Demographic and national data are far better for predicting the race than scrutinizing unreliable state-level polling.

hayestilden.banner.jpg
In 1876, Samuel Tilden (left) won the popular vote but lost the presidency to Rutherford B. Hayes. (Wikimedia Commons / Library of Congress)

It is a source of constant amusement to me that so many people obsess -- as if fiddling with a Rubik's Cube -- over the various combinations of states that could get either President Obama or Mitt Romney to the magic number of 270 votes in the Electoral College. The guilty include pros at both ends of the political spectrum; people who ought to know better; and armchair analysts who seem to think that they can crack the magic code.

The simple fact is that our nation has had 56 presidential elections. In 53 of them (94.6 percent), the winner in the Electoral College also happened to be the one with the most popular votes. Of course, we all recall the 2000 presidential election in which Al Gore prevailed in the popular vote, while George W. Bush was the Electoral College victor. But few remember that the previous divergence between popular and electoral votes was 112 years earlier, in 1888. Incumbent Grover Cleveland was the popular-vote winner. However, challenger Benjamin Harrison carried the Electoral College and was declared the winner. The only other time that happened was in 1876, when Samuel Tilden prevailed in the popular vote but Rutherford B. Hayes won the Electoral College vote and the presidency.

Two points are worth keeping in mind: First, the chances are about 94.6 percent that the same person will win both the electoral and popular votes. So people are expending a lot of time and energy trying to figure out something that has about a 1-in-20 chance of happening.

But second, close races are just that: close races. Notwithstanding the almost daily emails that I get protesting that the election will be a slam dunk for either Obama or Romney, this race will be close. Obama has a high floor, meaning that he has a fervent base of support. He also has a low ceiling, meaning that he has large and adamant opposition. That high floor has prevented him from descending to the depths of low job-approval ratings that afflicted many of his predecessors. Obama's lowest Gallup job-approval rating was 40 percent, compared with George W. Bush's low point of 28 percent; Jimmy Carter's 29 percent; George H.W. Bush's 32 percent; Ronald Reagan's 37 percent; and Bill Clinton's 37 percent.

If a race is close nationally, it will be close in a lot of individual states, too. Given the dubious quality of most of the publicly available state polling, few of these two-bit surveys could offer unique insight into who is likely to win close contests -- because, well, they're close.

Of course, none of this cautionary tale about polls applies to the teams at the Obama headquarters in Chicago and the Romney headquarters in Boston. They are paid to look at things very closely, and they have the political equivalent of electron microscopes to take that look. The campaigns are each spending millions of dollars on extremely sophisticated survey research in each of the swing states.

They also have access to a considerable trove of other polling commissioned by Senate and gubernatorial candidates, state parties, and, in a few places perhaps, groups focused on a referendum or two. The Obama and Romney camps have tools for close examination unavailable to the news media or individuals. Comparing the caliber of state-level survey research that the presidential candidates have to what the average political aficionado has is like comparing the new Boeing 787 to a World War II-vintage DC-3.

All of this time and effort spent parsing state-level polls would be better spent more closely examining the national polling data, particularly looking at how the candidates are performing now compared with Obama and John McCain in 2008, and examining how likely the members of specific (and potentially decisive) demographic groups are to actually vote.

We are awash in high-quality data: The Pew Research Center released a large survey last week. Every Tuesday afternoon, Gallup publishes the detailed demographic breakouts for its three-week moving average -- usually about 9,000-registered-voter tracking surveys. And a new NBC News/Wall Street Journal poll was just released on Tuesday night.

The Gallup three-week moving average, among 9,362 registered voters, shows the race effectively tied: Romney ahead by 1 point, 46 percent to 45 percent, with 86 percent of Republicans saying they would definitely vote compared with 80 percent of Democrats saying so. Like before, African-Americans look very likely to vote. The younger and Hispanic voters, even after the president's announcement last week about leniency for students facing deportation proceedings, do not look as likely to do so.

In the just-released NBC/WSJ poll by Democrat Peter Hart and Republican Bill McInturff, Obama had a 3-point lead among all registered voters, 47 percent to 44 percent. Obama voters were more enthusiastic about their candidate. But key demographic groups for Romney looked more likely to vote than the ones likely to go strongly for the president. People who want to look into the weeds are better off looking into those weeds than into the swamp of mostly mediocre state polls.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.