They BTH weight by party identification (compared to their latest polls).
Someone that is smarter than me explain to me this... They find the voter intensity on Trump's side. Write about how big of a difference it is and then at the end of the poll, admit they are weighing to party ID.
Then there's this:
Democrats-Republicans-independents, in the full sample, and 36-28-31 among likely voters.
What the hell kind of sample did they get in that scenario? I don't even think it was D+8 in 2008.
There hasn't been a D+8 exit poll going back to 1992. There hasn't been a Republican number at 28 in that time period either. There must be a hell of a lot of Trump-supporting independents then or something...
No... wait.. Clinton wins independents...