So how can those votes be anywhere near accurate if no-one knows what the postal votes say? Surely to have any predictive power they would need to ask a sample of postal voters how they actually voted. In that case you have an effective “exit poll” being broadcast whilst the voting is still in play.
Nobody is reporting votes. Some states are reporting the numbers of mail ballots returned based on the party registration of the voter. Not all states have party registration, and a large chunk of the electorate in states that do have it choose not register with any party. But people who are registered with a given party are likely to vote for that party’s nominee. How likely they are can be somewhat predictable based on demographics, but that is of course just a statistical model, not a record of votes.
ETA: Some of this data also comes from exit polling at early voting locations.
They aren’t. They should be ignored. They are clinging to false hope. They only remotely useful one is in Nevada, and that’s because there is a particular expert there that is very good at sussing out the numbers.
The only practical use they have is for campaign staffers to know where outstanding ballots are to attempt to get them returned (if Miami-Dade county in Florida has a large percentage of unreturned Dem ballots, they may pump resources there to phone bank or text message voters to get them to drop them off).
I take that back, there is one other potential use for pundits. They perhaps give a sense of the overall size of the turnout. It seems likely that this year will have much higher turnout than any election since 2008, and possibly since 1980 (percentage-wise). Typically, high turnout helps Democrats, but we shall see.
around 11 million citizens have turned 18 since the last presidential election, and I’m assuming their turnout will be 30-50%
but there are still about 100 million Americans who didn’t vote in 2016 who probably won’t vote in 2020 either.
This is important to not just pundits. The more people who are exercising their political power by voting, the more healthy it makes our democracy. And it’s important to report it because it’s self-reinforcing. News of high turnout encourages more turnout from those who haven’t voted yet.
But we shouldn’t read anything into actual results from the high turnout so far, and I’d rather people wouldn’t draw any conclusions from it. It’s a repeat of the misunderstanding of what polls and poll predictions actually mean.
Does anyone have numbers about how many Americans age 65+ died since the 2016 election?
Are you saying that someone, somewhere knows that John A, John B and John C have definitely returned postal ballots and, though no-one has opened them up and checked, because there is an official record of John A, John B and John C being rep, rep, dem respectively, there is a reporting of what the anticipated split would be.
Apologies if I’m being thick on this but both the concept of voter registration and reporting on votes cast (even semi-speculatively like you suggest) is completely alien to me.
Yes, basically. But the reporting I’ve seen isn’t what the anticipated votes would be, but simply (in your example) 2/3rds of of the ballots returned are from registered Republicans.
And the “someone, somewhere” is the state elections office, which knows which people’s ballots have been returned and also knows which party their registered with. As noted, however, many states don’t have a mechanism to register by party.
OK, to be honest that doesn’t sound any better. Is there not a risk of skewing the results or the turnout by reporting even that potentially inexact figure?
i.e. dems and rep both see a high turnout for one side in those post-in figures and one-side thinks it is a done-deal and stays home, the other gets spooked and floods to the ballot-box. I don’t see how it doesn’t change behaviour.
I don’t think too many people are even paying attention to these figures. The headlines are focusing on total number of early voters and polls. The break down by party is available, but not highlighted.
And, at least on the Democratic side this year, no one is complacent or taking anything for granted.
I guess there is. But I’d wager it’s pretty small. I don’t think there is much evidence that perception of which candidate is winning or losing changes turnout that much.
A much bigger factor, for example, is west coast states having polls open while the eastern states are reporting results. For example, if Florida gets called for Biden early on election night, I could see that reducing turnout in California (for both sides, honestly).
We have lots of factors that effect turnout, and I don’t think that reporting early voting numbers is really one of them. It’s just something that wonks like to look at to either get excited or get scared. For example, right-wing boards are all over reduced black early vote turnout in North Carolina as a sign that Biden is doomed.
Even if I hadn’t already voted, such news out of Florida would not discourage me from voting for Biden. I want this to not just be a victory. I want a landslide. I want a blow-out. I don’t want Donny Two-Scoops defeated, I want him crushed. I’d walk on my knees through a mile of broken glass to accomplish that.
You are right, in that election data experts are saying we should not make any assumptions about the final vote based on early voting info. Both Nate Silver and Elliott Morris (of The Economist) have said it can be very misleading. It’s other election pundits who are trying to get meaningful info from the early vote. They should be ignored.
The year I hear mentioned is 1908, which is so long ago to essentially be meaningless. For all practical purposes this year will set the record.
I made this post a few weeks ago.
In the last 4 years roughly 15 million kids turned 18, while about 11 million Americans died. The vast majority of those Americans who died were silent generation.
So I’m guessing about ~10 million of that 11 million who died were senior citizens.
That the people who had voted were registered as Democratic or Republican; which is public record.
How the individual actually voted, however, is not public record, and is not known to anyone other than the individual (and whoever they tell, if they tell anybody, and don’t lie.) A person registered as Republican may vote for a Democrat, and vice versa.
In addition, many people aren’t registered as either.
In practice, people registered as Republican are a whole lot more likely to vote R than people registered as Democratic are, so party registration does provide some indication on average. But they’re not guaranteed to do so, and this is an election in which it’s possible some significant number of people may jump ship.
Yup.
It’s sometimes suggested that such information, as well as exit polling, should be forbidden before the polls close – or even that announcing definite results in early-closing states should be forbidden until the polls have closed in all states. I suspect there’d be immediate lawsuits on the grounds that it would be an unconstitutional restriction on free speech, though.
But I think the risk is less this year because of what happened in 2016. People on all sides are suspicious of the accuracy of predictions, no matter what they’re based on; and may this year also be suspicious even of official preliminary results, especially in states which accept mail-in votes that arrive after election day.
I seem to recall that there have even been instances in the past where the networks have called a state before voting was over in that state, based on most of the state being in one time zone but some small portion in another. It may have even been Florida, I believe the panhandle is Central time while the rest of the state is Eastern. The result was people waiting in line to vote simply went home.
Yeah, that has happened. Networks are much more careful about that now.
Video on calling states (ABC will not call unless 99.5% sure)
How many votes need to be counted to call Wyoming or Vermont?
Brian
A majority of them…you know, a few dozen. 
They can also poll the cattle in Wyoming and the dairy cows in Vermont. 