President Bush's margin last year over Sen. John Kerry of Massachusetts, 2.4 percentage points, was the smallest of any victorious presidential incumbent in history. He won a very close election.
But in most of the country, the 2004 race wasn't even close to being close. A Pew Research Center analysis shows that in the majority of the nation's 3,153 counties, the election was a landslide -- with either Mr. Bush or Mr. Kerry winning by a margin of at least 20 percentage points.
Never before in modern times has so competitive a national election been the sum of so many uncompetitive parts.
Some have dubbed this "the big sort." It's not a new political phenomenon, but it's more striking than it once was. In the two other very close presidential elections of the past half-century -- 1960 and 2000 -- fewer than half the nation's counties (44 percent and 45 percent, respectively) were decided by a margin of 20 percentage points or more. Last year, the number rose to 58 percent.
Unlike congressional districts, whose lines are redrawn every decade (often to create safe seats for incumbents), counties' geographic boundaries rarely change. So if it's not redistricting, what is it that sorts the electorate into partisan enclaves? Some leading theories:
(1) It's all about lifestyles. Everyone knows the stereotypes: Red Americans like barbecue, NASCAR and George W. Bush. Blue Americans go for Thai food, beat-up Volvos and anybody-but-W. With red and blue in such different universes, it's hardly a surprise they would gravitate toward different neighborhoods.
But wait a minute. People have liked living among people like themselves since the dawn of civilization. Why would this have more of an impact now on voting patterns?
One hypothesis is that, as the modern communications revolution has promoted ever more well-defined lifestyle groupings (each with its own cable channel), social identity plays a greater role in determining where people live as well as the candidates and parties they support.
Last year the Bush campaign did a great job organizing around affinity groups. They identified potential supporters by the magazines they read, the television shows they watch, the beer they drink -- to say nothing of the churches they attend -- and used this to support a sophisticated get-out-the-vote effort.
But when political scientists look at public opinion survey data, they consistently find what matters most in determining partisanship is not who you are but what you believe.
For example, married, religiously observant Southern white men are among the most Republican voting blocs, leaning Republican by a margin approaching 3-1. But members of this bloc with liberal policy preferences, they lean Democratic by 2-1.
The same is true for all major groups (except African-Americans): Ideology trumps social identity by decisive margins as a predictor of partisan affiliation.
(2) OK then, it's all about ideology. Republicans are more consistently conservative than they once were; Democrats more consistently liberal. Scholars who have analyzed survey data find the correlation between ideology and partisanship has increased by about one-third in the last generation.
Still, this doesn't by itself explain the growing congruence between geography and partisanship. Only in looking at the shift's regional basis does the picture come into sharper focus.
The most significant change in voting trends in the past half-century has been the realignment of the South from Democratic to Republican. White Southerners are no longer voting the way great granddaddy shot in the Civil War. They're supporting the party more in sync with their conservative values and beliefs -- and by doing so, have brought more ideological and geographic coherence to the electoral map.
(3) OK then, it's all about regions. In presidential elections, Republican totally dominate in an L-shaped swath of 26 states that runs horizontally through the South and vertically through the Great Plains and Mountain West. In 2004, President Bush carried the popular vote in this "L" by 58 percent-42 percent and the electoral votes by 232-0.
The Democrats, meantime, control the West Coast and the Northeast. In 2004, Mr. Kerry carried these regions by 55 percent-44 percent and their electoral votes by 194-5.
Yet there are limits to this rigid regional determinism. Some of the bluest states (California, New York, Massachusetts) have GOP governors, just as some of the reddest states are led by Democrats (Montana, Virginia).
And in many states, the urban-suburban-rural split is at least as important as the regional divide. In 2004, Mr. Kerry won seven states (Illinois, Michigan, Minnesota, Oregon, Pennsylvania, Washington and Wisconsin) by generating a big enough margin in the state's biggest city to overcome a sizable deficit in the rest of the state.
(4) It's all about W. There's a "perfect storm" that acts as a multiplier on all the polarizing factors listed above -- and its name is George W. Bush.
A Pew Research Center public opinion survey earlier this summer found 88 percent of Republicans approved of the job Mr. Bush is doing as president, while 76 percent of Democrats disapproved. None of the other two-term presidents of the modern era were so polarizing. The comparable figures for President Nixon at this stage of his second term were 67 percent approval from Republicans and 65 percent disapproval from Democrats; for President Reagan, 87 percent approval from Republicans and 50 percent disapproval from Democrats; for President Clinton, 84 percent approval from Democrats and 58 percent disapproval from Republicans.
Presidents who win second terms usually in some aspect of their leadership style or policy agenda cut across cultural, ideological or regional divisions. Mr. Bush hasn't governed that way. He has staked out a conservative posture on virtually all major domestic and foreign-policy issues.
Pew's polling last year found Mr. Bush's assertive leadership in the war on terror the single most important reason for his victory in a high-stakes, high-turnout election. But the polling also found that Mr. Bush's going to war in Iraq made red places redder and blue places bluer.
Ten months later, as public support for the war recedes across the board, the county is still where it was last Nov. 2: a portrait in red and blue, led by a divider, not a uniter.
Paul Taylor is vice president of the Pew Research Center and editor of "Mapping the Political Landscape: 2005," from which this commentary is drawn. The book can be read at pewresearch.org.