538 model & poll tracker thread

[ultraviolet]

He said in the aritcle that went up last night that Biden's odds would be 91% on Election Day if things stay consistent.

Given that 91% was exactly the odds he gave Obama 2012 on Election Day when Obama only had a 1.5% national polling average lead over Romney, it says a lot about how much more cautious this model is that it'd give Biden's much larger lead the same chance.

I'm not so sure about that. The model also detected structural advantages for the democrats in the electoral college in 2008 and 2012. But now it thinks Trump has a structural advantage.

Yeah because that’s the truth...

The tipping point will vote to the right of the nation this year while in 2008 and 2012 it was to the left.

[Antonio the Sixth]

He said in the aritcle that went up last night that Biden's odds would be 91% on Election Day if things stay consistent.

Given that 91% was exactly the odds he gave Obama 2012 on Election Day when Obama only had a 1.5% national polling average lead over Romney, it says a lot about how much more cautious this model is that it'd give Biden's much larger lead the same chance.

Obama had an electoral college advantage of about 2 points. Biden has an electoral college disadvantage of about 2 points. Adjust for that and the numbers are pretty comparable.

[Interlocutor is just not there yet]

I'm just happy that, after tomorrows jobs report, the 'economics' and 'uncertainty' caveat should be steadily fading away by next week

[Crumpets]

https://twitter.com/NateSilver538/status/1312023423619600386

[Figueira]

North Carolina is now expected to vote for Biden by exactly the same margin as Ohio (.6%).

[Dr. Arch]

North Carolina is now expected to vote for Biden by exactly the same margin as Ohio (.6%).

This is an effect of OH moving further left on the model, correct?

[Figueira]

North Carolina is now expected to vote for Biden by exactly the same margin as Ohio (.6%).

This is an effect of OH moving further left on the model, correct?

Yes. Both have been moving left, actually.

[Dr. Arch]

The 538 model no longer projects that Trump winning the popular vote is a possibility within 80% of the outcomes. The cone of uncertainty is no longer crossing over the margins of error for each candidate's potential final PV range.

[Figueira]

The 538 model no longer projects that Trump winning the popular vote is a possibility. The cone of uncertainty is no longer crossing over the margins of error for each candidate's potential final PV range.

It's a possibility, but it's not within the median 80%. The model currently shows a 10% chance of Trump winning the popular vote. You can see this if you scroll down just below this graph.

[Dr. Arch]

The 538 model no longer projects that Trump winning the popular vote is a possibility. The cone of uncertainty is no longer crossing over the margins of error for each candidate's potential final PV range.

It's a possibility, but it's not within the median 80%. The model currently shows a 10% chance of Trump winning the popular vote. You can see this if you scroll down just below this graph.

Yep, fixed. Thanks.

[Landslide Lyndon]

LOL! A single poll showing Biden +3 made Biden's lead in the poll aggregate crumble almost a point. And it's not like it is from an A+ pollster.

[emailking]

LOL! A single poll showing Biden +3 made Biden's lead in the poll aggregate crumble almost a point. And it's not like it is from an A+ pollster.

Can't simply re-running the model cause this?

[Figueira]

National polls don't affect the overall forecast much.

[Landslide Lyndon]

National polls don't affect the overall forecast much.

Not the model, the polling aggregate. It literally showed Biden +8.2 and after the IBD/TIPP poll was added it collapsed to 7.4. 

[emailking]

Oh whoops I misunderstood too.

[ElectionsGuy]

National polls don't affect the overall forecast much.

Here's the result of that. Every state's trend according to it's projected Biden +6.9 national vote.

NE3: D+1.5
WY: D+1.6
WV: D+3.3
ID: D+0.2
OK: D+5.2
ND: D+7.8
KY: D+4.5
SD: D+4.8
NE: D+1.9
AL: D+5.1
UT: R+4.1
TN: D+5.3
AR: D+7.7
IN: D+1.2
NE1: D+2.8
LA: D+1.9
KS: D+4.4
MS: D+1.6
MT: D+5.1
MO: D+4.1
SC: D+1.8
AK: D+2.3
TX: D+0.3
IA: D+2.0
GA: R+0.9
OH: D+3.5
NC: R+0.7
ME2: D+6.1
FL: R+2.0
AZ: D+0.8
NE2: D+0.6
PA: D+1.0
WI: D+1.8
NV: R+1.2
MI: D+3.0
NH: D+2.5
MN: D+1.6
CO: R+0.6
VA: D+1.0
ME: D+4.0
NM: D+0.1
OR: R+1.7
IL: R+5.5
NJ: R+0.4
CT: D+0.9
ME1: D+2.3
DE: D+6.0
WA: D+3.2
RI: D+3.7
NY: R+0.3
MD: R+2.6
CA: R+5.0
VT: R+1.3
MA: D+0.5
HI: R+3.6
DC: R+5.9

