I think they were running it for a while but started publishing it in August.
The Economist has published their election model. They have detailed outputs similar to FiveThirtyEight’s. Right, they have Biden at about 6 in 7 chance to win (versus 3 in 4 at 538).
I am not exaggerating when I say I am literally praying that these are accurate and Joe Biden can win this. Please, Lord, let this be the end of Trump in office. I can’t imagine the suffering of another four years.
I don’t have any strong thoughts about the 538 model versus the Economist model. I have read a bit of back-and-forth on Twitter but both seem to be created by smart people with reasonable differences. I suspect they will converge as we get closer to election day.
But man from a graphic design perspective, 538 is a disaster. Fivey Fox has been mentioned but the whole top of the page is chartjunk, with a huge number of maps which don’t really tell you anything important. By contrast the top of the Economist page gives you important information which is regularly updated: the probably of a win for Biden and Trump.
Also which genius thought it would be a good idea to have text boxes with slanted text on 538?
I see the Economist has the Dems with a 67% of controlling the Senate! Mind you, it has Dems winning 51.4 seats and Reps 48.6, so there’s no real room for error. But still…
It’s not just the election side of the house; the sports sides are no better. I cannot think offhand of any website today with such a massive disparity between high quality content and absolutely appalling design.
This was all on purpose, to slow the viewer down so that, by the time they get to the bottom, they understand what “Biden currently wins 77 of every 100 times we run the model” MEANS.
Nate et al. are understandably miffed that so many people in 2016 misunderstood what that 77 number (say) means, and blamed them for getting Clinton’s chances “wrong.”
I’m mostly okay with the “easing” into the projection that the formatting encourages. I think that’s a good idea. I do find the maps specifically to be ugly and overly busy. I thought they were just decorative (like an ugly wallpaper) until today when I ready the “Fivey Fox” comment. Pretty lame.
He doesn’t seem to get miffed when people say he got something “right.”
I don’t know of many situations where someone would actually get “miffed” about that conclusion about themselves.
I am sure that the polls are NOT accurate.
Because there is a huge issue that the polls do not (and can not) address: ballots which will not be counted.
From a New York times article back in 2012, (paywalled, but readable if it counts as one of your 5 free articles)
https://www.nytimes.com/2012/10/07/us/politics/as-more-vote-by-mail-faulty-ballots-could-impact-elections.html
The article mentions several problems with absentee ballots in general…(not related to Trump or to this years criminal behavior by the Post Office etc.).Sometimes signatures don’t match, sometimes people don’t follow instructions and fill out the absentee form incorrectly. Whatever the reason, if two percent of the ballots were discarded in 2012, when many fewer people used them, it is quite possible that this year, maybe three percent of ballots will be invalidated for the same type of mistakes. And this year , of course, we have a new situation never seen before in recent US history: outright criminal activity encouraged by the Republicans. I would expect another one percent or more of ballots to “accidentally” disappear. A few boxes in the basement of the city archives will be “forgotten”, others will never arrive, since they were “accidentally” sent to the Post Office’s department of lost letters. Polling places will be located in inaccessible places, etc.,And most of these “accidents” will occur in blue states.
So it seems very, very likely that as much as 5 % of the total vote will not be counted. And those lost votes may well split 4% Biden and 1% Trump–a 3% difference.
So subtract 3% of total votes from Biden’s current very narrow lead in the only states which matter.
Subtract another 3% due to the legitimate margin of error in the polls.
Subtract 6 from Biden, add 6 to Trump…and the race becomes much close to make any predictions.
So despite all the dry and detailed scientific analysis by highly qualified people at 538…there is one huge fact that is being overlooked-because it is unknown and untestable: :the number of ballots that will not be counted.
And sometimes, when science doesn’t provide an answer, you have to go with your gut level emotions:.
Trump’s supporters are far more enthusiastic, and they’ve controlled the narrative for the past 6 months, while Biden tries to play catch-up from behind. Nothing Trump says will cost him votes, while Biden’s gaffes leave his own supporters sweating with fear at every appearance.
I fully expect Trump to win the election.
(And I hope I;m wrong)
correction: that should say
“the race becomes much TOO close…”
My point is, if he says Candidate X has a 77% probability of winning:
If Candidate X loses and people say “You were wrong,” he says “No I wasn’t.”
If Candidate X wins and people say “You were right,” he DOESN’T say “No I wasn’t,” he sells his website to NYT.
They’re only been out for a few presidential election cycles. If they stick around for, say, 10, and manage to predict 9 out of the 10 with 75% probability, I can see him saying to people that say “he gets it right” “no, our confidence intervals should have been higher”.
These forecasts assume the usual circumstances and conditions, which may not be accurate this time. I doubt it factors the possibility of a contested election.
Not true. Nate has said several times that, if the higher-probabiliy-of-winning candidate always won (let’s say over five or ten elections), his models (one per cycle) would probably need some improving. Somethings not right if that 20% (or 25%, or 30%) probability NEVER comes true.
“Being right” (as you put it) about any single event has no meaning whatsoever to Nate Silver. I’m sure he’s never used the phrase.
Also, it less about who wins and who loses and more about if the result falls within the model. If Hillary had won, but the results for each state were way off from what was predicted, that’s just as wrong of a result.
I didn’t say HE ever said he was right. Other people do and he does a shoegaze or something.
He’s not doing it as a hobby. People thinking he was “right” means $$$$$.
Perhaps, but I meant that literally. Forget Nate Silver or any one human’s motives — what I tried to convey is that “being right about one event” has no meaning in probabilistic modeling.