(FORTUNE Magazine) – It is just after 7 a.m. on a Tuesday in the middle of August. Retired high school science teacher Harold McGuire is sitting at a long table surrounded by men he has known for decades. His shirt pocket is full of Kerry-Edwards bumper stickers.

"Hey, Rodney, you want one of these, don't you?" he says, offering a sticker to dentist Rodney Beard.

"Yeah, I want one to start a fire," Beard retorts. "We're all Bushwhackers in here, except the ones who voted for gay marriage."

McGuire starts to protest, then reconsiders. "I guess I did vote for it."

From across the table, retired trucker Frank Roth chimes in, "I voted for it because it's nobody's business."

So goes another morning at the Trails Inn Cafe in Liberty, Mo. Since the 1960s a motley assortment of locals has been gathering there every weekday between 6:30 and 7:30 to participate in what they call the Clay County Coffee Club and Debating Society. As one would expect in a town where voters have gone with the winner in every presidential race since 1960--and a county Al Gore won by a single vote in 2000--opinions at the breakfast table are sharply divided on matters concerning George Bush and John Kerry. The debate over Kerry's war record, Bush's draft avoidance, and the troubles in Iraq can get mighty heated.

One subject that does not stir passions this year is the economy. Even after I ask about it, no one can muster up any steam one way or the other. That may be because business is pretty good in Clay County, a land of factories, office parks, and tract houses across the Missouri River from downtown Kansas City. But then, by most measures, business is pretty good in the U.S. as well. This year, perhaps, it isn't the economy.

The idea that the state of the economy decides elections goes way back--at least to Franklin Roosevelt's victory over incumbent Herbert Hoover in the Depression year of 1932. The most famous distillations of it, though, are relatively recent: It was in 1980 that candidate Ronald Reagan looked into the television camera at the end of a debate with incumbent Jimmy Carter and asked, "Are you better off than you were four years ago?" And it was 1992 when campaign manager James Carville posted that sign in Bill Clinton's Arkansas headquarters that read, IT'S THE ECONOMY, STUPID!

Economic issues certainly did matter in those years. In 1980 a brief but sharp election-year recession induced by Federal Reserve interest-rate hikes lost votes for Jimmy Carter. In 1992 the recession was already more than a year in the past, but Americans remained gloomy about the economy's prospects and unconvinced that incumbent George Bush could do anything to improve them.

This year another George Bush is trying to hold on to office. He's having to contend with a recession in the recent past, Fed rate hikes in the present, and an atmosphere of lingering pessimism despite mostly positive economic data. Which raises an important question: How bad does the economy have to get to hurt a sitting President? Or to put it differently, How good does it have to get to help him?

It was in the economically troubled 1970s that political scientists first established a statistical link between economic performance and the reelection odds of incumbents. In the 1980s some began using economic data to predict election results--with reasonable success in 1984, when strong economic growth helped Reagan win reelection, and in 1988, when mostly benign conditions provided a tailwind for Vice President George Bush.

In 1992, Bush was thrown out of office even though GDP was growing and the unemployment rate falling, leading one political scientist--Marc J. Hetherington of Bowdoin College--to theorize that negative media coverage of the economy had driven Bush from office. Hetherington found that voters who regularly read newspapers and watched TV news were more likely to have a negative view of the economy than those who didn't. Maybe so, but the problem wasn't all bad press: Unemployment was high and didn't fall much in 1992, and new jobs were scarce (see chart).

In 1996 the economy was indisputably strong, and incumbent Bill Clinton won easily. In 2000 every economic indicator was again flashing positive. At the American Political Science Association (APSA) annual meeting in September of that year, the forecasting models wheeled out by the professors all forecast an easy Gore victory. They weren't completely wrong--Gore did win the popular vote, and the models generally had nothing to say about the electoral college breakdown--but the result, which was much closer than expected, was a blow for the economics-is-political-destiny approach.

Where the forecasting models have so far failed is in sorting out why the economy matters more in some elections than in others. Voters don't just automatically respond to economic impulses: Evidence from polls and election results indicates that their assessment of the national economy weighs heavier than their own financial situation. One promising new line of research even suggests that voters are able to sift out real evidence of candidates' competence on economic matters from the inevitable ups and downs of the economic cycle.

One piece of evidence for that view, unearthed by Michael Ebeid of Boston University and Jonathan Rodden of MIT, is that voters in gubernatorial elections tend to pay relatively little attention to the economy in states where farming and natural resources dominate--and economic cycles are thus almost entirely out of the hands of local decision-makers. Another is found in recent work by Raymond Duch of the University of Houston and Randy Stevenson of Rice University, who examined the "economic vote" in elections in the U.S., Canada, and Western Europe for the past 20 years and found that economic issues weigh heaviest in voters' decisions when a nation's economic performance is dramatically different from that of its peers.

It is, of course, hard to imagine American voters checking other countries' economic statistics before going to the polls. But in 1992 they didn't have to: "Last year Germany and Japan had productivity growth rates that were three and four times ours," Bill Clinton told those who sat through his economic stump speech. The most famous one-liner of the campaign was Ross Perot's nationally televised warning of a "giant sucking sound of jobs going south" if the North American Free Trade Agreement became reality. The economy wasn't just about people's self-interest, Clinton and Perot both seemed to be arguing. It was about patriotism!

