Jamelle’s post about Florida is excellent — he’s exactly right about Florida.
But I have a better suggestion: Just ignore the state polls entirely right now.
Regular readers have heard me write about this many times, but really — just ignore them. If either candidate wins by three percentage points or more, and probably if it’s more than a single full percentage point, the electoral college will take care of itself. And if it gets close enough that variation in the states matters, it’s too early to really be able to get reliable clues from the states.
That’s especially true because we’re at the mercy of relatively unreliable polls at the state level. That’s not (usually) because there’s anything wrong with the pollsters. It’s because there just aren’t enough surveys yet.
Consider. Since last Monday, Mark Blumenthal’s Pollster site has registered ten different national head-to-head polls, with Obama leads of 1, 0, 1, 3, 4, 2, -1, 3, 7 and -4. That’s a whopping 11 point range (and the largest Obama and the largest Romney lead were actually in the field at the same time). Over the same stretch, in Florida we have . . . just the one poll, with only one other this month, in early May. There’s just no way of knowing whether today’s poll is at the top end of the range, or the bottom, or square in the middle.
The much more reliable method at this point is to look at the national average (Obama by a slim margin), look at how Obama did in 2008 compared with how he did in Florida, and then apply that margin to Obama’s current lead. Which will get you, as Jamelle said, a (narrow) Obama win while Romney takes Florida. The thing is — not only would that work well now, when the current poll gives Romney a solid Florida lead, but it would have worked when we had that early May Florida poll that put Romney in the lead.
Or, the best plan: Just ignore the state polls for now. We have plenty of national polls, which give you all the polling information you need, given that it’s still a while before these polls are really good predictors of final results anyway. Thinking about the states just adds a layer of complexity without adding much understanding.