Lost amid the babble of election predictions? No kidding. Polls are not coming out
fast and furious every day that show Mitt Romney ahead by one. Or Barack Obama by two. Or Mitt Romney by three, with an error margin of two. Or Obama up by two with an error margin of three.
|You can’t say you weren’t warned.|
Worse, polls are being picked apart. This poll has Obama up by five, but Democrats are over-counted in the survey and so when that is normalized, Romney is up by three. What? Or this poll of a swing state shows Romney up by two, but because this poll uses a sampling of a given number of independents, it really means Obama is up by one. Humans do not understand this nonsense, which means simply this: if you do not understand how the poll concludes anything within two seconds of reading it, you should not read too much into it.
As we have tried to explain before, the most critical piece of information (intentionally left off most polls) is whether the participants are registered voters or likely voters. Polling companies like to use registered voters—if you a registered to vote anywhere (and for the most part, you better be), they have your name, gender, age, address, and phone number. That’s awesome sauce for a pollster: this way, they can tabulate your answer by all those categories and come up with statements like “42% of women in that metropolitan area favor this candidate.” It sounds very scientific, but may be based on only five people.
On the other hand, the likely voter tally requires a lot more work, which is why only a couple of polling agencies do it. A registered voter means that person is registered to vote; but if you look at voting turnouts, you see that only a minority of voters actually bother to go to the polls.
If you aren’t going to vote in the next election, sorry; but who gives a flying jay what you think about anything? If you don’t vote, you thereby concede your voice to someone else.
A likely voter is not only registered, but is someone who is motivated to vote and presently intends to do so, barring any personal emergencies or other events that prevent it. Sometimes, such as with Rasmussen, poll takers are somewhat subtle in their questioning, and ask things like “When thinking about the upcoming Presidential election, how strongly do you expect to go vote that day?” A range of choices follow, and the more likely you are to vote, the more serious the questions get. One supposes that if you say you have no intention to vote, the Rasmussen robot suggests a program on Cartoon Network might be on right now.
Anyway, that is why polls of likely voters should always be taken more seriously than the other type of poll. But even these polls can produce strange results, and this time around we still see that Romney, while ahead, is still quite close to Obama among likely voters.
There is another type of prediction method: modeling. Some folks use historical trends—no President has ever been re-elected when American confidence in the direction of the country is that low. Or, no President has ever been re-elected when independents are undeclared this late in the season. Those are all interesting, and these models do insist that President Obama stands little chance of re-election.
Of course, there is little reason to bet the house on that: just because something hasn’t happened yet does not prove impossibility. Speaking of betting, lots of folks use gambling odds as indicators. These suggest Obama will win re-election. However, when discussing real odds, it is important to remember that for every person willing to place a bet, there has to be someone willing to take that bet: a large pool for Obama also means that there are a lot of people willing to bet Romney will win. Gambling odds are great to measure expectation, mood, and instinct…but they do not predict actual outcomes.
Another model is a purely economic one. Are there factors that correlate well to political events? It turns out, there are. One such model weighs income changes, unemployment, and so on, at local levels and then maps them to the Electoral College. This method, developed at the University of Colorado, is interesting because it can measure the effects of a popular vote, but also tabulate the translation to the Electoral College. For example, it can predict an Al Gore popular vote but also a George Bush electoral win.
And indeed the model did predict that in 2000. The other interesting aspect of this method is that, unlike polls and other forecast models, it has so far been 100% correct since first used in 1980. The Czar understands that this model, when used with simulations of older elections where economic data was not tracked the same way it is today, still proves accurate.
We can always await that first exception, but the track record is very good here. And not just good, but very good—it consistently comes close to the actual Electoral College votes, and even predicted the memorable 1980 and 1992 elections in addition to the confusing 2000 election.
So what does this model say about our 2012 election? It predicts that Mitt Romney will win 53% of the popular vote, and also clinch the Electoral College. Barack Obama will win only 218 electoral votes, and will lose all the swing states.
Why? The model reveals that the whole Obama campaign breaks down when unemployment hits 5.6%. Interestingly, every other factor measured shows that the President could reclaim his job one way or another—which might indeed explain the neck-and-neck polling seen everywhere else. But once unemployment hits 5.6%, the other factors guarantee his loss to Mitt Romney.
The Czar assumes you understand that, with the time remaining before the election, getting the national unemployment rate down to 5.6% is an impossible task. Another model the Czar did not study in detail apparently reiterates this, and indicates that even if Romney or McCain were President right now, the unemployment rate would be enough to kill their chances, too.
Basically, these economic models say the same thing: the President should have focused on the economy like a laser, instead of playing golf, fundraising, and doing daytime television interviews. Now it will cost him his job.