Friday, October 31, 2008

The Law of Gravity or What Happens to Dead Cats After They Bounce



PROJECTED ELECTORAL VOTE: OBAMA/BIDEN 356, MCCAIN/PALIN 182
(unchanged)
PROJECTED POPULAR VOTE: OBAMA/BIDEN 52.0%, MCCAIN/PALIN 46.0%
(Obama up 0.2%, McCain down 0.2%)
TIME UNTIL FIRST POLLS OPEN: 3 DAYS, 11 HOURS, 54 MINUTES

Actually kind of a slow day for state polling data this late in the election. No real surprises or major moves. McCain loses ground in our tracking polls for the second straight day, he has given back most of the ground he had gained earlier in the week. In fact, if you look at our projection, it has operated in a very tight band for a long time with Obama's lead being between 4.6% and 7.6% every day from September 27th until now.

In terms of the state moves, there is not much of consequence. Obama gains a little ground in the Dakotas and Montana, but there are very few electoral votes in play there.

McCain, in the remaining 4 days of this campaign (really less now, but we'll give him the benefit of the doubt) would need to close 1.5% per day and flip 22 electoral votes every one of the next 4 days.

On Intrade McCain is now a 6.1:1 underdog, slightly worse than his 5.9:1 position yesterday but a lot better than his worst showing of 8.3:1 a few days ago. Honestly, 6.1:1 is starting to look like a rip off as McCain runs out of time.

One positive news of note for McCain is that he has now reduced Obama's lead in Pennsylvania to under 10% in our projection, for the first time in a long time. It is still not close enough to even be considered a fringe battleground and I doubt he can win it, but it is some confirmation that his strategy there is having some effect.

STATE CHANGES
None

KEY BATTLEGROUNDS
Montana -- promoted from serious to key battleground
North Dakota, Missouri and Indiana remain as key battlegrounds

SERIOUS BATTLEGROUNDS
North Carolina and Florida remain as serious battlegrounds

SUBSTANTIAL BATTLEGROUNDS
Arizona, Georgia and Ohio remain as substantial battlegrounds

FRINGE BATTLEGROUNDS
South Dakota -- added as fringe battleground
Nevada, Colorado, Virginia and New Mexico remain as fringe battlegrounds

On Polling Accuracy
With all of the discussion of polling errors and divergent polls and Bradley effects and reverse Bradley effects, I thought it would be valuable to take a look at historical poll accuracy in presidential races. The result might surprise you -- national polls have been incredibly accurate in every election.

The link below will take you to a site that analyzes final polls results versus general election results from 1936 to 2004.

http://www.newsu.org/angel/content/aapor_polling07/1a_why/accuracy.php

Here are the bottom line conclusions:
In 70 years, the average of polls have never been off by more than 6% on the final spread.
Since 1960, they have never been off by more than 3%
Since 1984,they have never been off by more than 2%
In 2004, they were off by less than 1%, with TIPP, CBS News and Gallup all within less than 0.5% and Harris, Zogby, ABC News, The Washington Post and NBC News within 1%. No major poll was off by more than 4%.

Could this year be different? Sure -- the dynamics are complicated. Figuring out African-American turnout, youth turnout, it's all difficult. And the Bradley Effect is still an unknown.

But the point is, more often than not, if it looks like and quacks like a victory for one candidate, it probably is.

And it looks an awful lot like a victory of Obama on Tuesday from here. The one thing McCain can take some comfort in is that we aren't up to the final polls yet, so he still has some time to close -- but that time is dwindling.

Battleground Early Voting Update
(# of votes, % of 2004 early vote, % edge Democrats-Republicans, % African American)
Colorado -- 1.3M, 125%, 2%, N/A
Florida -- 3.4M, 122%, 7%, N/A
Georgia -- 1.8M, 264%, N/A, 35%
Indiana -- 0.4M, 157%, N/A, N/A
Iowa -- 0.4M, 85%, 20%, N/A
Nevada -- 0.5M, 113%, N/A, N/A
New Mexico -- only 1 county with data but turnout is very high there
North Carolina -- 2.1M, 190%, 23%, 27%
Ohio -- incomplete data but with the 13 counties we do have -- at least 0.5M, 87%, N/A, N/A
Virginia -- only 1 county with data -- it is higher than 2004 but not extraordinary

Early voting ends most places on Saturday, so by Sunday (or Monday at the latest if some states don't report until the next business day) we will have a final picture of early voting.

It's kind of silly for me to give a "time until the first polls open" when twenty-some million have already voted.

The numbers in Georgia and North Carolina continue to be staggering and very favorable to Obama. Other states are also up but without the pronounced Democratic edge, although the edge is growing in Florida (it was about even by party two days ago and now favors the Democrats by 7%.)

Trail Dust
The schedules are now completely insane and I expect them to stay that way for the last weekend.

Today:
McCain -- 3 more stops in Ohio (he has spent two solid days here and 6 stops)
Obama -- 1 stop in Indiana and then home to Illinois to take his daughter's trick-or-treating
No word on VP candidate whereabouts today.

Tomorrow:
Obama -- Missouri then Nevada.
McCain -- Virginia.
Biden -- Indiana then Ohio.
Palin -- no word yet

Obama has also shared his intention to make his final campaign stop on Monday in Virginia, the state that has been the tipping point for him to get to 270 for much of this campaign.

No word on where McCain will finish, but the McCain camp's strategy at this point appears to be to pound the heck out of Ohio and Pennsylvania, so I wouldn't be surprised for the last stop to be in one of those two.

Getting down to the end folks....get out and vote early if you can. If not, please vote on Tuesday.

As much of a political junkie as I am and as much as I've enjoyed the past two years, I'm ready to have an election already!

No comments: