Vanguard Analysis of Presidential Election – For as long as I have been writing on the subject, I have believed that this was 2004 in reverse. I remember believing up until the very end, when it became clear that Cuyahoga County was not going to be enough that John Kerry would defeat President George Bush.
I was wrong in 2004, I had underestimated Karl Rove and overestimated the willingness of people to make a leap of faith. But it was a critical point in my understanding of politics. I have felt since last year that conservatives had made the same mistake as liberals in 2004.
In most ways, 2012 has played out to form, but I started to waver early last week. I told people for the first time that I thought Romney would win and I believed it. That changed on Thursday and Friday.
First of all, I think if it weren’t for Nate Silver a lot of liberals would have jumped off a bridge by now. But he may be right. As he noted, 19 of 20 battleground polls on Friday showed Obama in a very small, mostly statistically insignificant lead. Another two showed ties.
I used to do this stuff for a living, so you would think I would be well versed on it. Polls are based on sample size, probability, and an 85% confidence interval that accounts for standard errors in the sampling. The result is that most polls produce an error of 3% or 5% on either side of the result.
Averaging polls is a common way thought to reduce the error – but it probably does not do so in any meaningful way.
There are some sophisticated ways to fix these problems that I’m not going to bore my patient readers with. So I asked a few people who know far more math than I do how to interpret a number of very consistent but very small differences in polling.
One answer makes a good deal of sense and it is similar to what Nate Silver does – use the polling averages as an indicator and rank the states sequentially to assess how much error would have to occur in order to shift the outcome.
This is similar to what Intrade and Nate Silver do. Intrade puts Obama at the 60 to 70% range while Nate Silver puts it a bit higher at 85% this morning.
Mr. Silver now puts the figure at 85%, but we can dig deeper into his numbers: “My argument, rather, is this: we’ve about reached the point where if Mr. Romney wins, it can only be because the polls have been biased against him. Almost all of the chance that Mr. Romney has in the FiveThirtyEight forecast, about 16 percent to win the Electoral College, reflects this possibility.”
But here’s the problem, most of the polls are based not only on a sampling but a prediction about what the electoral composition will look like, and what if the other polls are wrong in how they modeled it and what if Rasmussen is correct?
You’re telling me the probability that turnout models are wrong is just 16% (now 14.9%)? I do not buy that.
So let’s look elsewhere.
Some are believing that the Superstorm that hit the east coast may play a pivotal role in the outcome here, even though it is not registering in the polls.
The nugget that is more interesting to me at this point is from Jan Crawford of CBS News, who reports: “For eight straight days, polls showed him picking up support. The campaign’s internal polling, which is using different turnout models than most public polls, had him on solid ground in Florida, Virginia, North Carolina and Iowa. He had a slight lead or was tied in Ohio, New Hampshire and Wisconsin and was in striking distance in Pennsylvania, a state Republicans hadn’t won since Ronald Reagan in 1984.
“Those leads in Florida, Virginia, North Carolina and Iowa still hold in the internal polls, campaign sources say, but Romney’s movement flattened out or, as the campaign likes to say, ‘paused.’ Nevada is now off the table, and those neck-and-neck swing states are even tighter.”
Do I really believe in this? No. I think Romney’s internal polling is probably wishful thinking in terms of turnout models, but forget that point for a moment. The key point is the trendline. I agree that for about eight days, give or take, Romney looked like he was gaining momentum and that that has stopped and now flatlined, if not declined.
That, of course, accounts for my changed thinking.
Make no mistake, the Republicans will point to Chris Christie with some justification if Romney ends up losing.
Now, in fairness to Governor Christie, his state has been utterly devastated, he needs federal resources and in the aftermath of the largest disaster in state history he would be shirking his duties to even consider politics.
“I have no idea, nor am I the least bit concerned or interested,” Governor Christie said. “I have got a job to do here in New Jersey that is much bigger than presidential politics. And I could care less about any of that stuff.”
“I have a job to do. If you think right now I give a damn about presidential politics, then you don’t know me,” he would later add.
And the governor was quick to rightly praise the President: “He’s done — as far as I’m concerned — a great job for New Jersey.”
I will add this point because it may or may not weigh in, in the voters’ minds – there have been two large crises in the last two months of this election and both times, the instinct of some in Romney’s camp was to play politics.
They held their tongue better this time than Benghazi, but the leaks coming out about Governor Christie made their feelings very clear.
The fact that President Obama got to use his incumbency advantage, while Mitt Romney was a spectator, very much plays a role in the change in the dynamics in the last week.
It is the chief reason why I once again believe he will narrowly prevail on Tuesday.
Still, I think Mr. Silver’s much too sanguine on his modeling and discounts too much the probability that the modeling is wrong. We learned in 2004 that candidates can actually artificially alter voter composition – something that Democrats have tried to replicate without nearly as much success – though I must say the California online voter registration model looks promising.
So the bottom line is that, while I don’t think it’s too close to call at this point, I am not nearly as confident as Nate Silver in the turnout models and so right now I think 65 to 70 percent is about right.
Tomorrow I’ll offer a more philosophical analysis of the race.
—David M. Greenwald reporting