US Poll-itics

[Kalbear]

3 minutes ago, Maithanet said:

Ok, this is more reasonable, but the problem is that you always tie this "the polls could be wrong!" to an argument that Trump is likely to prevail in a crazy election.  In a normal year, someone polling +9 nationally on election night would have like a 98% chance of winning the presidency.  If your point is that this particular election is extremely high variance, then perhaps that number is only 90% or 80%, but it's definitely not 50%, that's just not what variance is.  The fact that national polls continue to find significantly more people want Biden to be president is in fact a big advantage in an election. 

I think that because Trump has the federal office, the DOJ is his lapdog and the courts are in the bag for him that crazy favors him. Wisconsin is a good example of this - polling is +17 ! for Biden, but they also have the most stupid system for counting and the most pro republican court in the battleground states. That would be a great example of a state that will show MASSIVE shenanigans, I think. Pennsylvania is not out of the woods either, as they want scotus to throw out the +3 days rule for their ballots. 

That is the sort of thing that Trump has which Biden doesn't and why I think it favors him more. 

My current thought is that Biden will absurdly win the popular vote - possibly by as much as 10 million. Itll be a high turnout election in most states. But we won't have an official winner for months, because literally millions of votes are going to be attempted to be ruled ineligible. The scale won't be a few hundred thousands - itll be massive. 

[Kalbear]

Also, @DMC, while you did make those predictions before I had assumed that you might change them with new info. If you haven't that's fine! You're just uncurious, but you're not a coward. 

[Fragile Bird]

The thought that crossed my mind is that if Biden wins by that much, any attempt to overturn the election will be met with wide spread anger and Republicans who did get elected will disavow Trump. 
 

* crazy, optimistic Canadian POV*
 

The Science office in the WH issued a report listing “ending the Covid-19 epidemic” as one of Trump’s accomplishments. A spokesperson has apparently tried to explain that away by saying it was supposed to be listed as one of the WH’s planned future accomplishments.

[williamjm]

3 hours ago, DireWolfSpirit said:

That's what I had thought of as soon as I saw your first post. 

Maybe it's a bit like how we hear the British do not have to use postage stamps because they invented the postage stamps or something?

Britain is the only country that doesn't include its name on its stamps (we have a picture of the Queen's head instead), which is probably what you're thinking of.

[DMC]

1 minute ago, Killjoybear said:

You're just uncurious

I'd argue wholly dismissing the metrics and analysis of people that devote their lives trying to figure out what's going on and instead relying on your gut feelings is "uncurious," by YMMV.

[Ser Scot A Ellison]

Can we have a seperate thread to discuss the intricacies of statistical modeling in polling?

Please?

[Kalbear]

2 minutes ago, DMC said:

I'd argue wholly dismissing the metrics and analysis of people that devote their lives trying to figure out what's going on and instead relying on your gut feelings is "uncurious," by YMMV.

I give them as much credit as I gave Sam Wang last cycle. When they suck less they get more credit. 

[maarsen]

18 minutes ago, Killjoybear said:

I mean, in a sane world...

 

Does anyone else think the people around Trump are feeding him crazy ideas and trying to top each other by getting Trump to say it out loud?

[DMC]

2 minutes ago, Killjoybear said:

I give them as much credit as I gave Sam Wang last cycle. When they suck less they get more credit. 

Which demonstrates either your ignorance or intentional bias - there were plenty of (virtually all) other modelers and pollsters showing a much greater chance for Trump in 2016 than the Princeton model.

[Mr. Chatywin et al.]

59 minutes ago, The Anti-Targ said:

It's only really an evolution on one of the aspects of ritual cannibalism, that by eating a person you take on their power. 'Twas a thing believed by a lot of pre-industrial societies, and undertaken, I believe, especially by the victor after a battle. It's not really surprising, when you think about it, when this conspiracy is mostly believed by people who believe demonic possession is an actual thing. It is also metaphorically represented in 'Highlander' with the whole chop the head off and absorb the power thing, to become "The One", once you've eliminated all your competition.

Darn you! As I began reading I was already thinking how I could make a Highlander joke. You stole my thunder.

[Kalbear]

2 minutes ago, DMC said:

Which demonstrates either your ignorance or intentional bias - there were plenty of (virtually all) other modelers and pollsters showing a much greater chance for Trump in 2016 than the Princeton model.

Sure, but none of them got it right except the guy who didn't look at polls at all. If all the models all say the same thing and all of them get it wrong, that's not a great sign. 

I'm honestly more in line with Silver this election who says that in a normal election Biden cruises- but this isn't a normal election and that those things (fraud, theft, corruption, court cases) are not very predictable. Still, no one - not silver, not the economist, no one - is modeling with polling being potentially very off. 

[Ran]

SCOTUS rejects PA GOP's second effort to get them to intervene to prevent mailed ballots from being counted if they're received up to 3 days after the election. Interesting, really, for two reasons: Alito, Thomas, and Gorsuch dissented (though Alito argues that the GOP can bring the case again after the election if those late ballots have a meaningful impact on the result), Kavanaugh -- whose concurrence opinion in the WI case just yesterday has been lambasted for being incredibly hackish and sloppy -- with the majority... and Barrett, who the PA Dems filed a motion suggesting she needed to recuse herself, doesn't take part because she didn't have time to review the material, side-stepping having to engage with the Dem argument.

 

[DMC]

5 minutes ago, Killjoybear said:

Sure, but none of them got it right except the guy who didn't look at polls at all.

Probabilistic models are not predictions.  Silver has been trying to emphasize this for the past four years.  As for you being "in line" with Silver, that's especially rich even for you.

[Ser Scot A Ellison]

5 minutes ago, Ran said:

SCOTUS rejects PA GOP's second effort to get them to intervene to prevent mailed ballots from being counted if they're received up to 3 days after the election. Interesting, really, for two reasons: Alito, Thomas, and Gorsuch dissented (though Alito argues that the GOP can bring the case again after the election if those late ballots have a meaningful impact on the result), Kavanaugh -- whose concurrence opinion in the WI case just yesterday has been lambasted for being incredibly hackish and sloppy -- with the majority... and Barrett, who the PA Dems filed a motion suggesting she needed to recuse herself, doesn't take part because she didn't have time to review the material, side-stepping having to engage with the Dem argument.

 

If Barrett actually recuses herself from all Election related cases would that impact anyone’s view of Barrett?

[Ran]

Just now, Ser Scot A Ellison said:

If Barrett actually recuses herself from all Election related cases would that impact anyone’s view of Barrett?

The way she put it makes people think she's not planning to recuse herself from all election cases, but to my eye it feels like she's basically walking the tightrope and we'll have to see what happens. I certainly think she'll need to address it if a party to a case argues that she must recuse, one way or the other.

[Gaston de Foix]

1 minute ago, Ser Scot A Ellison said:

If Barrett actually recuses herself from all Election related cases would that impact anyone’s view of Barrett?

Not me at least.  My concern with Barrett is not she is an unethical person or a Republican hack.  She may be both for all I know or not.  It is, as she herself has said, "her judicial philosophy is Scalia's".  Scalia believed (even if he never quite said it) that Brown v Board of Education was wrongly decided.   

[Kalbear]

Just now, DMC said:

Probabilistic models are not predictions.  Silver has been trying to emphasize this for the past four years.  As for you being "in line" with Silver, that's especially rich even for you.

Even if probabilistic models are not predictions they still did not give much of a chance for Trump to win, despite a lot of other indicators. Silver was the best and I personally did bring up that polls are only as accurate as the assumptions they're based on,, and in response @lokisnow laughed at me. More importantly ALL the models were off in exactly the same way, which tends to indicate that the basis being used is flawed and misunderstood. 

And yes. I think silvers caveat is more right. I don't know why thats weird - I've been saying that for as long as you've said that Biden would comfortably win. I think he is aware of that being an issue and is moving on knowing it could be totally wrong. What other choice does he have? He's built his entire website on this. But that model right away assumes certain priors and if those are wrong, the whole thing fails. Now again I ask - what do you think is more right - that polling will have roughly the same error as the last 40 years with all the shit that is going on, or polling will be less accurate and / or may not matter because so many votes are going to be challenged, uncounted or won't arrive?

[Mr. Chatywin et al.]

26 minutes ago, maarsen said:

Does anyone else think the people around Trump are feeding him crazy ideas and trying to top each other by getting Trump to say it out loud?

He's Ron Burgundy. 

[Larry of the Lawn]

3 minutes ago, Killjoybear said:

Even if probabilistic models are not predictions they still did not give much of a chance for Trump to win, despite a lot of other indicators. Silver was the best and I personally did bring up that polls are only as accurate as the assumptions they're based on,, and in response @lokisnow laughed at me. More importantly ALL the models were off in exactly the same way, which tends to indicate that the basis being used is flawed and misunderstood. 

And yes. I think silvers caveat is more right. I don't know why thats weird - I've been saying that for as long as you've said that Biden would comfortably win. I think he is aware of that being an issue and is moving on knowing it could be totally wrong. What other choice does he have? He's built his entire website on this. But that model right away assumes certain priors and if those are wrong, the whole thing fails. Now again I ask - what do you think is more right - that polling will have roughly the same error as the last 40 years with all the shit that is going on, or polling will be less accurate and / or may not matter because so many votes are going to be challenged, uncounted or won't arrive?

As someone completely uninformed about the vagaries of polling, totally agree with this.  Even if you correct for the shit that was off in 2016, how do you factor in all the chaos of this year between the pandemic and it's effect on turnout, SCOTUS, and the possibility of bunch of election shennigans in places like WI and PA?

Random anecdote- I fell down a board-hole about a month ago and was reading the Sept-Oct  US Politics threads from 2016 and lokisnow speculated that WI and maybe PA would just barely go for Trump due to bad polling and give him a narrow win.  

[DMC]

Just now, Killjoybear said:

Even if probabilistic models are not predictions they still did not give much of a chance for Trump to win, despite a lot of other indicators.

A 30 percent chance - which is what the final 538 forecast gave Trump - is a pretty damn good chance.  As for "all the models being off in the same way," well yes, all the polling the models were based off of would have to be off in the same way to elicit a significant polling error - otherwise they'll cancel each other out and there won't be an aggregate polling error.  In 2016 it went Trump's way.  In 2012 it went Obama's.

2 minutes ago, Killjoybear said:

And yes. I think silvers caveat is more right. I don't know why thats weird - I've been saying that for as long as you've said that Biden would comfortably win. I think he is aware of that being an issue and is moving on knowing it could be totally wrong.

I haven't said Biden would comfortably win.  It's like you're incapable of understanding nuance.  I've said if Biden maintains the current lead in aggregate polls - both nationally and in swing states - it would require a massive systemic polling error that would be significantly larger than the 2016 error.  I also have said I agree if Trump loses closely, which would inherently only require a more muted polling error, I agree he'll be successful in stealing the election with the help of the SC.  I've also mentioned - albeit I don't think to you directly - that I don't take 538's current probabilities of a Biden victory at face value.  I do not think Biden actually has an 88% chance of prevailing - and I've said before I doubt Silver actually does either - because of all the factors you highlight and I tend to agree with, just not at the magnitude you do.