538 model & poll tracker thread

[GeorgiaModerate]

@LimoLiberal or a mod: can you please remove (model coming on Wednesday) from the title?

[wbrocks67]

Can someone explain how the odds dropped from Biden 73/100 to 70/100 overnight?

Biden's nat'l average literally hit +9.3 yesterday and is down to +8.8, but only b/c of Ipsos (7) and Change (, but those are still on par with recent.

Only state polls I've seen are Trafalgar, and the Change polls which were 4/6 better than two weeks ago.

I'm finding it odd how when Biden gets good polls, the odds barely budge up, but then if there's like 1 or 2 bad (not even really that bad) polls, it drops a whole 3%? What?

[Gustaf]

Can someone explain how the odds dropped from Biden 73/100 to 70/100 overnight?

Biden's nat'l average literally hit +9.3 yesterday and is down to +8.8, but only b/c of Ipsos (7) and Change (, but those are still on par with recent.

Only state polls I've seen are Trafalgar, and the Change polls which were 4/6 better than two weeks ago.

I'm finding it odd how when Biden gets good polls, the odds barely budge up, but then if there's like 1 or 2 bad (not even really that bad) polls, it drops a whole 3%? What?

Doesn't the model have some sort of convention bounce effect? That might affect it. It might think Biden should be on a convention bounce high right now or something.

[GeorgiaModerate]

Meanwhile, Biden's chances in the Economist model have been gradually climbing back up and were 90% in last night's run...let's say if they change significantly this morning.  They update the model 2 or 3 times a day, usually morning and evening, and sometimes in the middle of the day.

[wbrocks67]

OK, yeah, I'm convinced Nate's model is junk. His Twitter thread makes no sense. He admits that the model is not expecting a convention bump this year. And then he basically says that because of a few Change polls (that 4/6 of them are actually UP from where they were two weeks ago), that that why it dropped.

He also says that the convention polls have not been good when most have seen an uptick of about 1 pt (YouGov, MC) and them, besides the Change polls, have BEEN THE ONLY POLLS WE'VE GOTTEN! There has been NO high quality live phone polls. And yet, here we are, with Biden dropping 3% in the model.

I really can't

https://twitter.com/NateSilver538/status/1298601607689580544

[wbrocks67]

G Elliott seems to agree

https://twitter.com/gelliottmorris/status/1298604889124347904

[GeorgiaModerate]

Elliott also has a good thread on why boosting Republican turnout is not, by itself, enough to give Trump a win.  It starts here:

https://twitter.com/gelliottmorris/status/1298469783164391425

[Annatar]

Looking at how Nate has described the 538 model vs the Economist model it seems the main difference is the 538 model is basically completely driven by state polling with national polls only being used to apply a trend adjustment whereas the Economist model is built to a larger extent around national polls.

Because of this it is quite possible Trump's chances of winning won't drop that much in the 538 model even as we get closer to election day as it is based off state polls and if polls like Marquette which had Trump up 4% in WI and Muhlenberg which had Trump up 4% in PA keep coming out, even on election day Trump will have a reasonable chance of winning due to the possibility of a polling error.

The 538 and Economist model also use different polls, for example Morris has stated their model does not use Emerson at all because they don't like the polls' methodology whereas 538 does use Emerson so the 2 models differ on what polls they use as well.   

[American2020]

https://twitter.com/NateSilver538/status/1298601607689580544

[TiltsAreUnderrated]

Have they confirmed whether or not they're doing a Senate model this year yet? I haven't found mention of one.

[ProgressiveModerate]

Have they confirmed whether or not they're doing a Senate model this year yet? I haven't found mention of one.

Really want them too. The senate is highly competative this year and we have no good statistical senate models, except from Decision Desk, which I don't like just for the fact AL is "Lean R"

[ultraviolet]

It’s really bothersome how little information the design of this model makes easily available to viewers.  It seems like the only message they want to communicate is “I guess anything could happen?!?”.

Where’s the “polls only” forecast?
Where’s the “now cast”?
Where’s the easy link to the database of polls used to generate the projections?
Where’s the accounting for how individual “fundamentals” factors influence the projections?
Where’s the map??

All of this was available in the 2016 model, which actually performed quite well.  This is such a huge step backward.

He has said since 2017 that he's never doing another Nowcast and it was a mistake to make it, so I'm not surprised it's gone.

The rest, though, should be there. I really want that list of probabilities of various interesting things (Electoral tie, landslide, EV/PV split, etc) that they had in 2016.

