The electorate has been polled, polled and polled again. Campaign workers have knocked on millions of doors with millions more to hit before Nov. 2. Voter registration figures in some states show big increases. And voter interest in the presidential election appears to be at record levels.
But the biggest mobilization in modern presidential politics cannot answer the big question that could determine the outcome of the race between President Bush and Sen. John F. Kerry: Is there an invisible army of voters out there -- and if there is, will it tip the balance toward the incumbent or the challenger?
All indicators point to an increase in turnout compared with 2000, when nearly 106 million voted. Curtis Gans of the nonpartisan Committee for the Study of the American Electorate predicts participation levels equivalent to 1992, which would mean 118 million to 121 million voters this year.
But the composition of the electorate, perhaps more than its size, is the crucial unknown heading into the final week of campaigning. Will there be a surge of young voters, a spike in participation by social conservatives, an outpouring of Republicans loyal to Bush, an energized cadre of sporadic voters hostile to the president, an invigorated black or Latino vote -- or will all of the above occur?
The belief in a hidden army of voters stems from signs of voter interest in the election, registration figures reported by the states and confusion generated by a profusion of polls that sometimes appear sharply at odds with one another about the state of the presidential race.
Many partisans say that they believe polls this year are not capturing the views of new voters or that they do not accurately predict who the likeliest voters are. Pollsters say they are working to make sure that new registrants and sporadic voters who are certain to vote this year are included in their samples -- but there are limitations to their ability to predict the shape of the electorate.
"I think we're likely to see, certainly, a higher turnout than in 2000 and the 1990s, and possibly the highest in our lifetime," Republican pollster Whit Ayres said. "Is it going to be higher across the board, or will some demographic groups turn out disproportionately? That really is the critical question, and I don't think anybody has a good handle on that."
The Bush campaign, led by White House senior adviser Karl Rove, consciously set out after 2000 to alter the shape of the electorate in 2004, hoping to narrow or eliminate what has been a historic Democratic advantage in party identification on Election Day. In the past three presidential elections, 35 percent of voters identified themselves as Republicans, while 38 percent to 39 percent said they were Democrats.
Because Bush lost the popular vote in 2000, his advisers launched an effort to register millions of new GOP voters, calculating that, by raising the overall GOP percentage a point or two, they could go a long way toward ensuring the president's reelection. The Bush team concentrated efforts in heavily Republican precincts, particularly in fast-growing exurban counties, and last month, Republican National Committee Chairman Ed Gillespie announced that the party had achieved its goal of 3 million new registrants.
The GOP efforts, however, produced a counter effort among Democrats, fueled by strong anti-Bush sentiment. Democrats, aided by independent but allied groups such as America Coming Together and ACORN, mounted registration drives in cities, aimed at minority voters. Final figures are not available in all states, but the arms race in registration may have canceled out efforts on both sides, with an energized right clashing against an equally energized left.
"When we look at this after the election, we're going to look real hard at whether Rove succeeded in getting enough energy and enthusiasm among his voters but also whether he produced more anti-Bush voters than Bush voters," said Democratic pollster Stan Greenberg, a Kerry adviser.
There are several potentially significant voter blocs in this election. One consists of voters younger than 30, who normally turn out in numbers disproportionate to their large share of the population, but who have been the focus of great attention in this campaign.
The 2000 campaign provided few incentives for young voters to participate, with much of the debate between Bush and Vice president Al Gore focused on Medicare, Social Security, prescription drugs and other issues of interest mainly to older voters. But the economy, Iraq and speculation about a military draft have younger voters paying more attention.
A poll by Harvard University's Institute of Politics, released Thursday, showed college students splitting for the Massachusetts senator 52 percent to 39 percent. But other polls suggest that the universe of young voters may be more evenly divided. "This might just be the election where they turn out," Andrew Kohut of the Pew Research Center said of the under-30 voters. "But we don't find a consistent Kerry or consistent Bush trend. It moves around a lot."
Republicans have made a special effort to mobilize social and religious conservatives. Rove has said that there are potentially 4 million such voters who did not turn out in 2000 who should be recruited this year, but some doubt the reliability of that figure.
John C. Green, director of the Ray C. Bliss Institute of Applied Politics at the University of Akron, said he and two colleagues from other universities examined exit polls and other data to try to determine whether the 4-million estimate is plausible. "We couldn't figure out where it comes from," he said.
Four years ago, white evangelicals voted at about the same proportion as the overall population. To increase that vote, Republicans have built a program for targeting social conservatives that goes far beyond the days of dropping literature about the candidates' positions at churches -- the outreach is now handled by the Bush-Cheney operation in a much more systemic way, which includes phone banks and direct mail. "That whole church-based lit drop is so Moral Majority, 1980s," one Bush adviser said.
After looking at the Bush campaign voter mobilization in Ohio, Green said: "If effort counts, turnout among white evangelicals in the battleground states ought to go up." Just how much is the question.
On the Democratic side, black and Latino votes are as critical, perhaps more so, as social conservatives are to Republicans. A recent poll for the Joint Center for Political and Economic Studies showed Bush receiving 18 percent of the black vote, double his share in 2000.
That spells trouble for Kerry, if it holds true on Election Day, but no one in either party believes it is likely. Some national polls that show a close race may be understating the share of the black vote Kerry will receive, which could affect his overall vote by a full percentage point.
Still, Democrats are concerned about the lack of enthusiasm for Kerry among black voters and have launched efforts to boost turnout. That is particularly true in Wisconsin, where Democrats have targeted blacks in Milwaukee for special attention.
Both parties have competed for a Latino vote that could grow in line with population increases, with the Republicans trying to gain a larger share than the 35 percent Bush received in 2000. But Democratic groups such as the New Democrat Network have spent millions of dollars on Spanish-language ads in battleground states, designed to boost Democratic turnout among Hispanics.
Gauging the impact of newly registered voters, sporadic voters and the relatively few undecided voters is equally difficult. Some polls suggest that newly registered voters of all ages favor Kerry, but who knows how many will vote. "When you start talking about new voters, historically you're talking about people whose intentions are better than their performance," Gans said, "but this year that may not be true."
Many infrequent voters are unmarried men or unmarried women. Unmarried women vote strongly Democratic, when they cast ballots, but four years ago, unmarried men split almost evenly between Bush and Gore. In Iowa, America Coming Together has put much of its effort into identifying and turning out citizens who have voted infrequently, and the Democratic National Committee has targeted these voters in battleground states.
Undecided voters, while fewer than in the past, are still numerous enough to make a difference in a close election. Historically, undecided voters break against an incumbent president, and Kerry advisers say their analysis suggests that those voters are eager for a change of direction and will side with Kerry. But Bush advisers say these voters are lukewarm toward Kerry -- something Democrats privately confirm -- and believe Bush has a chance to win a bigger than normal share of them.
Finally, there is another oddity in some polling this year. In many national polls, Kerry runs better in a subsample of voters in battleground states than he does overall, suggesting that the electorate that has been bombarded by television ads and courted with numerous visits by the candidates may see Bush and Kerry differently than do other voters.
Matthew Dowd, senior strategist for the Bush-Cheney campaign, disagrees, saying his analysis shows that the battleground states generally track national polls. Kerry strategists believe that battleground voters view Bush more negatively than the overall electorate and that that gives Kerry an important advantage.
Ultimately, the election will test the president's strategy of creating a new GOP electorate and Rove's bet that the key to victory is an energized GOP base. But Democratic pollster Peter Hart, noting the enthusiasm among Democrats, said Rove's calculations may not be taking into consideration an outpouring of anti-Bush votes. "Karl Rove may be energizing too small a percentage" of the electorate.
As the campaigns gear up their final get-out-the-vote operations, they know that is something that can be answered only on Election Day.