10 Things You Should Know About the Post-Debate Polls
You may have heard that there was a debate last week between Mitt Romney and Barack Obama. Romney won, and then the political press turned his victory into a blow-out of epic proportions (theCBS' snap poll immediately following the debate showed that a majority of uncommitted voters thought either that Obama won (22 percent) or that it was a tie (32 percent)).
Since then, several polls have been released showing Romney making significant gains, most notably a Pew poll released on Monday that had the challenger up by 4 points nationally. This state of affairs has Republicans measuring the drapes of the Oval Office and some Obama supporters on suicide watch.
But it's not wise to read too much into the polls immediately after a significant development like Romney's winning performance in the first debate. Here are a few things to keep in mind when reading the polls this week.
1. I'm An Obama Supporter – Should I Feel Panicky/ Start Looking at Canadian Real Estate?
No, you should remain calm and wait for additional survey data. The thing that stands out about the polling right now is that it's very noisy – different pollsters are getting very different results. The best bet right now is to take individuals polls with a grain of salt and look instead at polling averages like those maintained by Real Clear Politics or Talking-Points Memo, or broader analyses like Nate Silver's 538 model, which incorporates polling with economic data to survey the state of the race. These are less likely to over-react to short-term, news-driven shifts in the race.
2. What's the State of the Race Right Now?
Again, it's pretty noisy. But it appears that Romney got a significant bump in the race – maybe as much as 5 or 6 points in surveys conducted on Thursday, Friday and Saturday -- which, according to Gallup, Ipsos/Reuters and Public Policy Polling, then began to recede on Sunday.
3. Does That Mean Voters Are Shifting Their Support From Obama to Romney
No. Romney may have pulled a few weak supporters from Obama, but the shift we're seeing is almost entirely about Republicans who didn't care much for Romney coming back into the GOP fold.
4. So, Did Romney Change the Fundamentals of the Race or Not?
It's really too soon to tell. Most of the polling we're seeing right now was conducted in the immediate aftermath of the debate. Some of it was conducted after the debate but before Friday's decent jobs numbers were released.
The Pew poll, for example, was conducted between Thursday and Sunday, but only 13 percent of the interviews were done on Sunday, so it's a bit front-loaded. Similarly, a Zogby poll released today finds the two candidates in a tie, but it was conducted Friday through Sunday.
The question is whether Romney got a bounce or a bump. We'll know in a few days. The rest is punditry.
5. Isn't This Tighter Race What Everyone Expected All Along?
Yes. After the conventions, Obama opened up a very healthy lead, but the fundamentals have always suggested that the 2012 election would be a tight race, with Obama the favorite but not hugely so. The reason Obama looked like he was headed toward a landslide win was that, as I mentioned above, a lot of Republicans and independents who vote for Republicans thought Romney was lame and were unenthusiastic about supporting him. With a strong debate performance, they are now finding him acceptable and we appear to be seeing the race we were supposed to see – tight, with Obama a slight favorite.
6. Should I Dismiss That Pew Poll Because It “Oversamples” Republicans?
No. That's making the same fundamental error as the “poll truthers” on the right. Partisan ID, unlike demographics such as age, gender or ethnicity, is highly fluid. And that's not because people switch from Democrat to Republican at the drop of a hat, or vice-versa. It's kind of a proxy for voter enthusiasm – when people feel good about a party, they're less likely to tell pollsters that they're independents. Similarly, only about 10 percent of people reached by pollsters participate in their surveys, and when people feel good about their candidate they may be less likely to hang up when that call from Gallup comes in. Good polls using industry standard methodology can and do find wide variation in party ID – leave the trutherism to the nutjobs at Fox News.
7. Can Obama Rebound?
Obviously, and the idea that he can't – that Romney landed a “knock-out punch” and the election may be over – is a bit odd. It should go without saying that if one is going to lose a debate, the first one is a lot better than the last. Obama and Romney have two more debates and Joe Biden and Paul Ryan face off in a vice-presidential debate this week. Expect Obama to be a lot more aggressive.
But here's the thing: it's unlikely that the race will rebound to the wide lead Obama was enjoying after the conventions because after Romney's performance in the first debate, we're not going to see large numbers of Republicans and Republican leaners regarding him as a disastrous, hopelessly bumbling loser. That's pretty much baked into the cake at this point.
8. What About the State Polling?
In the swing states, Romney has gotten a bump just as he has in the national polls, but it's even less clear whether it's a sustainable shift or a short-term bump. There are fewer state-level polls and more of them are conducted by pollsters with shaky track records. We need more data.
But here's an important point: it remains the case that the Electoral College map favors Obama. As the Washington Post's Chris Cillizza and Aaron Blake put it:
For the sake of argument, say that Florida, Ohio, North Carolina and Virginia all move toward Romney — and he winds up winning them as well as all of the more reliably Republican states leaning or solidly in his camp today. He still loses the electoral vote to Obama. We’ve written extensively about this often-overlooked reality in recent months but it’s worth reiterating again: Even if Romney surges in a handful of swing states, his path to 270 electoral votes remains tough. For Romney to win at this point — assuming he can’t put Michigan, Pennsylvania or Wisconsin play — he needs to all but sweep the remaining toss-up states.
9. Did Romney Just Get a 7-Point Bump in Gallup's Tracking Poll This Morning?
No, Gallup switched to its likely voter model today, which resulted in a shift from Obama up by 5 to Romney up by 2. But there's an important caveat, as Steve Singiser notes, “Gallup's likely voter screen has a pretty chequered past.”
It crapped the bed a little bit in 2008, when it overestimated Barack Obama's margin of victory by four points. But then the likely voter screen for Gallup really took a dump in 2010, when their RV/LV gap on the generic ballot was a cartoonish eleven points. Among registered voters, Gallup saw an R+4 electorate. Among "likely voters", Gallup saw an R+15 electorate. The final outcome? R+6. So...yeah, let's say that the assumptions made in likely voter screens can be, shall we say, a little off.
It's also worth noting that 53 percent of respondents had a favorable view of the incumbent – one point less than his high-water mark for the year.
10. OK, So the Takeaway Is?
Take a deep breath, and don't panic. A black guy dealing with 8 percent unemployment was never going to run away with it. It seems like we have the race we were supposed to have, but we need to wait for more polling to be certain.
In the meantime, Nate Silver -- whose model now gives Obama a 75 percent chance of victory in November – offered this advice:
If you can trust yourself to take the polls in stride, then I would encourage you to do so. If your impression of the race is changing radically every few minutes, however, then you’re best off looking at the forecasts and projections that we and our competitors publish, along with Vegas betting lines and prediction markets.
All of these methods have slightly different ways of accounting for new information, but they do involve people who are risking either money or reputation to get it right, and who have systematic ways to weigh the evidence rather than doing so on an ad hoc basis.