Beyond red and blue

ONE OF THE MOST awful prospects of the next presidential election is the return of . . . that damn map.

Depicting the results of the 2000 election, the reigning graphic of American politics divides the United States into two colors, red for Republican and blue for Democratic. It is also the basis of a lot of simplistic political analysis. "The 2000 election map highlighted a deep cultural tension between the cities (the blue states) and the sticks (the red states)," Matt Bai wrote in The New York Times Magazine earlier this year. Or, as David Brooks put it in The Atlantic Monthly in 2001, "In Red America churches are everywhere. In Blue America Thai restaurants are everywhere."

But this primary-color collage resonates only because it turns up the contrast. Given that more than 40 percent of voters in the blue states backed Bush and more than 40 percent of voters in the red states backed Gore, doesn't the red vs. blue model seem, well, a bit black-and-white?

So, working under the auspices of CommonWealth Magazine, I decided to make a map of my own. Aiming somewhere between the reductionist red-and-blue model and the most accurate (but least useful) subdivision of the United States into infinity, I split the country into 10 regions, each with a distinct political character. These 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. Each represents about one-tenth of the national electorate, casting between 10.4 million and 10.8 million votes in the 2000 presidential election.

Some states fall entirely within a region, but many are split between two or more. Electoral votes follow state boundaries, but populations don't, and the social characteristics that influence politics spill over jurisdictional lines. Rural sections of adjacent states often have more in common, culturally and politically, with each other than with the urban and suburban population centers of their states. If political campaigns can translate media markets into electoral votes, why not regional identities that cross state lines? Furthermore, upstate-downstate divisions are well-established dynamics in elections for statewide offices. Why should it be a surprise that they play a role in the Electoral College tally for president?

No winner of a presidential election has carried fewer than five regions in at least three decades. In the razor-close 2000 presidential election, George W. Bush and Al Gore each won five regions, but it was Bush's hair's-breadth victory in the region I call Southern Lowlands -- which stretches from Maryland past the banks of the Mississippi in Louisiana -- that carried the day.

Although the purpose of the map is not prediction, its explanatory power is evident: If either Bush or the eventual Democratic nominee in 2004 can carry a sixth region, as Bill Clinton did in both 1992 and `96, he is virtually assured to win in November. As political campaigns pull out their maps and sharpen their pencils, setting a course for Nov. 2, they should consult the new cartography -- if only to determine where their opportunities lie, and where they are wasting their time.

Three of the regions -- Sagebrush, Southern Comfort, and Farm Belt -- have voted Republican in every election since 1964. Two others -- Appalachia and Southern Lowlands -- lean Republican, but have boosted Democrats from time to time. Appalachia, which follows the mountain range from Pennsylvania to Mississippi, supported Jimmy Carter in 1976 but abandoned him in `80 and has backed the GOP ever since. Southern Lowlands stayed with Carter in `80 and supported Clinton twice in the `90s but rejected Northerners Walter Mondale and Michael Dukakis, not to mention Gore in `00.

Three regions -- Upper Coasts (which is anchored by Boston in the East and San Francisco in the West), Great Lakes, and Big River -- have flip-flopped in a dramatic way, voting for Carter in `76, switching to Reagan in `80 and `84, then going Democratic in the past four elections. Northeast Corridor, which runs from Bridgeport to Bethesda, followed the same course, except that it snubbed Dukakis and waited until `92 to switch back to the Democrats -- and stayed there. Finally, El Norte, which stretches from Los Angeles to Brownsville, Texas, and also includes the Miami area, backed Republican candidates from `68 through `88 but more recently supported Clinton and Gore.

Move over, soccer moms

Of course, the 10-region model is not the only way to analyze national politics. The pundit class is especially fond of those based on demographic groups, whether clearly defined (such as college graduates and Latino voters) or rather squishy (such as Soccer Moms and Office Park Dads).

But in comparison to the others, the 10-region model has certain advantages. First, it is based on election returns, not on polls or focus groups or sociological speculation. Indeed, a geographic analysis shows that a supposed "swing" group like Soccer Moms is so large and ill-defined to be meaningless: The results from 2000 strongly suggest that Soccer Moms in the Southern Comfort region simply cancelled out their counterparts in the Upper Coasts, with the former supporting Bush and the latter voting strongly for Gore.

