A question for those of you familiar with Presidential polls

[ArcturusMengsk]

According to 538, Obama's likelihood of victory according to simulations based on polling averages comes out to about 90%, as of today. That sounds like great news, but being that this is the very first election I've ever payed close attention to, and the first I've ever participated in, I'm still very jittery. So I'd like to know: has there ever been a historical precedent for someone as far behind in the polls as McCain is currently to come back from behind at the last minute and win the election? I'm vaguely familiar with the circumstances of the 1948 election, when most pollsters and pundits were predicting a Dewey triumph only to be proven wrong, but I'm also aware of the difficulties of polling at that time (telephone landlines then were usually owned only by the well-to-do, who were more apt to be supportive of the Republican Dewey rather than Harry Truman), and I've heard that a similar situation exists today, in that poorer households which are more likely to lean to Obama frequently have only cellular phones, which are rarely polled.

So to set my mind at ease, and barring such dubious notions as the Bradley effect, is there any reason to believe based on past evidence that a come-from-behind victory for McCain is possible this late in the game? If so, is it possible to access such polls on-line?

[Ariphaos]

That's essentially what the 10% is - Nate's model is predicting that that's the sort of chance McCain has to stage that level of a comeback, based on how violent swings can occur in such a short period. That number gets smaller with every day for obvious reasons.

[RedImperator]

In 1948, Gallup stopped polling in mid October, so nobody saw Truman's surge. The thing to remember about that election, though, is that Dewey ran what is likely the most anemic campaign of the modern era. Truman was unpopular and the Democratic party had split three ways in that election, and so Dewey's strategy was, literally, to say nothing of consequence during the entire campaign, under the assumption that Truman couldn't possibly win unless Dewey made some kind of horrible gaffe. So while Truman was crisscrossing the country, tearing into Dewey and the Republicans in Congress at every whistle stop along the way, Dewey was mouthing some pablum about "unity". And he still managed to commit a major gaffe: on his own whistle stop tour, his train slipped forward a few yards while it was stopped, and Dewey snarled something about having the engineer shot in front of a bunch of reporters and small-town voters.

So the important differences between 1948 and today are 1) the polling will continue right up to Election Day, and 2) Obama is not going to take his boot off McCain's neck until 5 November.

Now, as for the Bradley effect. Here's a link to a Harvard study concluding that the Bradley effect did exist, but no longer does. The link is on the page; it's a PDF, so I can't quote it here. It's not the last word on the subject, but it's encouraging. It's also worth considering that in 57 primary contests against a white candidate, Obama underperformed his poll numbers only three times, while he considerably over performed them all across the South, indicating the possible existence of a reverse Bradley effect in areas where white voters might be reluctant to admit they will vote for a black candidate.

[CmdrWilkens]

With the size of Obama's lead at THIS point in the game it would be a unique and new comeback for McCain. The closest analogy would be Bush in 1992. After Perot essentially dropped off the radar screen Bush went from about -14 to -5 during October. The things to note are 1) Obviously Bush didn't win, 2) He started on an upwards trajectory starting with the convention in August (though it leveled off for a bit in September) and 3) Mccain doesnt' even have all of October left.

Simply put there are 3 campaign weeks left at this point and next week the narrative will kick off with Troopergate and the continuing economic crisis. This means at BEST McCain can hope to salvage a change in narrative on Tuesday after the next debate. However, and this is the however, if he fails to shift the narrative Monday night then that is at least another week where obama will drive the narrative. Now we get into the final two weeks of whcih the final week Obama has already reserved a HUGE chunk of national airtime to get his message out. There is no way that McCain can compete with that kind of ad buy so he really only would have the week after the debates to change the narrative and swing the eleciton by 6-7 points. Simply put he is on such a compressed time scale that there is simply no comparable period in the late stages of an election for such a drastic shift.