby diluting the power of the image)


Posted by: J at November 4, 2004 01:24 PM


Another reason to take down the EC? Please explain.

(In the interest of full disclosure, I should let you know that I voted for Gore in 2000 and Kerry in 2004. And I actually like the EC a whole lot.)

Posted by: JR at November 4, 2004 01:30 PM

That's da debbil's territory, that burnin' red. We need to throw some nice cool blue water over them.
And J is entirely right about communication (not to forget substance).

Posted by: PW at November 4, 2004 01:34 PM

Adopt a Republican:

Dems everywhere - identify a repub in your neighborhood and befriend him, teach him to read, teach him to analyze and teach him logic - in short adopt him and raise him like your own.

Then maybe, just maybe, we will see the hue go blue.

The Perlustrator

Posted by: The Perlustrator at November 4, 2004 01:42 PM

Joe - You can't use Eddie T. when you're only PhotoShopping something done by someone at USelessAToday.

JR - The best argument against the electoral college is that it allows the candidate to ignore the plurality of the voters (the cities and the major suburbs), concentrate only on issues that impact the exurbs and rural areas.

This leads to economic inefficiencies (e.g., Homeland Security budgets that consider Carhenge as much of a target as "Democracy Plaza") and especially excessive subsidization. (The 13 cents extra per dollar in taxes the blue states pay goes to "maintaining" infrastucture that is not in need of repair [e.g., sparsely-travelled highways between Denver and Tempe].)

In short, the EC encourages poor resource allocation.

Posted by: Ken Houghton at November 4, 2004 01:48 PM

But, Brad, so what?
Yes it shows that sane Americans and insane Americans live cheek-by-jowel across the country. BUT, BUT, BUT --- the US elects based on FPTP, not PR, so it doesn't freaking matter. And gerrymandering doesn't help --- yes it means there are some districts that are pre-designated as Dem rather than Repub, so Dems get into congress; but it also means that (on both Dem and Repub sides) the real election is in the primaries, not election day, and extremists on both sides frequently have the advantage in primaries.

This election seems a perfect example of what political scientists have said for years --- that there are real problems inherent in the US system, based on both an elected president and FPTP.

Posted by: Maynard Handley at November 4, 2004 01:52 PM


Fair enough, but that doesn't have much to do with the purple map. In fact, the standard red/blue dichotomy is better suited to the point you're trying to make.

Incidentally, I find the resource distribution argument more compelling than the standard cries of "one-man-one-vote". Resource allocation, however, has more to do with congressional logrolling than with presidential campaigning. The nitty gritty distrobution issues are handled on the Hill.

And presidential candidates have other, better ways to deliver the rural vote. (1) Appeal to "values." (2) Make unrealistic promises.

Posted by: JR at November 4, 2004 02:07 PM

I presume this was inspired by the map on Pat Sullivan's site proving that Bush really won by a landslide despite the fact that he only got 51%, because he carried a lot of big rural counties with few people in them? (Well, Pat had to justify his earlier confident prediction that Kerry would lose by a landslide SOMEHOW. I can't resist mentioning, though, that he's just accused me of "lacking minimal connection with reality".)

Posted by: Bruce Moomaw at November 4, 2004 02:12 PM

"Another reason to take down the EC? Please explain."

JR: One thing I really dislike about the electoral college is how is forces liberals to put forward a pseudo-conservative electoral program. (Rovian conservatives don't seemed binded by the reverse constrain anymore...) Liberal presidential candidates are incapable of tapping into Southern liberal votes since they generally can't tip the balance in the South anyway.

I was just wondering what would be the effect on political discourse, the balance of power in the presidential elections, and ultimately policy, if electoral college votes were to be allocated proportionally to votes gathered by each candidate in each state. This is not the same as getting rid of the electoral college alltogether but I think that it would at least give back to voters in all states the ability to influence the election.

In any case, I formulated this remark as a question, because I don't know enough to answer these questions myself.

Posted by: Jean-Philippe Stijns at November 4, 2004 02:26 PM

Could someone *please* post the same map with percentage changes from 2000 to 2004? Or point me to the raw data? Thanks!

