Key excerpt (But it's worth reading the full thing):
But the real value-add of the model is not just in calculating who’s ahead in the polling average. Rather, it’s in understanding the uncertainties in the data: how accurate polls are in practice, and how these errors are correlated between the states. The final margins on Tuesday were actually quite close to the polling averages in the swing states, though less so in blue states, as I’ll discuss in a moment. But this was more or less a textbook illustration of the normal-sized polling error that we frequently wrote about [paid only; basically says that the polling errors could be correlated be correlated between states]. When polls miss low on Trump in one key state, they probably also will in most or all of the others.
In fact, because polling errors are highly correlated between states — and because Trump was ahead in 5 of the 7 swing states anyway — a Trump sweep of the swing states was actually our most common scenario, occurring in 20 percent of simulations. Following the same logic, the second most common outcome, happening 14 percent of the time, was a Harris swing state sweep.6
[Interactive table]
Relatedly, the final Electoral College tally will be 312 electoral votes for Trump and 226 for Harris. And Trump @ 312 was by far the most common outcome in our simulations, occurring 6 percent of the time. In fact, Trump 312/Harris 226 is the huge spike you see in our electoral vote distribution chart:
[Interactive graph]
The difference between 20 percent (the share of times Trump won all 7 swing states) and 6 percent (his getting exactly 312 electoral votes) is because sometimes, Trump winning all the swing states was part of a complete landslide where he penetrated further into blue territory. Conditional on winning all 7 swing states, for instance, Trump had a 22 percent chance of also winning New Mexico, a 21 percent chance at Minnesota, 19 percent in New Hampshire, 16 percent in Maine, 11 percent in Nebraska’s 2nd Congressional District, and 10 percent in Virginia. Trump won more than 312 electoral votes in 16 percent of our simulations.
But on Tuesday, there weren’t any upsets in the other states. So not only did Trump win with exactly 312 electoral votes, he also won with the exact map that occurred most often in our simulations, counting all 50 states, the District of Columbia and the congressional districts in Nebraska and Maine.
I don't know of an intuitive test for whether a forecast of a non-repeating event was well-reasoned (see, also, the lively debate over the performance of prediction markets), but this is Silver's initial defense of his 50-50 forecast. I'm unconvinced - if the modal outcome of the model was the actual result of the election, does that vindicate its internal correlations, indict its confidence in its output, both, neither... ? But I don't think it's irreconcilable that the model's modal outcome being real vindicates its internal correlations AND that its certainty was limited by the quality of the available data, so this hasn't lowered my opinion of Silver, either.
Jump in the discussion.
No email address required.
Notes -
I supposed we will have to agree to disagree that it is unfair to request evidence to substantiate an accusation, particularly when that accusation is a form of argument.
'The accused did not defend themselves' is so flawed a standard for determining truth that Americans prohibit it as a legal standard at a Constitutional level as an element of self-incrimination protections (you cannot be forced to testify, but also your refusal to testify is not itself self-incrimination), and while we are not a court of law it does lead to why it's a poor argument on grounds of reputational damage. Even Nate defending himself in this twitter at this point can inflict reputation damage to Nate- after all, it could be presented / accused as Nate lying / trying to deflect from the poling-result mismatch / doing a condemnable thing in and of itself. General decorum is to not air legal conflicts on your pseudo-business reputational Twitter account, particularly when they are about a prediction that you got wrong- just re-raising the topic is causing reputational harm.
And this is without other possible reasons of reputational harm for not going into detail of why an offer might not have been carried through. Like, say, the fact that Nate Silver is a married men, and married women are often quite willing to have strong words with their husbands when said husbands start offering 5-figure bets. Or maybe his finances are not so liquid and stable as to set a 5-figure risk. Or that he was drunk and not in a good state of mind when he made the offer. Nate is already suffering reputational harm for coming off as a bet-welcher, but he could suffer more reputational harm if he provided more context... and this is regardless of whether there was an actual acceptance-offer made.