Second, because the data is based on geography rather than on vague demographic groups, the 10 regions can be studied over time, and redrawn to reflect changes in voting patterns -- in the same way that satellite weather maps track approaching storms.

Finally, from a strategic perspective, the 10-region model shows each party where it can (or has to) win in order to take the presidency. In `00, both parties won four regions by solid margins; no amount of politicking was going to change that. But four isn't enough to win, no matter how much you pump up the vote in your base. As we saw that year, even winning the popular vote doesn't do the trick if you can't nail down a fifth region, and a sixth is better still. Candidates can rack up votes -- popular and electoral -- by driving up turnout in the regions that like them best; tipping the outcome in states divided by region; or by expansionism, spreading the influence of friendly regions into adjacent territory.

But the real beauty of the 10-region map is that it gets beyond red-vs.-blue reductionism, introducing shades of purple. The American electorate is a big, variegated mass of humanity, and a small shift of votes in the right place can swing an election.

A Strategy for both parties

From the 10-region perspective, Bush has a strategic decision to make in 2004. He could try squeezing more votes out of the four regions he won handily in `00 (Appalachia, the Farm Belt, Sagebrush, Southern Comfort) and solidifying his position in Southern Lowlands, which he carried by fewer than 100,000 votes. That strategy would preserve his slight advantage in the Electoral College, but it might result in another popular-vote loss.

His best chance to capture a sixth region is Big River, the most closely divided in the last election. Iowa, Minnesota, and Wisconsin will surely be on Bush's hit list next November. But he'll also likely target states where the Upper Coasts and Sagebrush regions combine to produce closely divided electorates, such as Oregon and Washington.

For the Democrats, the El Norte region offers the best chance to pick up more electoral votes. It's the only region carried by Gore in `00 that has above-average population growth, meaning that the Democrats may benefit from get-out-the-vote efforts in states carried by Bush last time -- notably Arizona, Colorado, and Florida.

If the Democrats can carry a sixth region, it will almost certainly be the Southern Lowlands, but the distribution of electoral votes works against the party. Democrats do extremely well in counties with large black populations, but these strongholds are divided among 10 states, and they don't come close to a statewide majority in any of them. Still, gubernatorial victories in Louisiana and Virginia offer some hope that this year's nominee, like Clinton in the `90s, can carry a couple of states in the region.

Bleeding colors

In one respect, the red-vs.-blue model makes an important point about changes in national politics over the past 30 years. Though we think of the `00 election as being the closest in memory, many more states were competitive in the `76 contest between Jimmy Carter and Gerald Ford. That's reflected in the 10-region model: In `76, only one region was won by more than 10 points (Southern Lowlands, which Carter carried 57-42), but in `00 eight regions were won by double-digit margins.

One reason for the closeness of so many states and counties in `76 was that both Carter and Ford came from the moderate (some would say "electable") wings of their respective parties. The differences between the Democratic and Republican nominees have become sharper in just about every election since, particularly on social issues such as abortion. And regardless of how many voters tell pollsters that they're "independent," Americans are increasingly likely to engage in straight-ticket voting. Some analysts believe that the two major parties will only encourage this habit in `04, moving away from their longtime emphasis on "swing voters."

But even if the trend toward partisanship feeds the perception of the United States as a 50/50, red-blue nation, it's unwise to assume that this is a permanent condition. American voters also have a habit of rebelling against one-party states. In 2002, on a district-by-district basis, the results of congressional elections were eerily close to the results of the `00 presidential race. But in governor's races, where candidates are less closely identified with their national parties, several states rejected their "red" or "blue" labels. Support for third-party presidential candidates in `92 and `96 also indicates some discomfort with the limited palette.

And in the long run, the colors on the red-vs.-blue map may start to bleed. How red are the John McCain voters who supported Bush in 2000? Does the popularity of moderate Republicans such as Rudolph Giuliani in true-blue New York mean that Northeast Corridor could again become part of the GOP base? And does the success of rural-oriented Democrats such as Governor Mark Warner in Virginia mean that the South's Dixiecrats can someday rise again?

Maybe the good-vs.-evil nature of the war against terrorism has made it too easy to fit domestic politics into a similar kind of dichotomy. But the red-vs.-blue model papers over too many real differences in the national electorate. The party that understands this may gain an advantage in 2004 -- but it can't expect to maintain this advantage for long.

Robert David Sullivan is an associate editor of CommonWealth Magazine. A complete version of this article can be read at www.massinc.org.