Posted by: DaveV at November 4, 2004 02:38 PM

OK, so let's CAN the big purple map on basic psychophysics grounds, shall we? The way to do the map you want is to do the national map the way CNN does the state/county maps: shades of red and blue proportional to the difference between Democratic and Republican votes, converging on white or grey. This would show you that many counties are in fact not very biased one way or another without making you have to discriminate shades of purple, which is a Big Lose for communication value.

The reader above who suggested the "dots" approach to map coloring has another valid point, except that it should be noted that the differences in population density in the US are large enough to make this quite difficult to pull of in practice; the dots get way too crowded in the high density areas if you have any dots at all in the hinterlands.

Posted by: Jonathan King at November 4, 2004 02:43 PM

M writes:
> Where are the stupid voters? Guess...

OK, so let's not do it quite this way, shall we? I think a far more obvious approach is to plot something less incendiary like proportion of residents over 25 who have a BA or better. And somebody has done this one, too, for 2000:

Note that this leaves out the middle (which is unfortunate). But also note that we flipped the top 10 state New Hampshire this time. If I had time (and I don't) I'd regress Bush/Kerry proportions against the census county data for degrees, and presumably show (for some correctly transformed variables) a shockingly good correlation. But then I could also do this for population density, and a lot of other similarly correlated variables.

Posted by: Jonathan King at November 4, 2004 02:55 PM

Does anyone have any idea what the deep blue county in the middle of Idaho is, and what it's doing there? It's not Ada county, where Boise is, and it's not w/ the University of Idaho or Idaho State is- I'm from Idaho and have no idea why this particular county should have such a deep blue- is it an Indian reservation? If anyone knows, please let me know.

Posted by: Matt at November 4, 2004 02:55 PM

re: Idaho

Blaine county, if that helps. 5,992 (59%) votes vs. 4,034 (40%).

Apparently Sun Valley area.

Does that help?


Posted by: fatbear at November 4, 2004 03:28 PM

Malkin's map was run back after the 2000 election. What is she advocating? One acre, one vote? OK, red state voters live on a lot of undesirable land. SO?

Posted by: pgl at November 4, 2004 03:50 PM

Maybe to show population you could construct one with relief. You can pretty much do that in your head by simply having the blue areas reach tothe sky, although few Republicans would have the perceprive powers to see it this way.

Posted by: remo williams at November 4, 2004 04:36 PM

Actually, Taki used the same technique after the 2000 election to prove that Bush was REALLY the overwhelming choice of Americans even if Gore did win the popular vote.

Posted by: Bruce Moomaw at November 4, 2004 05:02 PM

Ah- Sun Valley. I should have thought of that. Thanks, Fatbear. I wonder who Bruce Willis voted. I guess all the snow-boarding trips Kerry made there paid off. ;)

Posted by: Matt at November 4, 2004 05:44 PM

Blue: America
Red: Flyover country

Posted by: ogmb at November 4, 2004 06:28 PM

The real vote map?! What we need is an electronic paperless voting machine map.

"COLUMBUS - The head of a company vying to sell voting machines in Ohio told Republicans in a recent fund-raising letter that he is "committed to helping Ohio deliver its electoral votes to the president next year."

Walden O'Dell, chief executive of Diebold Inc. attended a strategy pow-wow with wealthy Bush benefactors - known as Rangers and Pioneers - at the president's Crawford, Texas, ranch earlier this month.

Cleveland Plain Dealer - August 28, 2003

"Diebold and ES&S, will count (using both computerized ballot scanners and touchscreen machines) about 80 percent of all votes cast in the upcoming U.S. presidential election.

Both ES&S and Diebold have been caught installing uncertified software in their machines. Even if states or counties hire their own technicians to re-program Diebold or ES&S software (or software from other companies), experts say that permanently installed software, called firmware, still resides inside both electronic scanners and touchscreen machines and is capable of manipulating votes."

Online Journal - April 28, 2004

Posted by: Dubblblind at November 4, 2004 07:22 PM

I heartily endorse this map, with one objection: no Alaska or Hawaii! What happened to them?!

