Over the last week, fights have broken out about the polls. Here’s all you need to know:
1. Barack Obama has a small lead in the national vote, and a somewhat larger lead in the electoral college. While small, it’s a real lead, and large enough that we can say that if the polls are essentially correct, Obama will win.
2. It’s the job of any candidate losing in the polls to pretend that there’s a good chance the polls are wrong, as the Romney campaign is doing. All losing campaigns do this. They would be irresponsible not to. That’s especially true for presidential campaigns; if they admit that they’re losing, it’s possible that they could hurt the rest of the ticket. But all losing campaigns will, and should, talk up their chances until the bitter end.
3. Some of this spin has been directed at poll-based predictors and polling aggregators, such as Nate Silver. It’s silly. The problem for Mitt Romney is the polls, not the way people are manipulating them.
4. The press is normally a bit of an ally for losing campaigns in their attempts to spin away what the polls are clearly showing. That’s because the press — both individual reporters on the campaign beat, and publishers and broadcasters who want readers and viewers — have incentives to portray the outcome as unknown. Cliffhangers sell! What’s more, conventional standards of neutrality lead the press to treat claims about the polls from both sides equally.
5. And yet: Obama’s actual lead in the polls, while quite real, is small enough that it’s not impossible at all to envision a Mitt Romney victory. Obama’s lead is not like Bill Clinton’s over Bob Dole in 1996 here, much less like Ronald Reagan’s over Walter Mondale 1984. It is quite unlikely that the polls will be wrong — and if so, wrong by enough to enable Romney to win. But it’s not entirely implausible.
Some of the critics of the polls, or of the polling-based predictors, have identified reasonable arguments for why the polls might be off; some critics haven’t. There are plausible reasons why polling could be wrong by two or three percentage points. But the polls could be wrong in Obama’s favor, too. To yield a Romney victory, it pretty much has to be the case that all the plausible-but-unlikely reasons that the polls are inflating Obama’s chances must turn out to be correct, while none of the reasons that the polls could be biased in Romney’s favor (making Obama’s real lead larger than what the polls say) is correct. That’s not impossible. But it’s highly, highly unlikely.