The Tech - Online EditionMIT's oldest and largest
newspaper & the first
newspaper published
on the web
Boston Weather: 59.0°F | Fair
Article Tools

A wise man once said, “It’s difficult to make predictions, especially about the future.” But judging by the enormous volume of guesses made in the national print, merely making a prediction seems to take no effort at all. Bewilderingly, this election cycle has predictions that run the gamut from an imminent Romney victory, to a completely tied race, to a foregone conclusion for Obama, all made with straight faces by normally reasonable people.

Why the huge spread in the odds-making? The answer is that pundits are taking fundamentally different approaches to the poll numbers that have been observed over the past week. Here are the three major prevailing views on how to interpret the information we have.

The view of the mainstream media

The mainstream media benefits from saying it is a close race. And by just looking at the headline national poll figures, you’d be hard pressed to argue otherwise. The most recent polls from USA Today, CNN, and Politico show a tied race. Gallup and Rasmussen have the race one point in Romney’s favor, while NBC and ABC have it one point in Obama’s favor. Pew has Obama up by 3 points, but most major national polls conducted in the past few days have put down the race as a dead heat, anyone’s game.

The 24-hour news organizations have, for the most part, decided not to look this gift horse in the mouth. The standard newscast these past few days has been a simple three-step: The anchor brings up the latest poll indicating the race is close, the discussion is kicked over to a red-and-blue shaded map of the United States where analysts jab at all the states that are conceivably in play (and Ohio twice for good measure) before sending it back to the anchor, who registers their amazement with all the state poking and reminds viewers to turn in Tuesday night for the best coverage of the most amazing election in the history of democracy.

National polls have it tied, it must be tied.

Conservative counterargument

Many conservatives think the national polls are underestimating their man. How can it be that in all of the national polls that show the race tied, Mitt Romney is leading among independents by double digits? They take issue with pollsters’ subjective judgments on what distribution of party identification, race, age, and gender is likely to turn out on election day. If the make-up of the electorate on November 6th is really 41 percent Democrat, 30 percent Republican, and 29 percent Independent, then a tied race would make sense. But this distribution is backward-looking, conservatives argue, too influenced by the anomaly of 2008, where Democrats benefited from a large enthusiasm gap. To conservatives, the likely voter model employed by most polling agencies is patently flawed when it gives Democrats an 11-point advantage in a country that has long tacked to the right nationally. To boot, conservatives see most undecideds breaking toward Romney; a majority of voters identify the economy as their number one issue and a similar majority cite Mitt Romney as the man they trust most to handle the economy.

Romney is ahead among independents, he must be winning.

The liberal rebuff

Liberals agree with conservatives that the headline national numbers are too simplistic. But their claim is not that the likely voter models being used are wrong (if they are wrong, they’re just as likely to over-estimate Romney as they would Obama). Instead, they call attention to the extensive state-by-state polling that should be brought in to complement national polls.

The inclusion of state polling highlights the difficult electoral path Mitt Romney has to walk. His most likely winning combination is taking Florida, Virginia, Colorado, and Ohio. If Obama takes a single one of these states, Romney’s path to the presidency becomes almost impossible. Colorado could conceivably be replaced by Iowa or New Hampshire, but losing any of the other three would mean some large upset in Wisconsin or Pennsylvania would have to occur. Romney is already behind in the state polling across every one of these states with the exception of Florida (where he has a very slight lead), so it seems natural to give Obama 85 percent odds of winning the election, as Nate Silver, the election model runner for the New York Times, does.

This might suggest that the possibility of Romney winning the popular vote but losing the electoral vote is very large. But here we get to the real reason why the state polls are game changing: The state polls, in total, find Obama a favorite nationally. That is, if you took each of the state polls, weighted them by expected turnout in that state, and averaged them to get an estimate of the popular vote distribution, you wouldn’t see a tie, you’d see Obama ahead by as much as three points. And in fact, the argument continues, this bootstrapped estimate of the national picture is a more relevant one than that provided by the national polls, because the state polls, focusing as they do on battleground states, have greater accuracy where it matters most.

As for the conservative claim that the likely voter models being used are skewed, the response from liberals is simple: Pollsters are a lot smarter than they are being given credit for. They aren’t just looking at 2008 and copy-pasting a turnout distribution — these are professionals who make these judgment calls for a living. And the reason Republicans are given less weight in their polling is because many conservatives, not fond of the GOP’s brand, have switched identification from Republican to Independent. If they’d maintained their party identification, pollsters would give a higher weight to Republicans in their models. But then again, if that were the case, Romney wouldn’t be leading among independents.

National polls have it tied, state polls have Obama ahead. Average the two, and Obama must be winning.

For what it’s worth, on Intrade, an online prediction market where participants have staked tens of millions of dollars on the outcome of presidential election, Obama holds a stable 2-1 advantage over Romney.

It is impossible to tell which view is correct, and I don’t know who is going to win tonight. But I do know that this one is worth watching.