ARG had a pretty accurate final Iowa poll 4 years ago. They also polled Iowa a lot in 2007.
In Mid-December 2007, Paul only had 5% support and 5-8% support right before the caucus. He ended up with 10%. So I would say he has a damn good chance to win this, unless Santorum really gains in the final weeks ... 
Seriously, I suspect Paul's support is greater than traditional polling would indicate. After all, a sizable group of Paul voters are probably "non-traditional" but more importantly, he attracts the folks who probably are a bit, how do I say, paranoid?
