Electoral Vote Projections

[Old Blood and Guts]

I can't see Arkansas or Missouri going for Bush. It just isn't going to happen. The fact that Kerry isn't using any resources in that state gives evidence to the claim, just as Bush isn't using resources in Michigan, because Bush simply isn't going to win Michigan (I would assign equal probabilities to Bush winning Michigan and Kerry winning Arkansas and/or Missouri). Also, I don't see the Colorado ballot initiative passing. An interesting article I read this week stated that given past history on ballot initiatives in Colorado, an initiative is unlikely to pass if it isn't polling at least 50% of the vote a couple of weeks prior to the election, and the Colorado initiative isn't. If it does pass, I expect it to get overturned in the courts since it is a nakedly political move funded by sources outside of Colorado.

If I were a betting man (and I'm not), I would bet on Bush winning 300 electoral votes (all the states he won on 2000 plus Wisconsin and Iowa).

[Jason A]

Bush below 50%, but still up by five (47%-42%) on Kerry in Colorado.

http://www.rockymountainnews.com/drmn/election/article/0,1299,DRMN_36_3262788,00.html

[Rover]

If I were a betting man (and I'm not), I would bet on Bush winning 300 electoral votes (all the states he won on 2000 plus Wisconsin and Iowa).

Its a good thing you're not a betting man with that prediction. ;)

While we still have a couple of weeks left, states are starting to go Kerry's way. Minnesota is getting talked about less and less for a reason. Similarly, the last few polls are showing Kerry pulling ahead in Wisconsin. Iowa is even at this point, which is better for Kerry than two weeks ago. A couple of Florida and Ohio polls also show Kerry ahead.

Right now, it appears that Kerry can afford to lose either a Wisconsin or an Iowa with an Ohio win. If he loses both, he'll have to pick up NH (probable) and NV (50/50). That's doable either way and if I were Bush I'd spend half of the remaining days in Ohio. Essentially he's cooked without it.

[Kepler]

Unless we have a reversal of history, Bush probably needs about a 3-4% lead in his essential battleground states on election day morning, or the usual last minute challenger rush will kill him.

But it's still 50/50. Kerry's poll momentum has slowed down -- maybe even stopped -- with the spinner right on the line. Now it's a matter of whether the political winds blow it to one side or the other over the next 2 weeks.

Hopefully it won't be so close that voting, er, irregularities decide it.

[Kepler]

This is a prediction by John Zogby, the founder of the Zogby service. Hardly definitive (he gets one vote like the rest of us), but interesting.

http://www.zogby.com/news/ReadNews.dbm?ID=894

'Polling can be uncertain only if you rely on statistics alone,' is Mr Zogby's riposte. 'That's why an effective pollster has to rely on culture, history and sociology. I repeat, polling is the study of human behaviour, not simply a sampling of people's preferences.'

That may well explain his success. His big test will come on Nov 2. In Singapore last Friday, he flatly predicted that Mr Kerry would become the 44th President of the United States.

[dropthatpuck]

This is a prediction by John Zogby, the founder of the Zogby service. Hardly definitive (he gets one vote like the rest of us), but interesting.

http://www.zogby.com/news/ReadNews.dbm?ID=894

'Polling can be uncertain only if you rely on statistics alone,' is Mr Zogby's riposte. 'That's why an effective pollster has to rely on culture, history and sociology. I repeat, polling is the study of human behaviour, not simply a sampling of people's preferences.'

That may well explain his success. His big test will come on Nov 2. In Singapore last Friday, he flatly predicted that Mr Kerry would become the 44th President of the United States.

His own Zogby/Reuters polls have not been in agreement with that statement, until exactly today when the latest Zogby/Reuters poll shows them dead even. Up until yesterday, he had Bush at +4%. No way does that happen in a day. He's certainly respected considering his .1% prediction/realization in '00.

I've not been posting any of these polls, because they are just that and they continue to move each day.

IMO, a respected man like that shouldn't be making those sort of predictions in a foreign nation....but it is his business.

[Kepler]

His own Zogby/Reuters polls have not been in agreement with that statement, until exactly today when the latest Zogby/Reuters poll shows them dead even. Up until yesterday, he had Bush at +4%. No way does that happen in a day.
I'm not sure whether Z/R is rolling average or standalone, but with rolling averages polls, if you get a result at one end of the MoE dropping off on the day a result on the other end of the MoE comes on, you can get a 4 point swing easily.

