> 2. Polls attempt to account for the simple error that people are more likely to say that they'll vote than will actually vote. (see here: http://www.gallup.co

> 2. Polls attempt to account for the simple error that people are more likely to say that they'll vote than will actually vote. (see here: http://www.gallup.com/poll/111268/how-gallups-likely-voter-m... http://www.gallup.com/poll/111268/how-gallups-likely-voter-m...) The Romney campaign was predicated on the idea that GOP voters were likely to get out and vote, while Obama voters would say they were going to vote but then drop out on Election Day. That is a polling blind side that Nate Silver couldn't account for. Actually Silver did account for something similar: That polls could be systematically biased against Romney. Not for political reasons, but incorrect LV models. The accurate pollsters of this cycle used an LV model that was shaped by 2008 exit polls. Others used those based on 2004 or 2010 (Rasmussen). Those that banked in this being an '04 or '10 electorate were way off. He eludes to this here: http://fivethirtyeight.blogs.nytimes.com/2012/11/02/nov-1-the-simple-case-for-saying-obama-is-the-favorite/ http://fivethirtyeight.blogs.nytimes.com/2012/11/02/nov-1-th... and: http://fivethirtyeight.blogs.nytimes.com/2012/11/01/oct-31-obamas-electoral-college-firewall-holding-in-polls/ http://fivethirtyeight.blogs.nytimes.com/2012/11/01/oct-31-o...