Posted by: Julian Elson at November 4, 2004 07:24 PM

"But then I could also do this for population density, and a lot of other similarly correlated variables."

I correlated pop density, median income and a schooling score (%univ - % w/o high school) on the Gore vote share, and all three correlate highly ***on the state level***. I assume it's the same on the county level but if you look at per capita survey data the GOP seems to hold the advantage in income and schooling. I haven't come up with a good explanation for the discrepancy yet. Race?

Btw, I want to see the same map with lightness reflecting pop density. My guess is it will be straight blue and white.

Posted by: ogmb at November 4, 2004 08:38 PM

Postscript: Patrick Sullivan is now screaming at me that I and Josh Marshall are "dumb" because, when Bush called his victory "broad and nationwide", he wasn't really falsely claiming that his narrow victory was "big" -- he was just bragging, for some mysterious reason, that it covered most of the country's geographical area. After all, them city folk ain't REAL Americans. Living all crowded together like that -- t'ain't natural. In short, we have here still further proof that the governor is broken on Mr. Sullivan's steam engine.

Posted by: Bruce Moomaw at November 4, 2004 11:04 PM

I object. Once again, Hawai'i is left out of the map. It happens when gas prices are discussed, too. ($2.39/gal for regular today).

Posted by: Linkmeister at November 4, 2004 11:24 PM

Once more the Republicans trumpet that although the Dems have as many or more people, they have all the cows.

Posted by: rea at November 5, 2004 04:50 AM

Tracking thread here.
collection of maps.

From the map, I must conclude It is KERRY's campaign strategy that is screwed up. The map clearly indicates the countries has a lot of purple and blue area.

Posted by: aar at November 5, 2004 09:27 AM

Actually, Gallup's first post election poll ( ) is now showing some things very interesting and totally counter to everything we've been hearing about this election, although I'd need to look at the actual voting figures to confirm it. Specifically, they're saying that the urban-rural split this time was much LESS serious than it was in 2000 -- Bush gained 9% more of urban voters and 3% more of suburban ones, but LOST 6% more of rural voters than last time. They're also saying that most of Bush's gains this time were among nonwhites -- 4-5% as against only a 1-point gain among whites. If true, does this mean that Rove's anti-gay campaign was much more successful among nonwhites than among whites? (Gallup does confirm that Bush's gains since 2000 were mostly among enthusiastic church-goers -- but he also shows that they were bigger among the more educated than among the less educated! So, what: Bush was most successful among well-educated nonwhite homophobes? Or he was most successful among nonwhite homophobes and the well-educated upper class?

Meanwhile, for comedy relief, DeLong's shrillest troll is now screaming that I'm "stupid" for not realizing that Bush was obviously bragging about the fact that he won America's rural areas and saying that the fact that he lost its cities is unimportant.

Posted by: Bruce Moomaw at November 5, 2004 12:09 PM

I think that some on this comment page have fell for a well-crafted hoax.

Posted by: Fraydog at November 5, 2004 11:31 PM

To visually represent the populations of the counties, use a cartogram:

There's another type of cartogram, not shown at the site above, that renders maps in 3D so that more populous counties tower above sparser ones.

Posted by: Michael D at November 7, 2004 04:24 AM


Posted by: xenical at November 7, 2004 10:09 PM

Actually, Gallup's first post election poll ( ) is now showing some things very interesting and totally counter to everything we've been hearing about this election, although I'd need to look at the actual voting figures to confirm it.

Post-election polls are notoriously unreliable: there's a true 'bandwagon' effect. See the polls out of Florida four years ago, for instance.

Posted by: nick at November 8, 2004 12:22 AM


Posted by: levitra at November 8, 2004 04:13 AM

Speaking of cartograms, there are Election 2004 cartograms at

Yes, this is my blog page :), and could be construed as shameless self-promotion, but Robert Vanderbei linked to this page as well :)

Posted by: Suresh at November 8, 2004 07:11 PM

Here are buttons, T-shirts, mugs, and other tchotchkes based on some of the cooler-looking purple maps and cartograms of the 2004 presidential election.

Posted by: Clear Mandate? at November 12, 2004 08:14 PM