Say MoE is +/-4% and the population is a steady Bush +2%. If you sample at:

+ Friday, Bush +6 (pop +4%)
+ Saturday, Bush +2 (pop)
+ Sunday, Bush +1 (pop -1%)
+ Monday, Kerry +2 (pop -4%)

then your three day rolling average is

Sunday, Bush +9/3 which rounds to Bush +3
Monday, Bush +1/3 which rounds to Even.

Now assume just a *one* point real swing in the population towards Kerry over the four day interval from Friday through Monday, and you have a 4-point sample swing between Sunday and Monday.

Easily possible; no conspiracy required.

[Bob Gray]

the usual last minute challenger rush will kill him.

Never heard of that before. I think the last minute rush can go either way depending on a variety of factors or there may not be a last minute rush.

[Kepler]

Never heard of that before. I think the last minute rush can go either way depending on a variety of factors or there may not be a last minute rush.
I'll dig around for documentation, but the numbers I have read are that undecideds in the final week of the election break against the incumbent 2:1. The idea being: nobody waits until the last minute to "get comfortable" with the incumbent, a known quantity, but plenty of people ease into voting for change.

A pretty standard observation, but it certainly needs documentation.'

Section 2.6 of the following seems to contain some useful information (yes, the site is partisan): http://www.eriposte.com/election04/appalling_polling.htm

Snippet:

"Using my own research as well as research sent to me by Nick Panagakis of the National Council on Public Polls, I have gone through 451 poll results since 1976. In all 451 cases, the poll was in the field for at least one day that was within seven days of the election. In every case, it was the final poll taken by the polling firm for the campaign in question. Also, I do not believe that any internal or partisan poll results were used. Unfortunately, outside of the Presidential race, I was severely lacking in data from 1996.

...The results were as follows:"

Year Polls Und. Inc. Chal.
President 28 2.4 14% 86%
1976-88 155 11.8 20% 80%
1994 101 11.2 35% 65%
1998 76 10.1 27% 73%
2000 31 8.6 40% 60%
2002-4 60 7.5 42% 58%
1992-04 283 8.9 34% 66%
Total 451 9.7 28% 72%


So by this guy's measure, the break was 72-28 for the challenger, albeit over a short interval. Note that I was reading exactly the same stuff back in '96 in favor of Dole, so while it's use this year is obviously a source of Dem delight, the indicator itself is nothing new.

[Bob Gray]

Sorry, I started reading that website, but I just couldn't take enough to see if the arguments about late polling make sense or not. Very partisan. I still think that it can go both ways though, depending mainly on late events/trends. I heard somebody on the local news radio talk station here in Phoenix (not the conservative one either) say that basically in 2000 Bush had a slight lead in the polls heading into election day and that that trend basically carried forward through the election results. I'm not saying your theory may not have some traction, but I just don't see it necessarily and haven't heard of it before.

[Kepler]

I'm not saying your theory may not have some traction, but I just don't see it necessarily and haven't heard of it before.
Totally valid, and happily, we shall see in two weeks.

Here is another citation of the 2:1 rule from the most partisan forum on the right, Free Republic: http://209.157.64.200/focus/f-news/1103820/posts

"Undecideds typically break for the challenger by a 2-1 margin or more."

[Bob Gray]

FYI. I don't like highly partisan stuff from either side. I find, regardless of which side it's on, that as partisanship increases, the usefullness of the information/view/spin being put forth decreases.

[Todd]

His own Zogby/Reuters polls have not been in agreement with that statement, until exactly today when the latest Zogby/Reuters poll shows them dead even. Up until yesterday, he had Bush at +4%. No way does that happen in a day. He's certainly respected considering his .1% prediction/realization in '00.

I've not been posting any of these polls, because they are just that and they continue to move each day.

IMO, a respected man like that shouldn't be making those sort of predictions in a foreign nation....but it is his business.
A) National % vs EV could account for your difference.
B) Why does it matter where he is when he says something? If Reuters flies him to Singapore to speak about polls and asks him the question, should he wait until he gets back to the US and email them his answer?

[Buff]

I still think that it can go both ways though, depending mainly on late events/trends. I heard somebody on the local news radio talk station here in Phoenix (not the conservative one either) say that basically in 2000 Bush had a slight lead in the polls heading into election day and that that trend basically carried forward through the election results. I'm not saying your theory may not have some traction, but I just don't see it necessarily and haven't heard of it before.


