US Politics XXIX: The Final Thread Before XXX

[LuckyNumber7]

Cohn > Silver

I’m glad I took the opportunity to shit on Silver the day before the election.

 

[namkcuR]

I’m not trying to compare Florida penis with Ohio penis okay?

Nor was I. Geez.

My main point in that post was not any comparison, just expressing my disappointment/frustration with my state for seemingly turning red.

 

[Hewson]

Nevada, huh? I wonder if they have any tourist locations I could visit to repay them for pushing us over the top.

there's probably an old-timey silver mine somewhere near carson city you could throw a 20 to for a tour.

Take the dam tour!

 

[DaveC]

i have to say, the move by paypal/venmo to shut down a bunch of militia-related accounts tonight is genius.

the proud boogaloos are all currently flipping out about that instead of paying attention to the election.

 

[mikal]

I’m glad I took the opportunity to shit on Silver the day before the election.

Yeah he looked brilliant in 2012 and now he’s been way off the last 2 elections.

 

[LuckyNumber7]

GA leading demographics for Biden voters:

Men-Women = 42-54
Black-White = 87-29
Age Breaking group was 30-44 at 52% for Biden

So basically anecdotally probably 8 in 10 white men voted Trump and arbitrarily most old white men voted for Trump.

 

[Hewson]

I’m glad I took the opportunity to shit on Silver the day before the election.

Hopefully you ate corn and peanuts first, that's what he deserves.

 

[gump]

Yeah he looked brilliant in 2012 and now he’s been way off the last 2 elections.

I think he came out the best across the modelers in 2016. Their model really captured Trump's significant chance of winning, even if it still gave Clinton a 2:1 chance.

But there's some reckoning that needs to happen with polling-based modeling and polling itself. I like Nate Cohn a lot (though I find his unedited version not as good as the edited one, as he's not the best communicator at times), but the Upshot polls and analysis was as far off as 538's this cycle.

On Silver again, despite the numbers, I feel they took great pains to explain the uncertainty in the race and that a 10% chance happens quite often.

 

[mikal]

Kind of sucks that the election won’t be called for Biden yet tonight. But is Arizona shapes up of then I think we can all sleep comfortably.

Not sure how much longer the other states will count tonight.

 

[gump]

1) Say a combination of AZ/NV/PA gets called for Biden. Does Trump formally concede?

2) If Biden wins, does Trump invite him for the traditional White House visit?

3) If Biden wins, does Trump write the usual letter that is left on the Oval Office desk? What does he write/draw?

 

[iron yuppie]

On Silver again, despite the numbers, I feel they took great pains to explain the uncertainty in the race and that a 10% chance happens quite often.

Am I wrong in thinking that the point of those models is not to predict a specific outcome but rather to present a range of probable outcomes? In that sense it seems that 538 has done just fine.

It seems to me that people (not anyone here necessarily) will see a 90% victory probability and assume that means a blow-out win when in fact it means the alternative still has a decent likelihood. Just like when basketball fans get surprised when Steph Curry misses a free throw even though there's a decent chance he will miss any given free throw he takes.

 

[LuckyNumber7]

Yeah he looked brilliant in 2012 and now he’s been way off the last 2 elections.

You hear me railing on this concept of “polling averages” a heck of a lot to a point where I sound like a broken record... I lost a lot of respect for Silver when he started chirping those out.

Also 538s “predictive” models were all obnoxiously based on some if-then probabilistic logic which I highly disagree with. It seems to suggest that because something has happened before therefore historical information is useful to do predictive analytics for the future... this hasn’t been true since like 2004. We are far more mobile and diffusive as a people now.

Oh, also, this notion that an election will ever feature an 8% gap is a red flag and any statistician conducting a social study should never allow such a large gap to be unaddressed.

 

[womanfish]

Cohn > Silver

Agreed

 

[mikal]

1) Say a combination of AZ/NV/PA gets called for Biden. Does Trump formally concede?

2) If Biden wins, does Trump invite him for the traditional White House visit?

3) If Biden wins, does Trump write the usual letter that is left on the Oval Office desk? What does he write/draw?

1. No
2. No
3. He’ll draw boobies

 

[Headache in a Suitcase]

1) Say a combination of AZ/NV/PA gets called for Biden. Does Trump formally concede?

2) If Biden wins, does Trump invite him for the traditional White House visit?

3) If Biden wins, does Trump write the usual letter that is left on the Oval Office desk? What does he write/draw?

 

[Hewson]

1) Say a combination of AZ/NV/PA gets called for Biden. Does Trump formally concede?

2) If Biden wins, does Trump invite him for the traditional White House visit?

3) If Biden wins, does Trump write the usual letter that is left on the Oval Office desk? What does he write/draw?

1. No
2. No
3. Person, woman, man, camera, TV (in orange crayon)

 

[Headache in a Suitcase]

Hey Kornacki woke up

 

[gump]

1. No

2. No

3. Person, woman, man, camera, TV (in orange crayon)

Winner!

 

[gump]

Am I wrong in thinking that the point of those models is not to predict a specific outcome but rather to present a range of probable outcomes? In that sense it seems that 538 has done just fine.

It seems to me that people (not anyone here necessarily) will see a 90% victory probability and assume that means a blow-out win when in fact it means the alternative still has a decent likelihood. Just like when basketball fans get surprised when Steph Curry misses a free throw even though there's a decent chance he will miss any given free throw he takes.

Yes, this exactly.

My concern with Silver is that sometimes he becomes a pundit rather than a data analyst. And you can criticize some modeling choices. But I think his work has been better at depicting uncertainty than the overconfident models that were competing with him (the Economist this cycle, the Princeton model last cycle, etc).

 

[womanfish]

I think when it shakes out, the biggest things that pollsters will have to address is why were many swing state polls spot on (AZ, MN, NC, GA) within the margins (FL, maybe PA) but WI, MI, IA, OH nearly outside the margin?
What other types of questions could pollsters ask that might tell them how a person would vote beyond what they tell you.

And bottom line is, when you do have a margin of error where one candidate can be 4 points lower and the other could be 4 points higher, people will rage about them, but there answer can always be - we were within the MOE.

I do wonder if Biden's internals were way off too or not. If so, it may really change the game of how campaigns work going forward.

 

[womanfish]

crazy that PA may come in with better numbers than WI or MI.

 

[LuckyNumber7]

I think when it shakes out, the biggest things that pollsters will have to address is why were many swing state polls spot on (AZ, MN, NC, GA) within the margins (FL, maybe PA) but WI, MI, IA, OH nearly outside the margin?
What other types of questions could pollsters ask that might tell them how a person would vote beyond what they tell you.

And bottom line is, when you do have a margin of error where one candidate can be 4 points lower and the other could be 4 points higher, people will rage about them, but there answer can always be - we were within the MOE.

I do wonder if Biden's internals were way off too or not. If so, it may really change the game of how campaigns work going forward.

It’s trump related. These people won’t answer questions honestly because they don’t trust pollsters because they’re indoctrinated or ashamed.

 

[mikal]

https://twitter.com/jdiamond1/status/1324142698081755142?s=21

 

[LemonMelon]

Come on GA I want to sleep easy tonight

 

[Headache in a Suitcase]

Joe's going to win Pennsylvania and Trump knows it. If they didn't know it, they'd be taking the fight to Arizona and Nevada. They know Pennsylvania is the whole ballgame regardless of what happens in AZ, NV or GA.

 

[dazzledbylight]

Just anyone tonight idc who

I second that!

 

[Headache in a Suitcase]

Oh hey they're trying to stop the count in Georgia now. That's a very good sign for where Georgia is going to end up - buy I do worry about Kemp there.

 

[BEAL]

I’m starting to wonder if the Bumble dating app is a better at polling than what we currently use....

 

[cobl04]

It's not a problem just in the US, in Australia the polls were wildly wrong in our federal election last year, and I think the Brexit polls were wrong too.

 

[LuckyNumber7]

It's not a problem just in the US, in Australia the polls were wildly wrong in our federal election last year, and I think the Brexit polls were wrong too.

For sure. And I think you can still cite this right wing populist uprising that outwardly rejects science as a left wing hoax and views the media as against them.