Normally when I'm skeptical of a poll it's because it's from a firm I don't trust or because I don't think its methodology makes it predictive of the actual election.
But here we have Rasmussen Reports polling likely voters, and it's by far the best poll I've seen for Republican Richard Burr in the North Carolina Senate race against Democrat Elaine Marshall.
Burr hadn't hit double figures since shortly after Independence Day. But now he's broken way past that with a 54-38 advantage (MoE 4.5). Almost all of the shift is on Burr's side, as well. Marshall loses two points from the previous poll, but Burr gains 5.
The difference, of course, is that this is Rasmussen's first poll to include leaners. This suggests that Burr has the potential support he needs to hold off Marshall's challenge, but that support is not very enthusiastic. That's strange for this year, but Burr's inability to shake off Marshall has seemed strange to me all year long.
I will be interested to see if Public Policy Polling also shows Burr doing better after a shift to Likely Voters, or if this is a one-off thing from Rasmussen.