>30: 0-0.9
>40: 1-1.9
>50: 2-2.9
>60: 3-3.9
>70: 4-4.9
>80: 5-5.9
>90: 6+

Every single state that voted for Trump by more than 5 last time trends Democratic in 2020 (with one obvious exception: Utah). Meanwhile, Trump supposedly makes up for it in strongly Republican trending Illinois, California, and Maryland. Oh wait, this is scarily similar to the errors he's made in the past.

The Nebraska districts alone tell you everything you need to know. Supposedly rural Nebraska (NE3) will trend more Democratic than Omaha based NE2. It's absurd on it's face. In other areas, you can tell how heavily the model relies on polls. Why the difference between Oregon and Washinigton despite being very similar states? - relatively good polls for Trump in Oregon, and very bad ones for him in Washington. The polls in Maine, a rural, very white, and relatively non-college state, have been absurdly pro-Biden and that affects it to the point where Biden is favored to pick off ME2 (Hillary was also slightly favored there according to them in 2016).

It doesn't matter if Biden wins this election or not. This is such an obvious fraud of a model. This is nothing more than a fancy adjusted poll average, Silver and co at 538 refuse to change, and they will make the same errors in the same ways they did in both 2016 and 2018*. The state projections continue to be too good for Democrats compared to the national vote. It's possible that he's underestimating Biden in the popular vote, which is the only way some of these projections will end up being accurate, but it still stands: There continues to be a dissonance between state and national polls (and therefore his projections), especially in white non-college heavy states. And that has NOT been "fixed" in 2020, and it's clear as day at this point.

[emailking]

The Nebraska districts alone tell you everything you need to know. Supposedly rural Nebraska (NE3) will trend more Democratic than Omaha based NE2. It's absurd on it's face.

It's not. Here's a counter argument. It's much more Republican so it has more room to move Democratic in a polarized society. Here's another counter argument, you and others in this thread seem to base a large fraction of your criticism in how the tails are modeled, scenarios which are very unlikely to unfold and it agrees are very unlikely to unfold. I don't care if Oregon goes Republican before TX once in a blue moon. I care if Texas is going blue on Nov 3.

In other areas, you can tell how heavily the model relies on polls. Why the difference between Oregon and Washinigton despite being very similar states? - relatively good polls for Trump in Oregon, and very bad ones for him in Washington.

This is supposed to be a bad thing?

[Antonio the Sixth]

"Model that's explicitly polls-driven is reliant on polls (shocking!) and therefore affected by polling issues" is not a damning criticism. That's just the constraint inherent in the exercise.

You know, it's really telling how people are incapable to simply levy empirical criticism at the 538 model and try to analyze their roots and potential fixes that could be applied, but instead just go off on unhinged rants about how terrible he is and how Nate Silver is a disgusting fraud (by... building a statistical model with a perfectly transparent methodology and being extremely open about its limitations). That suggests pretty deep insecurities if you ask me.

[Interlocutor is just not there yet]

"It's only a model"

"Shhh!"

[woodley park]

Biden now up to 81 in the model, following this morning's bombshell poll releases.

[ProgressiveModerate]

Any good model SHOULD underestimate partisanship in safe states, since the distribution outcomes should be skewed towards the center (take NY, maybe on average Biden wins by 25%, but one standard deviation away would be 28% and 20%), with the mean being lower than the median, which is lower than the mode. If they're margins are based off the median or mean margin from their simulations, on average, the model should underestimate partisanship, right? Or is my logic flawed.

[Crumpets]

They now have Florida (barely) to the left of Arizona. Not sure if I agree.

[ProgressiveModerate]

They now have Florida (barely) to the left of Arizona. Not sure if I agree.

it's because of the NYT poll from yesterday

[cp]

"It's only a model"

"Shhh!"

Aaaaaaaaaaggh!

[Interlocutor is just not there yet]

When will the "Economics & incumbency" caveat start to fade? Was it expecting a much better jobs report that wouldn't be overshadowed by the President being hospitalized?

It's the only metric preventing Georgia from being 50/50 on the model