Friday, April 29th, 2016

Making the Election Day prediction sausage


Not that long ago, a local public official complained that I was bringing up something from the past. That prompted a response from me to the effect that the oracular function of the newspaper is usually restricted to the horoscopes, and that most of what appeared in a newspaper occurred in the past. I should probably heed my own words and leave the analysis for after the election. As I begin this, I’m reminding myself that I predicted former Wisconsin Supreme Court Justice Louis Butler would defeat now-Justice Michael Gabelman.

Making pre-election observations even more difficult this year is the wide variance in the quality of the polls. There are several ways to respond to this. One is to only look at the polls you like. Another is to immerse yourself in the polls you don’t like and wallow in self-pity. Or you can try to find some value in each poll, hoping that by reading a poll’s entrails you can discover a divine revelation.

I’m a believer in polls, but I’ve had my faith shaken in recent months. My doubts began during the recall election when pollsters were producing results that just didn’t make any sense even prior to a single ballot getting cast. The Republican presidential primary polling did not inspire faith, either, nor did Democratic polling firm Public Policy Polling’s comment after the primary that they were grateful the primaries were over because they had done so poorly.

I’m beginning to wonder if we are reaching the end of the effectiveness of telephone polling. Michael Barone, an analyst I greatly admire, said on Fox News election preview program on Sunday that Pew Research were only able to complete 9% of their phone surveys in 2011. The difference of those answering a poll and those not answering a poll is now an independent variable that can potentially affect the outcome of the poll. If the fate of Schrödinger’s cat is dependent upon us making the observation when we open the box, our ability to make the observation is increasingly dependent on whether the cat jumps in the box.

We saw a decline in the available polling data with exit polls. Brian Fraley, the managing editor of, has explained the wild inaccuracy of exit polling by saying that when Democrats are asked for whom they voted, they’re more than happy to answer. When a Republican is asked, they’re more likely to say, “F— you, it’s a secret ballot.” Then there are the additional problems of the kids asking the exit poll questions oversampling Democrats for a variety of reasons and the locations of the exit polls being more favorable to Democrats, and you have a polling tool that is only possibly useful in post-election analysis after the results are adjusted by the actual election results.

With all of that said, I give you Nate Silver of the New York Times:

Mr. Obama would be almost certain to win the Electoral College if he won the popular vote by a percentage point or so.

Instead, the model estimates that Mr. Romney would need to win the national popular vote by about one percentage point to avert a tossup, or a loss, in the Electoral College. A tied popular vote, as Mr. Romney’s better national surveys now indicate, would likely yield an unhappy outcome for him.

Mr. Romney would not be in much danger of losing the Electoral College if he won the popular vote by more than about 1.5 percentage points. For example, he would be about a 95 percent favorite in the Electoral College if he won the popular vote by two percentage points, according to the forecast model.

Silver now says President Barack Obama has an 85% chance of winning the electoral college, and only a 50.5% chance of winning the popular vote. It’s hard for me to square those numbers with the above statement. and is projected to win 50.5% of the popular vote (see comments). However, such a narrow margin of victory, would certainly cast doubt on so high of a percentage of a likely victory, especially when Gallup’s likely voter poll shows Romney winning by 1% as of Monday.

On the other hand, in an op-ed for the Daily Caller, Sean Davis re-creates Silver’s model walks the reader through the flaws, and at National Review’s Josh Jordan also does a great job of explaining the problems with Silver’s model.

But let’s take Silver at his word that Romney would win the election if he wins the popular vote by 1.5%.  The Gallup forecast for Tuesday shows Republicans with a +1 partisan advantage among likely voters, a number far more favorable to Republicans than recent public polls. If that number is correct, Romney’s 15% chance of winning the electoral college looks like a long shot that pays off. Or to look at it another way, the coin flip of the popular vote actually lands Romney’s way and he wins enough to win the electoral college.

As I wrote the other day, pollsters are having a hard time figuring out just what is a “likely” voter this time around, and the samples are showing a Democratic voter identification near 2008 levels. That just does not work with Gallup’s analysis of a final electorate. The electorate promises to be more favorable to the Republican nominee than the electorate in 2004.

Experience teaches us that undecideds tend to break towards the challenger in most but not all races. That’s in part because the incumbent is a known quantity, and if someone is undecided it’s largely because they are not inclined to support the candidate they know. We also know that independents are breaking for Romney in most polls. Former Carter strategist Pat Caddell is predicting that the election will be similar to 1976, when the challenger was able to beat an incumbent president in a narrow race, in large part because the electorate is not willing to support the incumbent.

Meanwhile, Michael Barone, the election knowledge guru, is not seeing the narrow race most other pundits see. Barone  sees Romney winning 315 electoral college votes. Karl Rove, an obviously more partisan analyst, sees Romney winning with 279 votes. Given the behavior of the two campaigns and the relative intensity advantage of Republican voters, I would be hard-pressed to disagree with either of them.

So let me offer some last thoughts before I post the map:

I fear election night will go long, especially if the race comes down to Ohio. That state will not begin counting provisional ballots until ten days after the election. The only way it will be a short night is if Romney wins Pennsylvania.

We will see early exit polls that are leaked to the press that show Obama ahead. Ignore them. They’re worthless data that cannot be evaluated independent of knowing the actual election outcome.

Just as Ohio will be decided by Hamilton and Cuyahoga counties, the Fox Valley and Brown County will play a big role in Wisconsin. In Pennsylvania, the suburbs of Philadelphia will the area to watch.

All that said, I’m predicting Romney will win the election with a minimum 275 electoral college votes. He could win as many as 331 votes, although I’m really leaning towards 295 votes.  He’ll win the popular vote with 51%-48% of the vote.

I’m more optimistic about Romney winning Pennsylvania and Ohio than I am Wisconsin. However, it’s entirely likely Romney could lose one of those states and win Wisconsin, and also win New Hampshire. However, if you look at the total, as long as Romney wins Colorado, which is increasingly likely, he only needs to win Ohio or Pennsylvania, not both. (I’m assuming that Florida, North Carolina and Virginia all go to Romney.)




Be Sociable, Share!

Print this entry