538's house prediction has us up to 86% likely to win control, average gain 40 seats

[unblock]

gotv!



https://projects.fivethirtyeight.com/2018-midterm-election-forecast/house/?ex_cid=midterms-header

[Scurrilous]

1. K&R

[C_U_L8R]

2. So much winning !

[Wounded Bear]

5. Not winning yet...

15 days to go until we get to celebrate.

GOTV!!!!

[Eyeball_Kid]

27. I'm still pessimistic. I fear that a rigged "miracle" will allow the GOP to keep

Congress. And it will be rigged for the GOP to keep control. There's no other way.

[sarisataka]

7. We haven't won

A damn thing yet

[C_U_L8R]

9. Of course of course

It’s just a Trump jab. GOTVLAMF!

[Quemado]

3. GET OUT AND VOTE!!!

Don't listen to anyone! Get out and vote!

[Baitball Blogger]

17. +1

[Wounded Bear]

4. WA08 is a tossup...

I can't vote there, it's next door to me. We're pretty solid in the 9th, there are 2 Dems running.

Dino Rossi has never won a state wide race, most polls had him leading so far. I think Kim Schrier is killing him on healthcare. She's a pediatrician, so she knows her ADA and CHIP facts.

[CousinIT]

6. Meh. GOTV. n/t

[MineralMan]

8. We'll see on Wednesday morning, the 7th.

I hope that prediction is correct.

[a kennedy]

10. Exactly, waiting for the final totals........

[Hassin Bin Sober]

13. It may take weeks to find out some of these elections.

People need to be reminded that many of these pick-ups will be in toss up elections that may come down to the hundreds, if not dozens, vote count margins.

There are bound to be recounts.

[Autumn]

14. Bernie says it may be tight.

GOTV

[Hassin Bin Sober]

15. Aggggg ummmm. Aaaaaaack you said THE word!

[Autumn]

16. OMG I did!!

[NeoGreen]

20. What?...

..."It"?

[Hassin Bin Sober]

32. Ber....NIe

[NeoGreen]

33. Oh that's good...

...because you can't get very far in life without saying "it".

[Tiggeroshii]

11. Vote vote vote! Manage expectations accordingly and VOTE!

[nycbos]

12. Go knock on doors. Make phone calls.

Take nothing for granted.

[Joe941]

18. This is going to be historic! I can't wait!

[melm00se]

19. Seat losses

The median number of House seats lost by the president's party at the midterm is 26.


The highest number of House seats lost was 71 in 1938.
The highest number of House seats gained was 9 in 1934.

The median number of Senate seats lost by the president's party at the midterm is 5.

The highest number of Senate seats lost was 13 in 1958.
The highest number of Senate seats gained was 9 in 1934.

https://www.presidency.ucsb.edu/statistics/data/seats-congress-gainedlost-the-presidents-party-mid-term-elections

[NickB79]

21. I hope to God they're right, but I'm not letting my guard down

Last time I did that, I went to bed expecting President Clinton and woke up to President Trump.

Everyone needs to get friends and family to the polls.

[OilemFirchen]

22. I know a lot of people talk about this blue wave and all that stuff, but I don't believe it.

[BannonsLiver]

23. Oh.

[OilemFirchen]

24. I'm in good company.

The most popular politician in the known universe said exactly the same thing.

[BannonsLiver]

25. Sounds more like wishful thinking to me.

Maybe looking for an I told you so moment more so than a Dem house?

[OilemFirchen]

28. Actually, it sounds stupid to me.

But I felt compelled to say it anyway.



“I know a lot of people talk about this blue wave and all that stuff, but I don’t believe it,”

[Azathoth]

29. Worth pointing out that the 538 "Lite" model has consistently shown an 8%-10% tighter race

Which is important because the "Lite" model only uses polling data, while the other models factor in "fundamentals" and all kinds of other things.

Nate Silver got himself in trouble in 2015-2016 because he started relying on additional "fundamentals" variables to dismiss Trump's chances of being nominated and winning the general.

[Amishman]

30. The lite model is pretty much just polls, which is inaccurate for the House

Many house races are not polled or under polled, which will throw it off

[Azathoth]

31. That's true, but polls are still the most reliable predictor of elections

Once you start modeling other variables, you get into all kinds of thorny issues. Are you cherry-picking variables? Are you properly evaluating and, if necessary, operationalizing those variables? Are you missing underlying "hidden" variables? Is a variable that was "fundamental" in 1962 just as fundamental in 2018, and vice a versa?

I'm not saying that fundamentals models are invalid or that 538's models are wrong, but I AM saying that properly-done statistical sampling is the only way to determine to a level of mathematical certainty what's actually going on in the electorate. When polls fail, it's usually because the sampling wasn't done properly, often because the pollsters were using some kind of fundamentals model that yielded incorrect assumptions (likely voter model, likely composition of the electorate, etc.)

538's 'Lite' model has consistently lagged behind the other models, which tells me that whatever "fundamentals" Nate and his team have plugged into the other models are consistently biasing their House predictions towards Democrats.

Again, I'm not saying 538 is wrong, but personally, as far as I'm concerned, any competitive district for which there is no decent polling should be treated as 50/50.