The one problem with your example, obviously, is that there was no incumbent in 2000.

[Red Cloud]

Arkansas
KTHV-TV: 51-46 Bush
Looks like a winner for the President.

Colorado
Rocky Mountain News: 47-42 Bush
Ballot initiative still expected to pass.

Florida
WFOR-TV: 50-49 Kerry

North Carolina
WBTV-TV: 50-47 Bush

Oregon
Riley Research: 48-43 Bush
Somewhat surprising result.

Pennsylvania
KDKA-TV: 51-45 Kerry
Keystone State should be blue in November.

Wisconsin
Rasmussen: 48-47 Kerry

[Jason A]

I'm not sure whether Z/R is rolling average or standalone, but with rolling averages polls, if you get a result at one end of the MoE dropping off on the day a result on the other end of the MoE comes on, you can get a 4 point swing easily.

Say MoE is +/-4% and the population is a steady Bush +2%. If you sample at:

+ Friday, Bush +6 (pop +4%)
+ Saturday, Bush +2 (pop)
+ Sunday, Bush +1 (pop -1%)
+ Monday, Kerry +2 (pop -4%)

then your three day rolling average is

Sunday, Bush +9/3 which rounds to Bush +3
Monday, Bush +1/3 which rounds to Even.

Now assume just a *one* point real swing in the population towards Kerry over the four day interval from Friday through Monday, and you have a 4-point sample swing between Sunday and Monday.

Easily possible; no conspiracy required.Zogby is rolling average. Bush led by four on Saturday, the lead was cut to two on Sunday, and it was a tie on Monday, I think. I haven't seen today's number yet.

A poll released yesterday shows Bush polling at 41% and trailing by five points in New Hampshire. A poll released the previous day showed him down by four points, so it seems the Granite State is leaning toward Kerry right now. But it has gone back and forth.

http://www.suffolk.edu/spotlight/poll_10_18.html

[Clifton]

Arkansas
KTHV-TV: 51-46 Bush
Looks like a winner for the President.

Colorado
Rocky Mountain News: 47-42 Bush
Ballot initiative still expected to pass.

Florida
WFOR-TV: 50-49 Kerry

North Carolina
WBTV-TV: 50-47 Bush

Oregon
Riley Research: 48-43 Bush
Somewhat surprising result.

Pennsylvania
KDKA-TV: 51-45 Kerry
Keystone State should be blue in November.

Wisconsin
Rasmussen: 48-47 Kerry
Just so you're aware, the Florida, North Carolina, Arkansas, and at least one other of those polls were conducted by SurveyUSA, which uses autodialers (press 1 for Bush, 2 for Kerry, etc.) for its polling. This method is considered extremely unreliable, and SurveyUSA's results this year have been wildly inconsistent.

[Rover]

Just so you're aware, the Florida, North Carolina, Arkansas, and at least one other of those polls were conducted by SurveyUSA, which uses autodialers (press 1 for Bush, 2 for Kerry, etc.) for its polling. This method is considered extremely unreliable, and SurveyUSA's results this year have been wildly inconsistent.

I agree. They are the most unreliable from what I've observed for both sides. Their polls are just all over the place. Beyond that, anything with Strategic Vision (R) or Lake Snell Perry (D) you can also dismiss as being partisan, because in truth they generally are.

I always thought Zogby to be the most accurate, and being a Dem hope that's the case this time around, because if you check out his battleground polls he tends to have states going Kerry more so than in some other polls (although he's generally in line with that American Research somethingorother that polls all 50 states).

Bottom line - who knows.

[Jason A]

New polls out today that tell us very little:

Oregon --

Bush by five in Oregon: http://www.rileyresearch.com/Riley%20Report%2010-14-04.pdf (PDF WARNING)

No, make that Kerry by six in Oregon: http://www.portlandtribune.com/archview.cgi?id=Z101804

New Jersey --

Kerry by 10 in New Jersey: http://slerp.rutgers.edu/retrieve.php?id=150-1 (PDF WARNING)

No, make that Kerry by four in New Jersey: http://www.quinnipiac.edu/x11376.xml?ReleaseID=481

[Patman]

This is a question for the "older" folks, but is it just me or have polls never have varied this wildly before? It would seem to me that in the past these things would fluxuate by a few points but these things seem to be erratic now.