FiveThirtyEight's election forecast is out!

I posted this in the “How big will Biden’s margin of victory be?” thread, but it deserves it’s own, to counter the preposterous complacency in that thread.

FiveThirtyEight’s election forecast is out today, and it gives Trump a 29% chance at winning. This is almost exactly the chances Trump had in 2016, and Trump Won.

Anyone thinking Trump has no chance and Biden will glide to a win is woefully mistaken. Trump could very definitely win, even with Covid and a bad economy and all the other stuff going on. Complacency and overconfidence will ensure that win. Cut it out.

Not to suggest complacency, but 29% two months before the election is different (in FiveThirtyEight terms) than 29% on election eve - the two-month out forecast gives chances of conditions/events changing (i.e. incorporates more uncertainty if I’m understanding correctly).

Unfortuntately, they don’t seem to have the “if the election was held today” percentages like they did last time to show the contrast. (And actually, as someone who’s been previously very impressed with how they display and communicate model results and uncertainty, that display design is absolutely god-awful, just one more “fluff over substance” currently-hip designs where you have to scroll scroll scroll for anything useful - fivey-fox? really??)

There’s a companion article that summarizes the current state of the race. And I’d think the “if it were held today” vs previously estimates clock starts today, so I’d expect that may be added to the forecast in the “how the forecast has changed” section.

The thing to remember that if another 30% chance event happens then it starts to cast doubt on the model. Sure its possible again for the margin of error to bite exactly right in the exact spot to make a difference but I would start to believe that 538 is undercounting their errors bars at that point.

Of course on the other side there is a non zero percent chance for Biden to win 500 EVs which would be hysterical.

Not sure I agree. If 100 30%-forecast wins happen, then yeah, the model is broken. Two? Not so much.

That article notes that the “election held today” odds are 93% for Biden.

Take what happens if we lie to our model and tell it that the election is going to be held today. It spits out that Biden has a 93 percent chance of winning.

I did say the model is broken if two 30%ers come through just that it starts to cast doubt on the model and modeling.

The best way to evaluate Silver’s models is state by state and clumping states with similar percentages. if he’s correct on 2/3 of the 66% states then we can say his percentages are close.

I hope they continue to show the “if it happened today” number.

I’m guessing that’s if all legally cast ballots are included and counted in a timely fashion?

IOW not what’s likely to happen, namely, dirty tricks, e.g., losing ballots; disqualifying ballots for spurious, arbitrary reasons; shutting down polling places so that a county of, say, 500,000 mostly minority voters has two polling places; making up rules that say you can’t carpool to the polling place; stationing armed federal troops near polling places to guarantee “security”; etc.?

Votes will be the smallest part of what decided this election IMHO. And I hope and pray that I’m wrong.

Yes, the article explicitly says they do not add an uncertainty factor based things like that.

We assume that there are reasonable efforts to allow eligible citizens to vote and to count all legal ballots, and that electors are awarded to the popular-vote winner in each state. The model also does not account for the possibility of extraconstitutional shenanigans by Trump or by anyone else, such as trying to prevent mail ballots from being counted.

I said more or less the same thing in other 538 thread. This election is going to be decided by counted votes and not by cast votes. The Republicans are going to bring out everything they can for this one, because the stakes are huge. I honestly believe that democracy vs. autocracy is up for grabs in this one. If the Republicans are able to blatantly steal the election, then I don’t see any reason why that wouldn’t become the new normal. America will still have elections, but you’ll have results like those in Belarus.

Sadly, the only people that can stop it are Republican voters. The Republican party needs to lose by so much that they return to normalcy and a commitment to democracy.

Very concisely put!

The best way to evaluate Silver’s model is Brier score or some other equivalent version of root mean squared error. That’s how Nate checks himself.

29% is way higher than I would like. Yikes.

And it does not factor in cheating, voter suppression, etc. Could be as high as 40% chance, I think.

I don’t know where to post this, but I’m only going to say it once, so here is as good as anywhere.

It’s not about how likely that stuff is to happen. It is very likely to happen. It’s about what the Democrats do about it when it does happen. If they remain true to form, they’ll roll over and play dead. And that just won’t cut it this time around.

And to address complacency. Please, for the love of Og, do not be complacent. Vote. One vote per eligible voter, please.

Don’t anyone assume that their side has this sewn up and there is no need for their vote.

If I can be all pie-in-the-sky about it, please make an informed vote. Check out what the candidates say about issues that matter to you and make an informed decision.

Alright folks, Biden’s now up to 72%

…it’s going to be a long 4 months.

As a counterpoint, Nate had Obama around 60-70% around this point in 2012 (can’t find a time series still up but a couple articles from July 2012 are still up with those numbers). It converged towards 90% as time passed and left fewer room for surprises. The signals that 2016 would stay uncertain (e.g. the number of undecided/third party voters) were being reported as unusual by Nate in August 2016. In terms of that raw uncertainty the polls look more like 2012 than 2016 right now.

Clippy had a previous engagement.