There is a link at the very bottom of the "National overview" page that will download model outputs. Besides more precise numbers, there is a file called "Scenario Analysis" that includes the chances of:

1: An Electoral College tie. (0.3%)
2: A presidential recount. (0.5%)
3: Trump winning the popular vote. (16.6%)
4: Biden winning the popular vote. (83.4%)
5: Trump winning the popular vote but losing the Electoral College. (0.1%)
6: Biden winning the popular vote but losing the Electoral College. (10.2%)
7: Trump winning a majority. (12.3%)
8: Biden winning a majority. (78.5%)
9: Trump winning by over ten points. (1.2%)
10: Biden winning by over ten points. (31.9%)
11: The same map as 2016. (0.0%)
12: Trump winning a Clinton state. (41.1%)
13: Biden winning a Trump state. (88.0%)

All of this is interesting, and it's valuable to have, but it is definitely odd that it requires a separate file download to find.

How is the 2016 map 0.0%?

[redjohn]

A model dropping one candidate's chances by 3% due to polls shifting in that candidate's favor seems... sketchy.

[ProgressiveModerate]

A model dropping one candidate's chances by 3% due to polls shifting in that candidate's favor seems... sketchy.

Maybe it's dropped beacuse they were expecting a bigger Biden bounce from the DNC? I've long thought that convention bounces are pretty much dead, and tbf we haven't gotten many new polls lately, but that's my guess

[wbrocks67]

A model dropping one candidate's chances by 3% due to polls shifting in that candidate's favor seems... sketchy.

Maybe it's dropped beacuse they were expecting a bigger Biden bounce from the DNC? I've long thought that convention bounces are pretty much dead, and tbf we haven't gotten many new polls lately, but that's my guess

Well it's confusing, b/c Nate literally said that the model was not anticipating a big bounce this year for either of them. But then he's saying the model is still being effected by there not being one...

[ProgressiveModerate]

A model dropping one candidate's chances by 3% due to polls shifting in that candidate's favor seems... sketchy.

Maybe it's dropped beacuse they were expecting a bigger Biden bounce from the DNC? I've long thought that convention bounces are pretty much dead, and tbf we haven't gotten many new polls lately, but that's my guess

Well it's confusing, b/c Nate literally said that the model was not anticipating a big bounce this year for either of them. But then he's saying the model is still being effected by there not being one...

I just don't get how he can say that when we've gotten so few polls the last few days.

[Gass3268]

Change Research shows Biden gaining two points in North Carolina compared to their last poll, but Trump's chances of winning the state go from tied to up 6?

[wbrocks67]

A model dropping one candidate's chances by 3% due to polls shifting in that candidate's favor seems... sketchy.

Maybe it's dropped beacuse they were expecting a bigger Biden bounce from the DNC? I've long thought that convention bounces are pretty much dead, and tbf we haven't gotten many new polls lately, but that's my guess

Well it's confusing, b/c Nate literally said that the model was not anticipating a big bounce this year for either of them. But then he's saying the model is still being effected by there not being one...

I just don't get how he can say that when we've gotten so few polls the last few days.

Well, that too. The fact that this model changed so much with a set of *Change Research* polls is also highly suspicious.

He admits in his Twitter thread that there hasn't been a ton of polling, yet, his model seems to be completely overreacting to it..

[wbrocks67]

Change Research shows Biden gaining two points in North Carolina compared to their last poll, but Trump's chances of winning the state go from tied to up 6?

Same reason I don't understand how his model has Trump winning 5 in Texas when we've gotten 75% of polls in the past 3 months saying the race is basically a dead heat.

[ProgressiveModerate]

Change Research shows Biden gaining two points in North Carolina compared to their last poll, but Trump's chances of winning the state go from tied to up 6?

Same reason I don't understand how his model has Trump winning 5 in Texas when we've gotten 75% of polls in the past 3 months saying the race is basically a dead heat.

It's all because of "fundementals", despite having nearly 200k deaths from COVID and high unemployment. He claims that polling will be weighted more as the election nears.

[Admiral Stockdale]

I am not sure how much the 538 model starts to remove a poll's "weight" the further back it is on the calendar.  Are they completely tossing a June or July poll if they have an August poll or does it just drop from a weighting perspective (e.g., 20% to 5%)?

Then there's the impact of the polster's rating and THAT weighting...  like a B/C polster gets reduced by 50%, but an A+ polster is 175%.

[Fmr. Gov. NickG]

It looks like the latest Rasmussen poll has dropped Biden down to only being "slightly favored" in the race (which I suppose it their categorization for 60%-70%?).

I have a lot of problems with this model.  But while I think the bump today is pretty baseless, I don't really blame the model for it. 

A model that relies so heavily on polls is only going to be as good as the quality of the polls it has available to input.  We've only been getting garbage polls, and so it's understandable when it spits out garbage.  The blame lies in the lack of quality polling since the DNC.

[Crumpets]

Here are the current swing and trend maps based on Nate's numbers:

Swing

Trend

[ProudModerate2]

Can someone explain how the odds dropped from Biden 73/100 to 70/100 overnight?
...

Now, later in the day, it stands at 69/100.
Ugggg.

[Fmr. Gov. NickG]

Don't talk to me about the race "tightening" until this changes: