Gun to my head, Hawley.
I keep going back to 2016. Trump underpolled by 6 points verus the RCP average. If it is a third of that this cycle, Hawley wins.
I keep wondering if the polling models in states that were outside the margin of error in 2016 (MO, IN, etc)... will experience an overcorrection in the polling models for 2018 as the pendulum of Trump support swing back a bit.
That is why I said, "even a third of that."
Excluding IL, something was way off in the mid-west in 2016 regarding the polling. Trump was leading in MO by 11 points; he took it by 18.5 points. Similar things happened in IA, WI and MN; it was 6-7.5 point in those states. The undecided breaking would not it. It had to be either a series of bad models or, in that region, it is a "shy Trump supporters" thing. It doesn't have to be a lot of either to give the state to Hawley.
Of course the opposite is true in NV and AZ.