I am confused how an incumbent wins after being down in double digits in August.

Tony Fabrizio had a good piece on this in NRO:
http://www.nationalreview.com/comment/fabrizio200407080940.asp

JAB:

Because the numbers aren't real, they just reflect blanket favorable coverage in the news media. If you check the polls John Edwards and John Kerry will have lower name recognition in August than in July.

Interesting. I did see that Kerry's name recognition went down after the primary.

Still, I am concerned that a big lead becomes a self fulfilling prophesy in this media climate.
I have no confidence in the WH communications strategy in combatting this risk.

How fast did Howard Dean go from sure thing to nobody?

No chance does Kerry ever go up by anywhere near 10 points. I want to see a poll where the first 2 questions are the names of the two major party candidates. That would tell you a lot about how meaningful (-less) the numbers are.

JAB:

President Mondale led Reagan, President Dukakis led Bush by double-digits, President Gore led by double-digits too. It's not real, just name recognition.

OJ
I'm sure you are correct and I don't dispute what you are saying at all, but the conclusion I come to is that the public is made up of clueless, fickle idiots and that we should stop this Democracy thing immediately before we hurt ourselves.

h:

That goes without saying. No reform would serve our future better than a drastic restriction of the franchise.

OJ notes the expectation - Kerry goes up due to the convention but then Bush catches up when the GOP has it's convention. I agree with Brian - I wonder if Kerry will get up the 15pts the GOP pollster predicted. So far he has gotten little bounce from naming Edwards.
Two caveats - Iraq or the economy worsening would hurt Bush. Also hearing that Kerry's fundraising has been so good he may forego the $75MM limit after the convention and may have more money than Bush.

I truly doubt that Kerry will be up by anything like 15 points after the Dem convention, I believe that what Matt Dowd (the GOP pollster) was trying to do by claiming Bush would be trailing by that large a margin, was trying to set expectations abnormally high for the Kerry camp, expecting them to fail to achieve them.

The strategic benefits are obvious - throw out a number too large to be ignored and you get instant coverage from the Kerry sympathetic press who want to believe that Bush will be trailing by that much, especially after he's managed to maintain parity (or even a slight lead) with Kerry after seven months of the worse press coverage they could muster. Then after the convention instead of a 15 point lead that everyone expects we'll see Kerry with around a 7 or 8 point lead (the historic norm) which immediately makes him look weaker than he should against Bush. Talk goes from "his commanding lead" to "is he under performing?".

Iraq will undoubtedly continue to improve (unless Syria or Iran decide to interfere overtly) and the economy continues to grow stronger, leaving the Kerry/Edwards with exactly what team to run on? Other than their hair, that is...

Way too much attention is being paid to polls. Many long years ago, in the Nixon Administration, before Watergate and before I went to law school, I was in graduate school. In the summer of 1971, I took a seminar on statistics. One of the things we were allowed to do was to spend a lot of time playing with their collection of polling data from the past couple of elections. I came to the conclusion that the polling data was meaningless.

Why? Probably because the populace does not understand the political jargon used by the pollsters. Furthermore, pollsters can and do manipulate results, consciously or unconsciously. The order and way in which questions are asked makes a huge difference.

Indeed, the mere fact that a poll may identify itself as being associated with a media organization such as ABC, might lead subjects to shade their answers to what they think Peter Jennings would want to hear. Finally, as time has gone on the validity of poll sample has continued to corrode. Many people do not answer the phone during certain times, screen calls, refuse unsolicited calls, have unlisted numbers, and increasingly are using non-wireline telecommunications devices such as cell-phones and voip that pollsters cannot access.

In 1980, the polls showed the race as being very close. At some point in October, ABC ran an automated 800 number call-in poll. Reagan won handily. Pundits rushed to pooh-pooh the "poll." In November, Reagan won handily.

This year the pundits and talking heads are telling us that the election is going to be very close, and the polls seem to confirm this conventional wisdom. Of course, if the pollsters wanted the polls to sing "O Solo Mio" in E flat they would. And at the prices the media is paying, the pollsters are not going to be making the blow-dries look like the buffoons they most assuredly are.

Conventional wisdom is almost always wrong, and, at any rate, you can't make money on it if it is not. So I am betting that Bush will win and win by a substantial margin. I am buttressed in my belief by the predictions of those economists who base their political predictions on economic models, not polling data.

The lead should swing in the Democrats favor by double digits over the course of the next couple weeks...

If we get lucky, even that may not happen. If the Democrats run a four-day Bush-Bash-a-palooza like I expect, the polling gains might be minimal. Everyone remember how George Bush Sr. won a grand total of three percentage points in national polls after the '92 GOP Confab?

In the last week before the 1984 election, Newsweek ran a cover (showing Reagan by Marine One) with the word LANDSLIDE? on top. The Democrats went berserk (but of course everyone knew what was coming). Will that happen this year if Bush is up 7-8 points on Oct. 28? No way.

The polls are meaningless, except to watch the press work themselves into an erotic frenzy. But that post-frenzy crash is fun to watch.