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 -
There's doing the thing and then there's talking about doing the thing.
One of the unfortunate results of social media and the internet more generally is that it allows people to show you what they're doing in another domain. This is everything from innocent cooking videos, to building things, all the way up to a statistical analyst (Silver) showing you what and how he is statistically analyzing. While this may let new people learn things they previously had limited information on, I believe it corrupts the thing itself (i.e. the cooking, the building, the numbers crunching).
Why? Because the focus of the "creator" turns to audience views, approval, satisfaction etc. instead of doing the thing itself well. Cormac McCarthy once said he never hung out with writers because, when writers get together, they talk about writing - where any writer who was serious would just go and write!
The French Gambler is a great antagonist to Nate Silver. He didn't care about explaining himself, he didn't need to find a way to capture and entertain an audience. He focus on the thing - betting intelligently on the election - and he executed it well.
I've toyed around with starting my own blog (I won't share about what because it's niche enough as to be identifiable) but this is the number one reason why I haven't so far. I worry that the thing I'll be writing about will suffer because my focus will be writing-about-the-thing instead of doing-the-thing.
I think that's what's happened to Nate. He's become so focused on writing about how is models work, or how the polls are biased that he's no longer spending the mental bandwidth necessary to build the best possible model. How could you even do that when you're setting daily print deadlines for yourself. Furthermore, a lot of the time, models have very weak explanatory and/or predictive power. How do you maintain an audience by writing "Eh, looks kind of ... like nothing. Whatever." Nope. You need 1,000 words that "examine the correlate intricacies and inherent cognitive blah blah blah blah."
I don't know what Nate Silver wants. Does he want to write about probability and statistics? Does he want to build models to predict events? Does he want to play professional poker? It doesn't matter, but he should probably choose one thing to do and another, different thing, to talk about doing. When you stack all of that on top of one another, you get a 50/50 chance of being either wrong or insufferable.
He wants to do all three. He's built a very succesful substack that earns plenty of money and I think he has a moderately succesful pro poker career too, I don't think he's missing out by not hyper specializing.
And if anyone thinks they have better estimates of who'll win elections than Nate's models, feel free to bet against him. Either through prediction markets, or if you're willing to bet a large sum he'll probably be down to do a direct bet against you through an escrow service.
I'm pretty sure the French gambler could've just bet directly against Nate Silver too and probably could've gotten better odds/smaller fees than going through Polymarkert, for at least a portion of his bet.
More options
Context Copy link
More options
Context Copy link