Public opinion, at least in the context of campaigns, can dart rapidly in one direction or the other, so this election is by no means a forgone conclusion. One presidential debate remains this Wednesday, and McCain will search once more for a "game-changer," however elusive this might be.
Three weeks of ads in targeted states will also provide additional contrasts between the candidates, and often resort to negative attacks. To date, McCain's attempts to link Obama to former Weather Underground domestic terrorist Bill Ayers and voter fraud perpetrated by ACORN have fallen mostly on deaf ears, but expect the campaign to continue to raise doubts about the junior Illinois senator.
Also, both campaigns are already turning to their so-called "ground games," the get-out-the-vote operations in swing states that focus on delivering the parties' respective bases to the polls on Election Day, or earlier (absentee balloting).
Poll numbers may move in response to a decisive debate performance by one candidate or the other, or a notable gaffe in this same environment. Attack ads may gain traction and push away undecided voters from one candidate to the other, or to sit out the election altogether. One party's 72-hour plan during the final three days of the campaign may defy so-called momentum and the suggestions of last-minute polls (I witnessed this myself in New Hampshire this year when Senator Hillary Clinton pulled off a narrow win over Obama when polls were suggesting a decisive win for the latter candidate).
A couple of poll-related phenomena also must be considered at this late hour of the 23-month campaign for president. One, the so-called "Bradley effect," named after the former Mayor of Los Angeles who lost a close election for governor despite polls predicting victory. Bradley was African-American, and the theory suggests that lower income voters overstate their support for black candidates in interviews with pollsters.
The academic community is split over the Bradley effect's continued relevance given that is surfaced in 1982 and Americans' views about race have arguably progressed in the intervening decades. Obama's candidacy itself is a testament to this, including his better-than-expected performances in Wisconsin and Virginia. However, he fared worse than anticipated in New Hampshire and California, and the primary electorate is vastly different than that of a general election.
Those who believe the Bradley effect lingers argue that lower income white voters are less likely to respond to pollsters questions to begin with, perhaps biasing the complexion of existing samples. Moreover, differences surface depending upon who is asking the questions. Individuals are more likely to express support for African-American candidates if and when they are questioned by a black survey administrator.
Others express skepticism, suggesting that polling techniques are increasingly refined nowadays, predicting the loss of African-American congressman Harold Ford in the 2006 race for Senate and the victory of Deveal Patrick as Governor of Massachusetts in the same year. Moreover, the Bradley effect is more likely to play a decisive role when an election has been "racialized," as wedge issues like affirmative action and welfare enter the fray. Although Reverend Wright has not yet been put to bed, this election is focused on the economy, gas prices, and health care, and to a lesser extent the Iraq and Afghanistan conflicts.
A second variable to keep on our collective radar screens in the closing days of this election is the bandwagon effect. This occurs when one candidate assumes a decisive lead in the polls, inviting voters to climb aboard. Should Obama continue to build upon his narrow, but widening lead, we may witness a run toward the Illinois senator that results in a decisive victory in the popular vote and a blowout in the Electoral College. If, however, the race returns to its previous dynamic, a nail biter within the polls' margins of error, the bandwagon effect exits stage left, and a long night of election returns looms.