Any honest prognosticator has to state that there is no meaningful data to enter into the models yet. Probably more so for a fundamentals heavy model that discounts polling at this point.
Might be nice to put up a statement to the effect.
Yes, Nate Silver has also suspended his predictions for a week or so, until polls of Harris-Trump start coming in.
At this point, yeah, we’re definitely in uncharted waters.
But there were still signs arguing against the reliability of the new 538 before this shakeup in the race.
The fact that the decades-experienced, consistently winning politicians all decided that Biden was dead in the water, while 538 was giving a 50/50, would say that we probably shouldn’t give it much credence.
About 1/3 of them, said that and 2/3rd said they back Biden.
538 speaks, giving preliminary state of play of the Harris-Trump election:
… Democrats will hope that her polling improves as she mounts an active campaign against Trump. For one thing, Harris could experience something of a quasi-convention bounce in the wake of the coverage of Biden’s departure and her now-active candidacy. Moreover, until Sunday (July 21 - b), any poll that tested Harris (or any other possible Democratic candidate) had been hypothetical for survey respondents. That is not the case anymore, which could at least partially reset the race and shift how some voters are thinking about it moving forward. For Democrats, Harris could represent an escape hatch from the doom spiral of negative media coverage that had hounded Biden following the June debate, as well as the heightened concerns about his capacity to serve at 81 years old and fears that he could not recover the momentum against Trump.
Ultimately, Harris’s seeming upward trend in the polls and the fact that she’s polled better than Biden in many recent surveys will give Democrats reason to hope that she has a higher ceiling of support than the president, whose polling numbers had remained pretty stagnant throughout the campaign. This seems especially true because fewer Americans have expressed an opinion about Harris than Biden in approval polling. In theory then, the public has less concrete views about the vice president, which could allow her to win over some voters who didn’t want to consider Biden.
That seems unlikely, given the result.
I’m also curious how you get to 3 from Biden, Jeffries, Obama, Pelosi, and Schumer? I count 5 and, as said, if the majority was voting that he was solid to win it then his decision to drop out seems pretty mysterious.
I was basing this on the NYT article posted in this forum lising Dem Politicos who either wanted Biden to withdraw, or backed him. The number who wanted him to with draw was 40, the number who backed him was 75.
And Obama and Pelosi for sure (not sure about the rest) just expressed 'concerns"- of course the polls are not great and Biden just got Covid- of course they would have concerns.
I’m doubting that he switched paths on the basis of polling of Democratic figureheads, speaking in public.
If Biden switched, it would be because he had some hard numbers, some professional campaign analysts, and some other professional campaigners looking over the same with him, together in a room, having some serious discussions. The best chance was to step aside.
Morris is quite familiar to members of #ElectionTwitter, which is a sort of loose accumulation of prognosticators and followers that coalesced in 2015 or 2016. He and Silver went at it back in 2020, and my general sense back then was that both guys got kind of personal very quickly…it’s no surprise if they still don’t get along.
None of the major election prediction folks had Biden dead in the water, though nearly all of them did give the advantage to Trump. That’s a rather odd way to judge how much credence to give to any of the models.
A major point of modern statistical analysis is that the gut instincts of decades-experienced, winning politicians (or baseball professionals - where Nate Silver started out) is often wrong, sometimes hilariously so.
If a particular model seems rosy, it may be. Or maybe they have some secret sauce. Unfortunately, how they generate the current 538 model is not publicly disclosed, so it cannot be validated against past results (a good way of checking it isn’t totally bogus).
Either way, deciding a particular model is insufficiently objective by comparing it to a subjective measure seems ironically counterproductive.
That’s true, but note the difference in information. We didn’t–still don’t–know the extent of Biden’s mental and physical health problems. Insiders have greater insight. The prediction model can, via polls, take some of this risk into account, but not all of it. Nate himself has said that his gut instinct was that a Biden win was more like 10-15% rather than the ~25% the model predicted. Probably because his own prior includes Biden actually being some degree of senile. The lower number is also (approximately) what the prediction markets gave.
(Emphasis added)
Biden wasn’t talked out of it by Nate Silver. He was talked out of it by Pelosi and Obama.
And Tony LaRussa with his decades of experience at winning took the White Sox to the World Series again…or not.
This is my point - relying on the gut instinct of “experienced” insider experts is kind of silly. We go to these models because we want something with some sort of testable and repeatable basis to it. You know…science.
Nobody knows what is happening in the 538 model. It may be total bunk or may have some validity to it.
But concluding it is bunk because it goes against those experts and not due to any actual analysis makes no sense. Might as well toss baseball analytics out the window because a lot of experts think their own instincts are better than the stats.
Right now, it is untested. I personally don’t trust it, but not because the experts say something different. I don’t trust it because it has no track record, and I don’t know what is going on under the hood, other than the fact that poll data is not a big part of the current model (which we only know because that’s what they tell us).
On the other hand, Nate was among the very first well-known pundits to call clearly for Biden to drop out. He did so the morning after the debate, five days before our “Biden may drop out” thread started. I’m not giving him credit/blame for Biden leaving; if it hadn’t been him someone else would have been first. I think this says a lot for his acumen as a pundit aside from his statistical expertise, though.
To be fair, there were some calling for that in the debate thread on the night of the debate itself, so preceding Silver
Sure, but was anyone doing that? People concluded it was bunk because:
- It didn’t match other models, like Silver’s or the Economist’s
- It was very far off from the prediction markets
- It had some weird inconsistencies, like state predictions that were even stronger than what the fundamentals said
- It didn’t seem to take into account obvious current events like Biden’s health, even indirectly
And in retrospect, giving ~50% to Biden winning seems very optimistic. Obviously a 50/50 in isolation can’t be said to be wrong. But it looks like we were in fact very far from Biden actually winning. It was so far from being a close race that he dropped out. You don’t do that for a 50/50 shot, or even a 25/75 shot.
…the post I was directly responding to. I even quoted it.
It denigrated the model because the “experts”, i.e. winning politicians with decades of experience, gainsaid it.
I think so, but until tell-alls, and book length journalism, appear, we do not know.
Did Nancy and Barack read both 538 and Nate, and side with Silver? If not, they probably listen to advisors, and the advisors turned against 538.
In any event, yesterday was a bad day for 538. I do not think anyone who thought the election was, pace 538, a complete toss-up, would have advised Joe to drop out.
Nate snarking on X:
“Can the President consistently speak in coherent sentences?” is a salient question, despite not being one of the Thirteen Keys.