Polling speculation

Since everyone seems to have a polling prediction, I'll add mine. I mentioned this a few weeks ago, but in the final stages of the election it is worth expanding. We’re hearing a lot about the swing states right now. What no one seems to be taking about, however, is that this is Romney’s second campaign of the year in several of those states: he also held vital primaries in Ohio, Florida, Iowa, and Michigan (and won all but Iowa).

So how did he do last time? Well, according to polling in each state in the final weeks before the election Romney was either tied or slightly behind--sometimes by as many as 10 points. And then he surged to win the state (except for Iowa, where he lost by 0.1 percentage point).
Take Ohio, for example. Here is the Real Clear Politics average of the polling going into the primary. Romney is shown in purple.

The same pattern appeared in Florida.

And Iowa (even though Romney ultimately didn't win). 

And Michigan.

In each of those states, Romney's opponent (often Santorum) saw a polling lead disappear in the final days before the vote.

And that's not all. A closer examination reveals something even more interesting. In all four of those states, polling showed a Romney surge in the final days. But when the results came in, Romney scored even better than the polls suggested. In Ohio, Romney finished 3.8 points ahead of the RCP average. Florida was 4.6 points above polling. Iowa was 1.7 points above polling. And Michigan was 3.8 points above polling. In other words, in each of those contests Romney did noticeably better than the polls predicted.

Now granted, the demographics are in the general and the primary. And this is a two-way race, rather than a multi-way race. But still, a close race with Romney in Ohio, Florida, Iowa, or Michigan, with one week to go, is not a promising spot.

Just ask Rick Santorum.