Conversation

This is pretty interesting

Quote Tweet

Oct 7, 2018

Maybe the easiest way to summarize this fact: the FiveThirtyEight forecast, last I looked, had Democrats favored to win +33 seats, well over the 23 they need. But FiveThirtyEight only had them favored in 218 individual contests, precisely the number needed for a majority

Show this thread

129

458

Nate Silver

@NateSilver538

Oct 7, 2018

With a generic ballot lead of D+7 or so, it's very possible for polls to about right overall, but for Dems to lose the House because the GOP ekes out wins in lots of the tossup districts. Once you get up to D+9 or so, the dam breaks and GOP would need a systematic polling error.

108

320

793

Nate Silver

@NateSilver538

We have the D's generic ballot/popular vote lead trending toward +7.5 or so. So while the topline numbers in our House forecast haven't changed much, we've crossed an important conceptual threshold where the eke-it-one-district-at-a-time scenario is back on the table for the GOP.

6:38 PM · Oct 7, 2018

Replying to

You're all (polsters) full of bull. Since the 2016 elections I don't believe one word from any of you.

Replying to

and

You might want to take a quick read of Wikipedia about how probabilities work. 538's prediction that Trump had a 30% chance of winning in 2016 is actually pretty damn good. That's about the same odds of flipping a coin and getting 3 heads in a row. Not likely, but possible.

Show replies

Replying to

UK's Labour is in a similar scenario. In the 2017 GE, a lot of their majorities (i.e. raw vote margins) in the constituencies they picked up from the Conservatives are < 1K.

Replying to

this underscores your point that the "Lean R" districts are where to really focus

Replying to

This is scary interesting and I’m hopeful democrats can push the needle in individual districts by being good candidates. Knock doors, kiss babies, be authentic with your empathy and understanding and Goethe fuck out there and GOTV!