There's a link to their final polls now.
"Fresh samples were drawn" they say but "As with all YouGov polls, respondents were selected from online panels."
Decimals and a ton of undecideds a few days before the election!
Brownback +0.8 is the closest they've ever shown that race.
This is the most blatant case I've seen this entire cycle of a pollster "massaging" the data of a bad sample to make it appear something vaguely appearing to be based on a random sample. How in the world does a pollster show quadruple the undecided voters compared to other people's polls, but still manage to show the margins between candidates looking like everyone elses', in eleven different tossup races?