Saturday, 18 November 2006

framing the election

One more in my continuing series, "Avoiding Major Papers Through Rants." As usual, apologies.

How should we understand last Tuesday's election results?

Of course, there's Red State, Blue State, One State, Two State. Besides 1984, when Reagan painted the map blue (yes, the colors were reversed--Mondale turned his home state of Minnesota Democratic red), a dichotomy is the opposite of informative if you're trying to understand anything about the election. Even USA Today is more informative than the Red-Blue map. (Okay, maybe not. In fact... note to self: write editor of USA Today to suggest that, next election, they send out the paper with a large uncolored pullout map--with state names printed in full, none of those confusing abbreviations--and red and blue crayons.)

Back in January 2004, Robert David Sullivan offered an alternative to the red-blue commonplace. Instead of two colors correlating with the two parties, he has 10 colors correlating with 10 regional voting blocs of equal population. "The regions are based on voting returns from both national and state elections, demographic data from the US Census, and certain geographic features such as mountain ranges and coastlines." The added granularity alone is a big help in showing some of the relevant patterns in recent elections (in particular highlighting the deep divisions inside some states--note that Ohio, Kentucky, Florida, Texas, and California all have regions of three of these divisions within them.

What do the regions "mean"? Sullivan doesn't really explain much about the factors he weighed or how he drew the lines. Here's a synopsis of what he does say (approximately left to right across the map):

Upper Coasts - high tech, independent progressives
Sagebrush - rural libertarians
Farm Belt - white farmers with high school educations
El Norte - young and Latino
Southern Comfort - Republican heaven
Big River - always highly contested, high school educated
Great Lakes - urban industrial centers versus suburbanites
Appalachia - poor Republican dirt farmers and coal miners
Southern Lowlands - African American Democrats and white Republicans duke it out in this region every election.
Northeast Corridor - big city liberal intellectuals.

I'll come back to these divisions in a moment, but first I'd like to show that they remain somewhat useful. Take a look at this county-level map of the 2004 election (scroll down 2/3 of the entry):

I like this map a lot more than red-blue, because it has a LOT of granularity to it. It's still somewhat unrepresentative since New York City is a tiny speck even though it has the population of that whole big red region in the middle there, but at least it shows purple gradations. But other than the gestalt effect, which I like, there's really too much information on this map to be useful, and that information is encoded in a way that makes it very hard to analyze in any meaningful way (at least, in the amount of time I intend to spend on it).

Now compare Sullivan's geo-demo-elector-graphic pre-election map with the post-election county map. Some similarities should pop out.

It's obvious that Upper Coasts and El Norte are Kerry-country. Sagebrush (other than the New England portion--hooray us!) went Bush, along with Southern Comfort and Appalachia. In some sense, we can conclude that Sullivan's map is successful in identifying regional differences in voting outcomes, and in some sense this might be considered "explanatory" (i.e., an instrumental or utilitarian sense). But it is completely unsatisfying if we seek to understand the differences. At the very least, we would need to spend some time constructing Rudyard Kipling style "just so" stories to describe the relationship between the geographic, demographic, and election history factors Sullivan used in creating the map.

Some of the divisions are immediately sensible, like dividing along geographic boundaries rather than political ones. Farmers on opposite sides of a state line are still more likely to agree with one another than with industrialists in their own states. But which geographical boundaries do you choose? Which ones are relevant to elections? How is it that Sullivan decided to make a Latino region? Don't pretend you didn't notice--he called it El Norte. There's no geographic component to this other than vicinity. Sullivan isn't dividing mountains from plains here, he's noticed that regions of the country with a heavy Latino representation vote as a bloc (or at least, the regions indicated do). There's no explanation for this, though--it's just a statement that if enough people in a region check the "hispanic" box on the census, then that region will likely follow a certain voting pattern. Is this what passes for explanation? (Granted, Sullivan never claims to be explaining anything, so this shouldn't be seen as a criticism of his project.)

Perhaps the trouble here is that I'm seeking an explanation where no explanation is to be had--or where any given explanation is just as satisfactory as the next. The trite, five-second post-election "analysis" says Tuesday's vote was a referendum on the war. Maybe it was. But maybe this is the same kind of analysis that accompanies the daily financial report: "The Dow was down 24 points today on news that ACME will lay off 2,500" Really? Is that the reason? Can you show me a causal map? No, of course not. To do so would be to point out how absurd the stock market is. It's not news about ACME that's important. It's how people react to news about ACME. More precisely, it's about how traders think people are going to react to news about ACME. Actually, that's not quite it either, it's really more about how traders think other traders think people are going to react to news about ACME. What's frightening is that these imaginary reasons for doing things become real when events unfold in just the way we would predict they would if those reasons obtained. This is, unfortunately, not a simple case of ad hoc, post hoc rationalization. It's a case of reification.*

If I leak news about ACME corporation to a bunch of stock traders, and they start buying or selling ACME stock on the basis of what they think OTHER stock traders are going to do when they hear the same news, surely the news really is the cause of what happens to ACME stock.

And surely, if I say that the election was a referendum on the war, and politicians start acting like the election was a referendum on the war, then the election really was a referendum on the war. Or it may as well have been.

Do election results have anything to do with, say, environmental concerns? The minimum wage? Social security? Universal healthcare? Stem cells? Congressional pages? Inheritance taxes? No. It's all about the war.

* In this analysis, I rely on ideas from Ian Hacking's 1982 essay "Experimentation and Scientific Realism." In it, he writes, "we are completely convinced of the reality of electrons when we regularly set out to build--and often enough succeed in building--new kinds of devices that use various well understood causal properties of electrons to interfere in other more hypothetical parts of nature." Or, if you want the shorter version: "If you can spray it, it's real." I mention this to give credit where due, but also because it's probably the first time anyone has ever supported an argument about politics with evidence from the philosophy of science.

No comments: