WASHINGTON — While most of America is watching the spread in the polls between President Bush and Sen. John F. Kerry, key strategists in both parties have their eyes on a different set of numbers: Bush's share of the vote and his job approval in the final surveys before election day.
Analysts watch the incumbent's numbers in the polls so closely because most voters who stay undecided until the very end of a presidential campaign traditionally break for the challenger. As a result, challengers often run ahead of their final poll results, while incumbents rarely exceed their last poll numbers.
"We know from the history of presidential elections that when a president is polling below 50% going into the election, he usually loses," said Alan I. Abramowitz, an Emory University political scientist. "That is true of incumbent office holders in general. The incumbent usually ends up getting the percentage that he is getting in the final polls -- that's it."
By that standard, the race today is teetering right on the knife's edge, though perhaps tilting slightly toward Bush after he regained the lead in five separate national polls released over the weekend. More importantly, for the first time since the debates, Bush in three of the latest surveys cracked the 50% level in support -- the best news GOP strategists have seen in weeks.
Surveys released Saturday by Newsweek and ABC/Washington Post put Bush's support at 50% among likely voters. On Sunday, Bush reached 52% among likely voters in a CNN/USA Today/Gallup Poll, opening an 8-percentage-point advantage over Kerry.
But a survey released over the weekend by Time placed Bush at 48% -- as did the Newsweek result among registered voters. And the daily tracking poll by independent pollster John Zogby on Sunday put Bush at 46% with likely voters. Among registered voters, Bush got 49% in the new Gallup Poll.
Bush's approval rating, another key indicator, is still running just below 50% in most polls.
In the final weeks of the campaign, Democrats are basing their hopes less on the difference between Bush and Kerry than on those surveys showing the president below 50% in support.
"This is a very well-known incumbent where people have strong views," said Democratic pollster Stanley B. Greenberg, a Kerry advisor. "His number [in the last polls] I believe is his number [on election day]."
Even some senior Republican strategists privately agreed that the experience of the last half-century supported that argument. In the history of polling dating back to 1952, no incumbent president has run even 1 full percentage point better on election day than he did in the final Gallup Poll before the vote.
But other GOP strategists argued that doubts about Kerry would allow Bush to capture enough late-deciding voters to win, even if he couldn't stay near 50% through election day.
"I'd rather be over 50 than under 50, but just because you're under 50 doesn't mean you are destined to lose this kind of race," said Republican pollster Whit Ayres. "Both Bush and Kerry have universal name recognition by now and this is a lot more than just a referendum on the incumbent."
Matthew Dowd, the Bush campaign's chief strategist, has privately raised similar arguments to leading Republicans. Dowd noted that Kerry has been unable to establish a clear lead over Bush in almost any poll, despite good reviews from viewers in all three debates. That pattern, like Kerry's failure to open a solid lead in polls after the well-received Democratic convention last July, suggested voters who were uncertain about Bush might remain hesitant to commit to the Democrat.
"If [Kerry's] first debate was his finest moment, and he didn't take a lead, what is going to cause an undecided voter ... to say, 'Now, I'm going to go with Kerry'?" Dowd said.
Three factors add to the uncertainty as the two sides anxiously pore over the late poll numbers. One is whether an unexpectedly large turnout -- which appears possible given all the signs of unusually high interest in the race -- could give either side an advantage not fully measured in late polling.
Another is that no one is certain how much of the vote will be needed to win this year -- since no one knows how many votes Ralph Nader and other third party candidates will siphon away.
The final factor is the inevitable imperfection of polling. "At 45% or 46% and tied or down two, that is a long road for Bush -- where I've had lots of unhappy outcomes in my career," said one leading Republican. "At 48-48, I've seen incumbents win by 15,000 votes. There is a big difference [for Bush] between 48 and 46. The problem is polling isn't that good."
History isn't always predictive, but races involving White House incumbents have produced a clear pattern over the last 50 years.