The Only U.S. Election Prediction You Can Trust - i.e. MINE

There must be something about living upside down that addles the mind…

we shall see…

It all depends on the electoral vote.  A narrow popular vote can translate into a large electoral landslide.

Hi David, I suppose it depends on whether the betting is tied to the polls. If so, then I wouldn’t put too much faith in them since the only polls showing O leading nationally are those with ludicrous samples (D+8 for example in the latest National Journal poll).

Let’s start with this:  Barack Obama in hopey-changey landslide mode took the popular vote 53 to 47% amidst Bush fatigue, war fatigue, a Republican candidate who routinely commented that he doesn’t really like movement conservatives in the Republican party.  (Hugh Hewitt calls John McCain a great American,  an ok senator, and a terrible Republican.)

Having seen 4 years of the autocratic, incompetent Obama, who’s piled up $6 Trillion in new debt, produced the weakest dollar since Jimma Carter so food and fuel prices have blown up and a tepid “recovery,” I can’t see how Obama doesn’t lose at least 3% of the popular vote.

Romney clearly isn’t an optimal candidate from a conservative perspective, but I don’t think any Republican can look at the damage Obama has done to the structural conditions of the American economic system and national defense, let alone his hard left social policies, and not want him GONE. 

He can pontificate from some endowed chair at some Ivy League school happily for the rest of his days or go on a world tour and apologize for what racists Americans are for voting him out of office for the rest of his days.

David, if Obama somehow gets re-elected, Australia will see an increase in boat people . . . from the United States.

The key factor here is “UK” betting site.  When I lived overseas for several years I discovered that people not in the US really believe what they see on CNN.  Things look quite different when you live here and talk to and observe Americans in their own environment.  Based on Karl Rove’s map in the WSJ this morning, Romney will win and lot of UK bettors are going to lose their money.

Newbie Anglican,

You should think long and hard about fleeing to the Land Down Under.  With “Red” Julia Gillard at the helm (and “red” here is not referring to her hair color), the Aussie state is headed down the same path as we are here in Obamaland.

Yep.  Here it is, according to Reuters:  90% of northern Europeans would vote for Obama.  They consider him a center-right sort of guy.  Here, he’s hard left.  These UK bets are placed by people who don’t understand American politics well.

When I go to the track, I check the horses by their records, their jockey, their musculature, coats, physical appearances on race day, and their spirit. I don’t think I’d place any bets on this race.

David brings up a good point.  But I think that the problem with his prediction is that the people making the bets might be using incorrect data to inform their betting.  The data used is the polling.  And one thing that we do know for sure about the upcoming election results is that we will learn something about polls.  See this article on how the major issue in polling right now is coming up with the correct “turnout model”.  Most of the media polling is adopting the Democrat expectation, which predicts a larger Democratic turnout then 2008.  Another good article on this is here.  I think that most people who are betting on this election will look to the major media polls.  But, as these articles point out, these polls may be employing misleading turnout models.

So the real issue is not what the polls seem to be showing (and which is driving the betting) but rather on what you think the underlying turnout model is.

The polling models are a problem.  Many of the polls, both state and national, are assuming something like the 2008 turnout, in numbers and in party identification.  If it’s more like 2010, which seems likely, those assumptions are poor.  Another problem is getting a valid random sample.  I saw an article at PJ Media about this; it’s been a few days, so I don’t have a link.  Imagine filling up a big cement mixer with a mixture of red and blue marbles.  You drive it around for a while to really mix the marbles.  Then you draw out a bucket of 1000 or so marbles and count the colors to get an estimate, with a 3% margin of error, on how many of each color there are in the mixer.  But wait!  Nine out of every ten marbles refuse to be counted!  That’s pretty much what the pollsters are facing.  They can’t get people to answer the phone, or to answer their questions.  Is the population which refuses to pick up the phone statistically similar to the 9% they finally talk to?  They don’t know.  Are people more or less likely to vote, or people inclined to vote for one party, answering the phone or refusing to take calls?  They don’t know.  I think the large Gallup poll showing consistently Romney +5% is valid, but really, neither Gallup nor I really know.

I hope we won’t see any risky exit poll calls on Tuesday night.  They won’t know what exit poll data to use until they see the total vote so they can properly adjust for party identification.

Katherine:

I think that they can probably create models to control for the 9/10 people who won’t answer the polling questions at all.  What the issue really boils down to is this - if you arrange your turnout model with a prediction of a turnout with a significant Democrat advantage, then it should be surprising that it predicts that the Democrat will be elected. 

Consider the following set of turnout models:

Model 1: R-30; D-38; I-32
Model 2: R-32; D-36; I-32
Model 3: R-33; D-35; I-32

Assume for this illustration that all Republicans vote for Romney and all Democrats vote for Obama.  Assume that the independent share is the same, and assume that Romney wins the majority of the independent vote - say the independents go 18 to Romney and 14 to Obama.  These are the raw data.

If the media assume turnout model 1, then the poll will show Obama 52% to Romney 48%.  If model 2 is assumed, then it will show them tied.  If model 3 is assumed, then it is Romney 51% to Obama 49%.  All with the same raw data.

We know that the polls showing Obama leading or tied are all assuming very high D turnout models which assume a turnout like 2008.  The raw polling data suggest that the crossovers will probably cancel each other out, and the independents are breaking fairly strongly for Romney.

Consider three possible outcomes in terms of voter turnout:
1) Best reasonable case for Obama - voter turnout resembles 2008.  Polls suggest in such a situation a close election with Obama squeaking through.
2) Best reasonable case for GOP - voter turnout resembles 2010.  With such a turnout model, all the polls would show a major Romney victory.
3) Middle-of-the-road turnout - voter turnout resembles 2004.  According to all the polls, such a turnout model would show a narrow but comfortable victory for Romney.

I suppose we will see what voters turn out when all is said and done.  Aside from the turnout question, the other big question will be whether Romney keeps winning the independents and whether either of them begins to open a crossover lead.

I agree that the polling models have flaws. I also note that serious spread market betters know this and adjust for it. The site I’ve linked to is interesting not least for the level of analysis that goes into this sort of thing.

So I still stand by my thesis. The people who are serious about putting their money where their mouth and their detailed analysis are, are putting that money on Obama.

We’ll see in 5 days! I look forward to my blogging colleagues all eating their various hats

From Fox News:  Obama seems to be getting a clear Hurricane Sandy bounce. 
http://www.foxnews.com/opinion/2012/11/02/sudden-danger-signs-in-polling-as-election-day-nears/

Karen B., I tend to take Dick Morris with a grain of salt in general.  There is no way to know if these are one-day wobbles in the polling averages or real changes, and one would think that polling will have been more difficult than usual in the northeast states affected by the storm.

Anecdotally, Romney is drawing huge enthusiastic crowds in the final days of the campaign, and Obama’s have been much smaller.  That may not mean any more than all these polls.  I’m not sure anyone knows what will happen.

I count on Michael Barone’s judgment far more than Dick Morris’s.  Barone’s prediction:  Romney 315, Obama 223 electoral votes.

Has anyone considered that it would be a relatively inexpensive matter for a political advocacy group to buy up shares of their candidate as a way of swaying public opinion?

not particularly pleased to be proved right