It's the unemployment rate, stupid. After all, no president since Franklin D. Roosevelt has been reelected when that number was above 7.2 percent.

No, it's the average inflation rate for the 15 quarters before the election, as measured by the "GDP deflator" (an index of the relative purchasing power of a dollar over time).

Or maybe it comes down to growth in the gross domestic product, particularly in the third quarter before the polls open; per capita disposable income; or the president's Gallup Poll approval rating.

By several of these traditional predictors, President Obama is teetering on the edge of defeat.

Consider: Unemployment stood at 9.2 percent in June and is projected to improve to no better than about 8 percent by November 2012. GDP is anemic, remaining flat in 2008, falling 2.6 percent in 2009, and rising just 2.9 percent last year. (So far in 2011, it's up 1.3 percent.)

Oh, and Obama's daily Gallup approval ratings averaged 43 percent for the week of July 18, the lowest weekly average of his presidency. Anything below 50 is generally trouble for an incumbent.

For years, political scientists have maintained that accurate forecasts of presidential elections are possible: Just plug in the right economic variables, and, presto, you have your likely result.

Some of the models they've built are elegant machines for teasing order from chaos that have decent records in predicting whether a president or the incumbent party will keep the White House. But there are always asterisks. Life is messy, and so are election campaigns.

The latest X-factor is the debate over raising the federal debt limit. Nobody can tell what effect it will have on the politics of 2012. Obama could end up looking like the mature adult compared with House Republicans, or voters might conclude he is not serious about restraining government spending. If the mess damages the U.S. credit rating, the damage to the already fragile economy could be decisive.

This is not an exact science.

Roosevelt won a second term in 1936 with the unemployment rate an estimated 17 percent, but that was a significant improvement from the roughly 24 percent rate under President Herbert Hoover in 1932. People felt there at least had been some progress. Incumbent parties lost the White House in five of the last 13 elections when unemployment was relatively high: 1960, 1976, 1980, 1992, and 2008.

Unemployment was 7.2 percent in 1984 when President Ronald Reagan won reelection in a landslide, but that was down from 10.8 percent in December 1982.

If unemployment is such a great predictor, then Democratic Vice President Hubert Humphrey should have won in 1968, when the jobless rate was a puny 3.4 percent. But the Vietnam War was the dominant issue then, along with a law-and-order backlash against the counterculture, and Republican Richard Nixon won the presidency.

In 2000, Democratic Vice President Al Gore had a 3.9 percent unemployment rate and good economic growth behind him, but government ethics and the tawdry scandals of the Clinton administration were arguably more important to voters.

Economic growth, measured by the gross domestic product, should be helpful in predicting an election. Political scientist Larry Sabato of the University of Virginia recently wrote about how a party's chance of holding the White House has correlated with the third-quarter GDP number in election years. It worked roughly half the time, but there were some big flaws.

Third-quarter GDP growth in 1980 was higher than it had been before the previous 13 presidential elections, 8.6 percent, yet President Jimmy Carter lost in a landslide. High inflation and interest rates, the Iranian hostage mess, and a general unease about the future apparently outweighed the good news.

A robust economy - GDP was growing 7.6 percent in the third quarter of 1976 - could not help President Gerald R. Ford overcome the sour aftermath of Watergate and his controversial pardon of Nixon.

Political statistician Nate Silver recently crunched presidential returns and economic data, concluding that GDP minus inflation explained 43 percent of the popular vote results from 1912 through 2008.

That leaves a lot of room for other factors to influence presidential elections, including unemployment, the percentage of voters who feel the country is on the right track, shifts in partisan leanings among the electorate, and controversial legislation, such as the health-insurance overhaul.

Then there's the opposition. What if a third, tea-party-oriented candidate got in the race, splitting the anti-Obama vote with the Republicans? Even without that unlikely scenario, the GOP could nominate a hard-liner who scares off independents - Michele Bachmann, maybe. (Democratic strategists sure hope so.)

Stay tuned.

Contact politics writer Thomas Fitzgerald at 215-854-2718,, or @tomfitzgerald on Twitter. Read his blog, "The Big Tent," at