It turned out, of course, that the sucking noises brought on by NAFTA weren't all that deafening, and by mid-1992 the German and Japanese economic miracles had both begun to sour. Voters can't see into the future, and they have trouble digesting economic news until it's a few months or even years old. What matters is that the economic danger seem real and that the difference between candidates be clear. That was certainly the case in 1980, when unemployment and inflation were both painfully high, and the incumbent President came across as indecisive and pessimistic. Ronald Reagan offered a straightforward and optimistic alternative --even if economists still argue about whether it was the right one.

So where does that leave us in 2004? By past standards GDP growth is strong enough and unemployment low enough that the current President Bush shouldn't have trouble getting reelected--and most of the "battleground" states that haven't been sewed up by one candidate or the other are doing even better than the nation as a whole (see map). The forecasting models that will be wheeled out at the APSA annual meeting in Chicago this month will probably all predict a Bush victory. But the models' keepers are skeptical. The war in Iraq is a big factor that can't be captured by GDP numbers, says Alan Abramowitz of Emory University, whose formula is forecasting a 53.7% Bush majority. "There sometimes is a gap between the objective economic indicators and people's subjective perceptions of the economy."

It might not be entirely subjective: While real GDP rose at a healthy 3.8% annual pace in the first half of 2004, real per capita disposable income--the after-tax measure of what actually makes it to Americans' pockets--is up only 2.1% annualized over that period, the lowest such increase in a presidential election year since 1980 (when it actually declined 3.2% in the first half of the year). Subtract from that rising gas prices and health insurance costs--and throw in a job market that, while it showed signs of life earlier this year, is far from booming--and you have millions of Americans who aren't better off than they were four years ago.

Even those Americans who are doing just fine see troubles in the economy: In virtually every poll taken this summer, more than 50% of Americans have said their own finances are in good or excellent shape, but less than 40% say the same about the national economy. When asked who is likely to do a better job dealing with economic issues, Kerry usually beats Bush by about a ten-point margin. "It's not good news for an incumbent," says Republican pollster Tony Fabrizio.

But is it really such bad news? Only 39% of Americans polled by Gallup in early August think the condition of the economy is good or excellent. But that's a lot better than the 10% who were similarly positive in August 1992. When Americans are asked these days what the most important problem facing the country is, the economy invariably gets more mentions than any other issue. But the picture changes dramatically when you group responses--as the Pew Research Center for the People and the Press did in August--into three broad categories: "war/foreign policy/terrorism" (41%), "economic issues" (26%), and "other domestic issues" (26%).

There are some obvious reasons why the war/foreign policy/terrorism troika wins out (war and terrorism being two of them). But it may also have something to do with the convergence of Democratic and Republican economic policies in the mid-1990s, when both parties agreed on reducing trade barriers, cutting the deficit, and keeping the dollar strong. The current Bush administration's fervor for tax cuts in the face of rising deficits has pulled that consensus apart, but Kerry has not so far articulated a radically different alternative (as Reagan did in 1980 and Clinton and Perot both did in 1992). If he were to denounce free trade and holler about soaking the rich, he probably could make the economy a bigger issue. But that is neither Kerry's style nor necessarily the wisest strategy in what appears to be an excruciatingly close race.

To illustrate, let us return to the Trails Inn Cafe, where, after repeated prodding from me, a couple of the breakfast eaters finally get an economy-related debate going. The subject is taxes. Sure, the wealthy have gotten big tax breaks from Bush, says real estate developer John Ferguson. But when they bought things with that money, it benefited whoever made, packaged, and transported those things. Retired trucker Roth allows that that would be so, if the rich folks had spent their tax windfall. "I don't believe they did," he says.

The debate continues, in the calmest of tones. Then comes an outburst from down the table. McGuire, the retired teacher, has been describing the anti-Bush sentiment he encountered during a trip to Europe earlier in the summer. Sitting next to him is Randy Hylton, who runs a nonprofit that finds jobs for the developmentally disabled. Hylton can finally contain himself no longer. "Are you going to vote because some German tells you what to think?" he almost shouts. "Who gives a damn what they think!?" All talk of taxes and the economy is forgotten. Now this is interesting!

As a method of plumbing national opinion, my breakfast visit is of course a sham. The crowd at the Trails Inn is a bunch of white men, most on the far side of 50, who in no way other than their split political allegiances can be said to represent the nation. Not long after breakfast, I visit the barbershop on Liberty's picturesque main square, where barber Steve Zink cites Ronald Reagan's "Are you better off?" standard. "I'm no worse off under Bush," he says. "My daughter is. She works for AT&T. She blames it all on Bush."

After 2000, when we all were reminded that the presidency can be decided by a few hundred votes, one cannot discount the possibility that this year's election will turn on the opinions of a few ticked-off telco employees (or laid-off telco employees) in Kansas City or Denver or Tallahassee. But a national referendum on the economy this vote is not. There are too many other things to get worked up about.

FEEDBACK jfox@fortunemail.com