Tuesday, October 28, 2008

The Race Is Getting Closer -- Maybe? Kind of?

(Obama down 0.4%, McCain up 0.4%)

Well, it's officially gotten confusing again. A little bit of poll chaos isn't rare in the week leading up to an election, so let's try to make sense of the madness.

First, for the second consecutive day, McCain has narrowed Obama's popular vote lead by 0.8%, exactly the amount he would need to continue to narrow it each day to be dead even by next Tuesday. This is according to the tracking polls I am using: Gallup, Hotline, Zogby, Battleground, IDP/TIPP and Rasmussen. Two days isn't quite a convincing trend, but if this continues tomorrow, we'll have a real race on our hands.

But, at the same time, just yesterday a Pew Research Poll was released showing Obama with a 15% lead. Since we are only using tracking polls because they are consistently released every day, this doesn't factor into the average. But it begs the question -- why is this result so out of line with the tracking polls?

Also, no state has swung during this two-day run for McCain and frankly, most don't appear to be tightening in the same way as the national polls. In fact, projecting the percentages we have for the states across the 2004 vote proportions would yield us the same 7.2% national lead we had two days ago -- that means we have a 1.8% deviation between the two.

So -- are the tracking polls noise? Is the Pew poll just wrong? Are more states in play than we think?

Let's start by examining our battlegrounds in detail:

Montana -- upgraded from serious to key -- we have 3 state polls from the past few weeks
with a spread from +4% Obama to +4% McCain with the most recent from 10/25 from Mason-Dixon showing McCain +4%.
Conclusion: Close, but McCain leading
Indiana -- no change -- 4 recent polls with a big spread from +10% Obama (Big 10) to +6% McCain (Zogby) -- the mathematical averages we have give Obama a slight lead here, but all polls are at least two days old
Conclusion: It more likely than not that McCain has overtaken Obama ever so slightly
Missouri -- no change -- very tight spread among 5 recent polls from +2% Obama (Zogby) to +1% McCain (Mason-Dixon) -- all polls are at least 2 days old
Conclusion: If there really has been 1.8% movement in the past two days, it seems more likely than not that McCain is ahead by a hair now
North Carolina -- no change -- past weeks polls range from +4% Obama (Zogby) to +1% McCain (Rasmussen) -- both the polls I mentioned are exactly two days old
Conclusion: Even with a 1.8% adjustment vs. the most recent polls, Obama still leads by a hair

Florida -- no change -- just released Bloomberg Poll shows Obama up by 7%, other two recent polls have him up by 4% and 5%
Conclusion: Obama still leads here

Nevada -- downgraded from serious to sginificant -- Just released Rasmussen poll has Obama up by 4%, other recent polls have him at +10% and +4% (Suffolk and Zogby)
Conclusion: Obama still leads here
Georgia -- no change -- McCain leads by 1% in just released poll (Insider Advantage), other recent polls have showed him up by 5 to 6%
Conclusion: McCain is ahead, although the margin is in question and the early voting demographics must have him concerned

Arizona -- downgraded from substantial to fringe -- McCain up by 5% to 8%, depending on which poll you pick
Ohio -- no change -- Obama up by 4% and 9% in polls released today
Colorado -- no change -- Obama up by 4% and 8% in polls released yesterday
Virginia -- no change -- Obama up by 4%, 7% and 9% in polls released yesterday
North Dakota -- no change -- no recent data to even hazard a guess with
Conclusion: There are no states that I believe have changed hands here

So, if we take the case that the tracking polls are right, it is entirely possible that Indiana and Missouri would flip back to McCain giving him 185 electoral votes and leaving Obama with 353. So, wouldn't change the outcome that much, but again, if McCain continues at this pace, it might get interesting again.

But, ok, this all assumes that the tracking polls are right and the Pew polls is wrong -- how do we know that's true? We don't. But the Pew poll has some oddities in the detail that make me suspect the tracking polls are closer to correct. The Pew poll has a 24% Republican sample vs. about 30% for most other polling firms, which would indicate some sample bias, if you believe the 30% is correct. This alone (assuming those Republicans primarily vote for McCain) would bring the poll down to a 9% edge for Obama if adjusted.

The Pew poll also does one other thing the tracking polls don't do -- it names the list of third party candidates. Interesting point here -- the difference in the Pew number is not so much that Obama is higher (his 52% is within the range of other polls) but that McCain's number is lower and the "undecided/other" bucket is higher. Could it be that voters that vote for McCain when presented two choices go for a third party candidate when presented with names like Bob Barr? I don't know. I do know that Ralph Nader was at 2 to 3% in a lot of polls going into 2004 and got almost no votes. So, I'm inclined to stick with the tracking polls but I'll keep watching.

The truth may be somewhere in between.

Obama's Second Convention
Obama has half an hour of primetime TV at 8 PM Eastern on the major networks tomorrow. This time will be unedited and unrebutted, much like a convention. Ah, the wonders of having a massive campaign war chest.

Will this give a convention-style bounce? Doubtful -- I have to imagine fewer people will watch. But Obama doesn't need a convention-style bounce -- all he has to do is freeze the polls for a few days and he wins -- and it might do that.

I'm certainly curious enough to be tuned in and I imagine a lot of people will watch, just as they did when Ross Perot bought his half hours on TV back in 1992.

Quick Early Voting Update
No real swings in the demographics, so let me just run down early votes cast and percentage of 2004 early votes cast for the battleground states:
Colorado -- 0.8M, 79%
Florida -- 2.1M, 75%
Georgia -- 1.2M, 180%
Indiana -- insufficient data
Iowa -- 0.3M, 79%
Nevada -- 0.3M, 78%
North Carolina -- 1.4M, 129%
Ohio -- insufficient data

As before, the only real states of note are Georgia and North Carolina. North Carolinans have already cast 40% as many votes as they did in TOTAL in the 2004 race.

Trail Dust
Obama is in Virginia and Pennsylvania today. He heads to North Carolina and Florida tomorrow.
McCain is in Pennsylvania and North Carolina today. He heads to Florida tomorrow.
Biden is in Florida today and will be there again tomorrow.
Palin is in Pennsylvania today and is headed to Ohio and Indiana tomorrow.

Expect to continue to see a rotating list of Virginia, Pennsylvania, North Carolina, Florida and Ohio as we close this thing out. Obama wins with any 2 of the 5.

No comments: