Welcome to DU! The truly grassroots left-of-center political community where regular people, not algorithms, drive the discussions and set the standards. Join the community: Create a free account Support DU (and get rid of ads!): Become a Star Member All Forums Issue Forums Culture Forums Alliance Forums Region Forums Support Forums Help & Search
34 replies = new reply since forum marked as read
Highlight: NoneDon't highlight anything 5 newestHighlight 5 most recent replies
Trump is down to 13 on Fivethirtyeight (Original Post) Le Roi de Pot Oct 2020 OP
Ugh, there's a tie scenario. unblock Oct 2020 #1
The tie scenario is irrelevant. TwilightZone Oct 2020 #4
And FL is key for the asshole. FL loss means game over for it. Le Roi de Pot Oct 2020 #7
I'm not complacent by any means, but I just don't see a path to victory in Florida for him. Pacifist Patriot Oct 2020 #16
The panhandle is certified stupid NorthOf270 Oct 2020 #24
I just don't see the Panhandle having the population to overcome the peninsula. Tommy_Carcetti Oct 2020 #28
I agree with you. Pacifist Patriot Oct 2020 #31
So you're saying the Democrats are too incompetent to pick Electors who will stick with Biden? brooklynite Oct 2020 #6
Your faith that zero out of 269 electors would crack unblock Oct 2020 #9
Almost every state has laws against that. nt Codeine Oct 2020 #14
And yet there are faithless electors nearly every 4 years unblock Oct 2020 #21
still has 43.6% approval and 53.2 % disapproval mucifer Oct 2020 #2
Loud and boorish ones too Le Roi de Pot Oct 2020 #5
if I never see another post about 538's odds it won;t be soon enough beachbumbob Oct 2020 #3
Agree Inablueway Oct 2020 #8
They can't be "wrong " zipplewrath Oct 2020 #11
Agreed! ProfessorGAC Oct 2020 #30
It's the nature of probability zipplewrath Oct 2020 #32
I've Used Monte Carlo ProfessorGAC Oct 2020 #33
They had many uses zipplewrath Oct 2020 #34
Tough luck, it is good news! Maybe take a break! nt USALiberal Oct 2020 #15
So ignore it then. BannonsLiver Oct 2020 #25
The clock is running out Johnny2X2X Oct 2020 #10
You are correct, the models don't account for early voting. whopis01 Oct 2020 #19
"(T)he same chance as flipping a coin and having it come up heads 3 times in a row" Tommy_Carcetti Oct 2020 #29
At this point even with a 2016 polling error Jamesyu Oct 2020 #12
Yep. And Georgia, Maine 2, and Iowa within 1% and moving toward Biden. yellowcanine Oct 2020 #13
On 10/17/2016 Fivethirtyeight had Trump at 12 and Clinton at 88 whopis01 Oct 2020 #17
anything more that 0% is way too much Le Roi de Pot Oct 2020 #18
Agreed n/t whopis01 Oct 2020 #20
Agree. Any chance at all is terrifying. n/t Chemisse Oct 2020 #22
I think the key is momentum. One side has momentum, and if that keeps up, Trump's chances become Doodley Oct 2020 #23
538 Poll average is now Biden +10.6!!! Johnny2X2X Oct 2020 #26
you mean beautiful? Le Roi de Pot Oct 2020 #27

unblock

(51,974 posts)
1. Ugh, there's a tie scenario.
Mon Oct 12, 2020, 08:30 AM
Oct 2020

We may win enough state delegations in the house to pull off a Biden win in that case.

But that assumes every elector votes as they're supposed to. I can guarantee that Donnie and his crooks would bribe and/or threaten a bunch of Biden electors and odds are one of them would crack.

Alternatively, they find done way to get something before a right-wing packed court and win that way....

Anyway that scenario would be a very ugly mess

TwilightZone

(25,342 posts)
4. The tie scenario is irrelevant.
Mon Oct 12, 2020, 08:34 AM
Oct 2020

If Biden wins by anywhere near the current margins, it'll all be over election night and no one will care what Trump says outside of his drooling supporters.

Pacifist Patriot

(24,647 posts)
16. I'm not complacent by any means, but I just don't see a path to victory in Florida for him.
Mon Oct 12, 2020, 09:15 AM
Oct 2020

I say this as someone sitting in a county that went for Trump by 16 points, and I have brazen Trump supporting neighbors flying huge Trump flags from their houses since yard signs are prohibited. I'm not blind to his support here by any means. However.

He won Florida with a margin of just 112,911 votes with a total of 9,122,861 cast.

Since that time, we've had four years of teenagers become eligible to vote and the "likely voter" numbers in the youngest voter demographic is now giving baby boomer likely voters a run for their money. That age group is significantly anti-Trump.

We've seen reports in deep red areas like The Villages and Vero Beach where democrats have historically kept their heads down to avoid negative attention, that there is significant Biden signage, golf cart parades, and airplanes flying over their communities with anti-Trump/Pence messaging.

The Florida legislature has been doing everything they can to muck up the felon voting rights restoration, but they haven't been able to prevent it entirely.

Hurricane Maria brought a decent number of Puerto Rican voters into our state who still reside here and are still furious with Trump.

I wouldn't read too much into the voter registration numbers (Democrats are still holding on to an edge, and we have over 700K new NPAs since the 2016 election) because it is all about turnout.

Anecdotally, I have encountered countless 2016 Trump and third-party voters who are all in for Biden this time. I have spoken to a number of Trump supporters (feeling like I need a shower afterwards) and every last one of them voted for him in 2016. I have yet to find a Trump voter who switched to him at some point in the last four years. Granted, maybe no one is willing to admit that publicly.

If Democrats keep up the pressure in Florida and turnout the vote like I think they are going to, I honestly don't see how he could legitimately win here.

Tommy_Carcetti

(43,080 posts)
28. I just don't see the Panhandle having the population to overcome the peninsula.
Mon Oct 12, 2020, 10:44 AM
Oct 2020

I think we'll see all the major metro cities in Florida go for Biden, even Jacksonville, the most conservative of them.

The Panhandle only has a few moderately large cities--Pensacola, Panama City, and Tallahassee. Tallahassee, a college town, already goes blue.

Pacifist Patriot

(24,647 posts)
31. I agree with you.
Mon Oct 12, 2020, 12:36 PM
Oct 2020

Yes, the peninsula is essentially Lower Alabama, but they aren't the tail that wags the dog.

brooklynite

(93,847 posts)
6. So you're saying the Democrats are too incompetent to pick Electors who will stick with Biden?
Mon Oct 12, 2020, 08:37 AM
Oct 2020

I hate lazy conspiracy theories.

unblock

(51,974 posts)
9. Your faith that zero out of 269 electors would crack
Mon Oct 12, 2020, 08:43 AM
Oct 2020

in the face of criminal pressure is touchingly naive.

At this point, it's not a "conspiracy theory" that Donnie and his gang are committing crimes to try to win the election any way they can. He's pretty much announced that.

Anyway, trying to win in the courts is more likely, and would come first anyway. Or getting states to refuse to certify elections.

Point is, they'll try all the dirty tricks post-election they can think of. Pressuring electors is merely one of them.

unblock

(51,974 posts)
21. And yet there are faithless electors nearly every 4 years
Mon Oct 12, 2020, 09:36 AM
Oct 2020

Granted it's never made a difference, and I don't know if any elector actually flipped to the opposition rather than just voting for their own aunt or something.

Still, it happens....

 

beachbumbob

(9,263 posts)
3. if I never see another post about 538's odds it won;t be soon enough
Mon Oct 12, 2020, 08:34 AM
Oct 2020

this stuff should be ignored as it means nothing until all the votes are cast and counted

 

Inablueway

(56 posts)
8. Agree
Mon Oct 12, 2020, 08:41 AM
Oct 2020

As far as I’m concerned, pollsters have a long way to go to recover any credibility after they crashed and burned in 2016. To be fair, 538 was less wrong than others but still missed by a wide margin.

zipplewrath

(16,646 posts)
11. They can't be "wrong "
Mon Oct 12, 2020, 08:57 AM
Oct 2020

538 just calculates probabilities. With HRC they had her around 65% in the last two weeks, which wasn't good. 86% is much better but I'd like it to be over 90%.

ProfessorGAC

(64,413 posts)
30. Agreed!
Mon Oct 12, 2020, 10:50 AM
Oct 2020

But, given how many scenarios come into play, I think the floor is nearly 10%.
I'm doubting it's possible we'll see Nate's model showing 93 .
Because what he doesn't do is perfectly weight the probability of each outcome. Not sure how that could be done, so I'm not criticizing Silver on that.
So, at some point the floor stops moving toward zero. I doubt we'll see a number higher than 92, no matter what PINO does to sink himself.

zipplewrath

(16,646 posts)
32. It's the nature of probability
Mon Oct 12, 2020, 08:02 PM
Oct 2020

Ultimately, even events with a probability of zero can happen, and those with one of 100% can still fail to happen. It's the nature of mathematics and that there is always an uncertainty or "limited number of significant digits" such that there is always a touch of uncertainty. And the math prevents calculations of probabilities over 100%. With these polls, the uncertainty is albeit single digits, they are often high single digits and when taken in aggregate, the uncertainty can easily grow to double digits. So as you suggest, getting much higher than the current 86% becomes increasingly difficult.

There is a type of analysis called "Monte Carlo" analysis in which models are built, and then run hundreds of times, varying the inputs for each run. I often criticized the nature of their selection of the inputs since they didn't always vary the inputs in a statistically appropriate manner (certain inputs were statistically less likely to happen and should have been inputs in very few runs). Alternately, there were combinations that were basically contradictory (cold hardware in the middle of hot deserts). But even when I forced them to correct the analysis for these things, the results often didn't change all that much.

HRC lost because of roughly 80,000 votes in three states. That amount was so small that most analysis wouldn't have even have differentiated between the cases where those votes went one way or another. It's a bit like Bush v. Gore where the difference in Florida was basically within the expected error rate of tabulation, even in recounts.

ProfessorGAC

(64,413 posts)
33. I've Used Monte Carlo
Mon Oct 12, 2020, 08:06 PM
Oct 2020

I was asked to see if I could come up with a predictive model for injury events in a manufacturing site.
It was actually quite accurate, but the most resistant folks were the safety professionals.
After I predicted 2 consecutive reportable events to within 48 hours, they doubted it far less.

zipplewrath

(16,646 posts)
34. They had many uses
Mon Oct 12, 2020, 08:24 PM
Oct 2020

One of their better uses was to identify those combinations of factors that could produce huge "outliers". Typically it was either a single variable, or a combination of two variables such that you could design the system to prevent them from happening to begin with. The other was identifying the variables that caused the largest variation. Basically it identified those that they system was most sensitive. Then you could either change the design, or restrict the exposure to wide variations in those variables.

Johnny2X2X

(18,744 posts)
10. The clock is running out
Mon Oct 12, 2020, 08:49 AM
Oct 2020

538 gave Trump a 30% chance to win on election day in 2016, so they were surprised at the results, but not shocked, if there's a 1 in 3 chance that something happens and it does, that's not a huge upset. If you have a worn tire on your car and I tell you you have about a 70% chance of driving it another 100 miles and then 50 miles into your journey you get a flat tire, was I wrong?

The 538 model has made some improvements, but more than anything the improvement we should realize is in the inputs, the polls. About half of all of the pollsters have updated their turnout models to reflect the 2016 results better. So the polls are more favorable to Don this year. The half that didn't change their turnout models didn't do so because they were closer to the results and they think the electorate will at least partially return to what their previous turnout models were.

So you got better models with better data going in. 13% is a 1 in 8 chance, the same chance as flipping a coin and having it come up heads 3 times in a row.

What I don't think these models have accounted for also, is that 10 million people have already voted. There was likely no way to anticipate or factor in early voting. I suspect 13% might be a couple points lower because so many people are voting early. A full 60% of the votes cast in this election could be cast before election day, that is beyond unprecedented.

whopis01

(3,467 posts)
19. You are correct, the models don't account for early voting.
Mon Oct 12, 2020, 09:31 AM
Oct 2020

During and after the 2016 election, Fivethirtyeight really played down the impact of early voting numbers. And they make the argument that early voting has little to no impact on turnout.


Tommy_Carcetti

(43,080 posts)
29. "(T)he same chance as flipping a coin and having it come up heads 3 times in a row"
Mon Oct 12, 2020, 10:47 AM
Oct 2020


Sorry, I just saw this movie (at long last) and your quote reminded me of this scene.

whopis01

(3,467 posts)
17. On 10/17/2016 Fivethirtyeight had Trump at 12 and Clinton at 88
Mon Oct 12, 2020, 09:25 AM
Oct 2020

The current trend is a great one. But even a 1 in 10 chance of Trump winning is way too much.

Doodley

(8,976 posts)
23. I think the key is momentum. One side has momentum, and if that keeps up, Trump's chances become
Mon Oct 12, 2020, 10:19 AM
Oct 2020

near to zero.

Johnny2X2X

(18,744 posts)
26. 538 Poll average is now Biden +10.6!!!
Mon Oct 12, 2020, 10:26 AM
Oct 2020

This is getting ugly.

So at this point, if Trump makes some huge comeback, the best he could hope for would be to be down by 7 3 weeks from now, IMO.

Also, in 2016 in the polls-plus forecast (Which is all Silver is doing this year), the best Hillary did was 84 after the Access Hollywood tape fallout drove the numbers. So Biden is officially in better shape than Hillary ever was both in the polls and the forecast.

Latest Discussions»General Discussion»Trump is down to 13 on Fi...