It's also a defense that doesn't really acknowledge the shifts in which reputations are being challenged how.
Note here that Nate's reputation isn't regarding whether he would fail to pay a bet. It's not even about if he made a bet. The reputation in question is if Nate's reputation as a political data analyst, i.e. whether his analysis of election, is sound enough to warrant respect.
Whether Nate is willing to actually to make a bet or not isn't actually a failure of his analytic capacity (i.e. whether he should be considered valuable input), it's a failure of an additional standard introduced as a basis of dismissal (if Nate Silver is not willing to go through with a bet, his work can be dismissed). Nate Silver's [Reputation as a political scientist] and [Reputation as a confident better] are two different reputations, which are being conflated to use the one that is easier to condemn [reputation as a betting man] to dismiss the other [reputation as a political scientist].
Since the [reputation as a betting man] is being used as a proxy argument for [Nate Silver's arguments can be dismissed without addressing], the role of evidence is in turn strengthening or weakening the proxy argument. This seems like a reasonable request, but I will admit that is my own opinion.
It is not his reputation as a confident bettor at stake, it is his reputation as an honest man which reflects on both the value of his words as a bettor and his words as a political scientist. Nate has shown himself to be a weasel with no integrity and that should absolutely inform your opinion of his political commentary as a good Bayesian
Hinging an argument on an accusation of dishonesty is precisely why I feel it is reasonable to request evidence of dishonesty, lest that accusation of dishonesty also be dishonest.
Ranger does not appear to have meant to make an argument about Silver's reputation as a bettor in the sense of upholding a bet, but if you wish to on the basis of Nate Silver being 'a weasel with no integrity,' I would make the same question of you: please provide the evidence that Nate Silver weaseled out of the bet.
I don't believe I have ever claimed to be a Bayesian, or ever particularly cared about others being Bayesian, particularly when the claims of being a good one or not revolve around character rather than statistical grounds.
In fact, for transparency I have a general skepticism of arguments about when other people are a 'good' [Insert Applause Category], since in my experience these often attempts to use assumed category requirements as a cudgel in either a No True Scotsman fallacy sense (if you wish to qualify as [Good Group], you must meet my critiera) or in ad hominem effect (this person is not [Good People]).
But no one accused him of dishonesty. Nate never said "I didn't get any contract", that's my entire point! It's his opponent that exposed himself to an accusation of dishonesty if and only if he didn't send the contract. This, and the fact that you thought it's his reputation as a better that's at stake, makes me think you're not really getting the logic behind my reasoning, but I don't know how to explain it any better.
...er, yes, there has been. That is one of the implications of the phrase 'weasel out of a bet,' which has been invoked in this Nate Silver context*. In forum and elsewhere, the Nate Silver's bet post is being used to charge Nate Silver of dishonest for not following through with his offer for a bet.
*Though not by Ranger specifically.
Nor, to my knowledge, has Nate ever said he did get any contract. Hence you do not have a point- you have an absence of information.
Hence why I am asking for some support that he received the contract, as opposed to working from a position of assuming he did.
Incorrect. Nate's opponent would only expose himself to an accusation of dishonest if and only if he claimed to send the contract but didn't.
However, if Nate's opponent has not made a claim, he would not be dishonest regardless of whether he sent a claim or not.
Unless I misread something earlier, at this point and in this thread, no evidence has been provided that Nate's opponent has claimed to have sent a contract. IF Nate's opponent has made a claim, THEN that claim could be looked at for evidence of credibility- for example, if the claim was made before the election (when results were still uncertain) that would be more credible than the same words made after the election (when the results are now hindsight)- but no consideration can be made absent of existence, and without existence of a claim from Nate's opponent then Nate's opponent cannot be dishonest about said claim.
No, I get your logic behind your reasoning, I just think it's a very poor counter to a request for information, and does not warrant accepting an assumption that is required for various arguments to be valid.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link