Polling Analysis and Election Forecasting

Month: July 2012

Q2 GDP Growth Released

The BEA released their first estimate of Q2 GDP growth today: 1.5%. This has been the key piece of missing information for the structural forecast I use as the starting point for my model. Until now, I’ve been relying on the Q1 growth rate, which is less strongly correlated with the election outcome. Starting with the next site update (either today or tomorrow), I will switch over to the model based on Q2 growth.

It’s helpful to have this new data point, but don’t expect a huge adjustment to the forecast. The Time-for-Change model using Q1 growth predicted that Obama would receive 52.6% of the two-party vote. With Q2 growth coming in at 1.5%, the baseline forecast falls all the way to… 52.3%. And this is before factoring in any adjustments from current polling.

In fact, the next update is also going to incorporate some important new polls. Two in particular stand out. First, we get a long-awaited reading in highly-competitive Missouri. The last polls in that state were done in late May/early June, and suggested a slight Romney lead. This latest survey indicates Romney is actually a few points farther ahead. Second, a poll in Illinois corroborates expectations of a large Obama lead there. So even though the forecast in Illinois won’t change that much, our certainty in it will increase considerably.

Stay tuned…

How to Follow the Campaign

Campaign reporters like to speculate about how events today will affect public opinion tomorrow. It’s the main way they justify why some new campaign tactic, news item, commercial, misstatement, gaffe, etc. is “important” or “matters” for the election. I appreciate that this is their job. But if public opinion is what we care about, why not just look at it directly? We have the data. So what are the polls telling us?

State-level presidential polls are now coming in at a rate of about one or two a day. And what’s happening is: Nothing. Voter preferences have been almost completely stable for nearly three months. The Obama lead we picked up on at the end of June is holding. The poll aggregators over at HuffPost Pollster and TPM are seeing the same thing. Nate Silver called it “The Incredibly Steady Presidential Race”. He’s right.

It wasn’t like this in 2008. In 2008, polling data showed a series of long-term, up-and-down trends in voters’ preferences for Obama vs. McCain. Support for Obama gradually declined all summer (leading to a bunch of “worried Democrat” stories like this one – really, one of my favorite genres of political reporting), then picked up around the Democratic convention, dropped off again following the Palin VP pick, and finally increased after the September financial crisis. Nothing like that has happened so far in 2012.

The question, of course, is will it? We still have three months to go. I’d be shocked if nothing happened between now and November. John Sides wrote about the primary factors to watch. But there are three considerations to keep in mind. First, even in relatively “volatile” 2008, voter preferences, by state, typically ranged just five to eight percent from their highest to lowest point over course of the campaign. Second, there are indications this year that an unusually large proportion of voters have already made up their minds. Voters are also saying they don’t have much more to learn about the candidates. Factor in (third) that current preferences are already very close to what the historical model is predicting, and we could be very nearly set.

Campaign news is entertaining, and it often feels like things that happen should matter. Don’t pay a lot of attention to the speculation. Watch the numbers. I’ll keep updating the site every few days. When public opinion starts to move, I’ll catch it. Then we can go back and look at what might have been the cause, and what it might mean going forward.


Theme by Anders NorenUp ↑