Today in polling: Obama improving?

By Frank Moraes

The big news is that Obama's poll numbers keep improving. On October 4, Nate Silver predicted that Obama had an 87.1% chance of winning the electoral college vote. This was his highest chance during the campaign because the first presidential debate happened on October 3. (Actually, the causality is not so clear; it looks like Romney was improving before the debate.) That was the start of the liberal freakout, where Silver's prediction went down steadily until reaching its nadir on October 12 with a 61.1% chance of winning. I'm sure you remember that week: it was a dark time for us all.

But since then, things have turned around. Obama has steadily increased his chances of winning. As of yesterday, Silver gives Obama a 74.4% chance of winning the electoral college. The popular vote is much tighter, but we see the same dynamic. On October 4, Obama was predicted to win 51.6% of the popular vote. This number dropped to 49.8% on October 12. Note: even on that day, Romney was only predicted to win 49.1% of the vote. So even at his lowest point, Obama was still looking okay. And it has only gotten better. Nate Silver now predicts that Obama will win 50.3% of the popular vote.

None of this means that Obama will win re-election. The race could change in the next ten days. And Republican voter suppression efforts could be highly effective. And the Frankenstorm could upset everything. And Silver's model could be all wrong. Perhaps he doesn't have sufficient height and weight to accurately predict elections. (See below.) But right now, the Obama campaign has to be a lot happier than the Romney campaign.

Undecideds

Last month I wrote about undecided voters and how they will vote. Whoever is winning always claims that this group will break heavily toward them. I've been in this situation many time. But in my experience, they generally go pretty much as the decided voters. It is pretty much never the case that the undecided votes are anything other than a fifty-fifty split.

John Sides over at The Monkey Cage looks at this question quantitatively, "How Will the Undecideds Break?" He finds that they will go 50.1% for Obama. However, looking only at those who claim they will definitely vote, they skew toward Romney with 56%. But even with this, it would add only 0.4% to Romney's total. The final prediction is that this will not be enough to give Romney the popular vote.

Bizarro World

You probably don't know who Dean Chambers is, but I'd bet you know about him. He's a little brain who decided that all the bad polling for Romney before the first debate was one vast left-wing conspiracy. So to counter it, he started the website UnSkewed Polls. Even though I feel pretty good about the polls going into this election, I'm willing to admit that it could go either way. The polls are indeed all over the map. But not according to Mr. Chambers. Using his unskewing procedure, every poll now shows Romney with a handsome lead over Obama. This morning, his poll average is Romney +5%.

According to Chambers, it isn't just the polling that has been skewed. The election models are in on the conspiracy. Yesterday at Examiner.com (more or less the dreadful San Francisco Examiner), he wrote an article attacking Nate Silver and his model. Remember that, if anything, Nate Silver is far more bullish on Mitt Romney than all the other models. But that doesn't matter to Chambers. Chambers knows Silver's model is wrong. How? Because Nate Silver is short and skinny:

Nate Silver is a man of very small stature, a thin and effeminate man with a soft-sounding voice that sounds almost exactly like the "Mr. New Castrati" voice used by Rush Limbaugh on his program. In fact, Silver could easily be the poster child for the New Castrati in both image and sound. Nate Silver, like most liberal and leftist celebrities and favorites, might be of average intelligence but is surely not the genius he's made out to be. His political analyses are average at best and his projections, at least this year, are extremely biased in favor of the Democrats.

Mostly, the article is just a proclamation that Nate Silver must be wrong because Dean Chambers doesn't like his results. In particular, he complains that Silver's model weights some polls higher than others. This is true. For example, he weights Rasmussen rather lightly because they have been shown year after year to be wrong. Apparently, Chambers doesn't understand the difference between weighting data according to their value and weighting them just because you don't like what they're telling you.

Chambers ends his article with another unsubstantiated claim:

[Nate Silver] claims to have been highly accurate in predicting the 2008 election results, and perhaps he was. But it's highly unlikely his current methods and projections will have the level of accuracy unless he changes them quite a lot between now and election day. The race has shifted profoundly in favor of Mitt Romney while Nate Silver is still projecting an Obama win. Unless he changes that, the credibility he earned in 2008 will be greatly diminished after this year's election.[1]

It isn't clear what exactly it would mean for Nate Silver to be right about this election, given that he is just providing probabilities. But let's suppose that Romney wins by a sizable margin. I'm sure this will cause Silver to go back and change his model. On the other hand, if his model turns out to be right and Obama wins by a comfortable margin, will Chambers admit that he was wrong? If I know conservatives, he will not. In 2016, he will still be predicting ridiculous Republican wins.

**********

[1] Note: I fixed about five grammar errors in the sections I quoted from Chamber's article. The awkward syntax was left in. I did this rather than throw in a bunch of [sic]s. I understand this is the Internet, but he's writing for kind of a real publication. Geez!

(Cross-posted at Frankly Curious.)