According to Wikipedia here are the approximate percentages of voting-eligible population that have participated in Presidential elections since Regan:
80: 54% (Regan/Carter/Anderson)
84: 55% (Regan/Mondale)
88: 52% (Bush I/Dukakis)
92: 58% (Clinton/Bush I/Perot)
96: 51% (Clinton/Dole/Perot)
00: 54% (Bush II/Gore)
04: 60% (Bush II/Kerry)
08: 63% (Obama/McCain)
12: 58% (Obama/Romney)
16: 59% (Trump/Clinton)
20: 67% (Biden/Trump)
I’m going to guess that turnout this time might be in the low-mid 60’s on this metric. Some of the changes that made it extra-easy to vote in many jurisdictions during the pandemic have lapsed, and some voters have short memories, taking the edge off the perception that the question of Trump/not-Trump is a rather existential one.
Also, election officials in some places have been more aggressive at playing games with the voter rolls…but I’m not certain how material an impact that would be on voter turnout rates.
The 2020 turnout was high for modern times. The first election I watched was Kennedy/Nixon in 1960 which had a shade under 64% voter turnout. There was an incredible buzz around JFK which might have been a factor in the good turnout.
It is disappointing to me that in some democratic countries over one-third of the population doesn’t bother voting. Canadian Federal Election turnouts are currently similar to US but for most of my life Canadian numbers were in the 70%+ range, nudging 80% in several elections. Folks here don’t care as much as they used to.
I’ve gotten some pretty bad mailers for local elections. Not mugshots, but just some bad stuff.
The town I live in has a mostly black population, with a small Jewish neighborhood at the edge of the town. A few years back, our incumbent mayor was a white man, running for election against a black woman. I received a mailer from her, with a picture of a black family, and the tagline “Vote for Someone Who Looks Like You!”
She ended up being elected as mayor. The night she was elected, she got caught “celebrating” with her boyfriend in a parked car on a local road. The police officer decided not to make an issue of it because she was just elected, and didn’t want to get the PD off on the wrong foot with her, but he wrote an internal memo about it to the chief. A year or so later, she got pulled over for speeding, running a red light, and using her cell phone, and threw a fit claiming the PD was out to get her - and tried to fire the chief. The memo was leaked to the press, along with the report.
What a dysfunctional town I live in…
If we’re doing anecdotes here’s one from a couple of days ago. I have a friend (I’ll call him Joe to make this easier to read) that runs a conservative PAC. Joe voted Trump 2016 I believe. After a year or two of President Trump Joe became a never Trumper in private. In public he still runs the conservative PAC. Joe informed me a friend of his confessed to voting for Harris on his mail in ballot. His friend is currently helping run the campaign for a currently sitting republican governor (not currently up for election). Joe says his friend is one of the most conservative guys he knows with decades of experience running conservative campaigns. The reason he gave Joe for voting Harris, probably the first D he’s ever voted for, “I had to vote for Democracy.”
It’s just one vote but I hope enough conservatives can make that same choice.
The “Silent Majority”?
One can only hope.
I mean, I assume GOP folks like to vote, right? Trump promised “you’ll never have to vote again.” They DO know what that means, right? Maybe they do. Maybe… they do.
The problem is that working class people are attracted to trump like flies are attracted to a lightbulb. Harris is polling (relatively) poorly with black men, and it’s basically a disaster with white men. Look at these proposals:
This is not serious. Trump is talking about putting tariffs on chinese stuff to save jobs in Michigan. What works best? Harris is proposing that black men sell weed. Again, this is not serious. Let’s be honest.
You can’t expect these % to ever be that high because the election is a done deal for many places where one party dominates the other. I would like to see the % in swing states, actually.
Trump: (sighs) I just haven’t gotten over (pause) something the justice department did yesterday where (pause) Virginia cleaned up its voter rolls and got rid of thousands and thousands of bad votes and the justice department sued them that they should be allowed to put those bad votes, those illegal votes back in and let the people vote. So I haven’t, I haven’t gotten, I haven’t gotten over it. A lot of people have seen that. A lot of people can’t believe it.
Reporter: The question is about Google, President Trump.
Editorial notes: It wasn’t yesterday, but was recent. The issue is that purging voter rolls within 90 days of the election, which Virginia did, is prohibited under Federal law.
How can polls be accurate within +/-3 points when turnout from one election to the next changes by 8%?
How much did COVID and mail in ballots increase turnout? My state did not really accomodate this and turnout was up 3-4%
You could argue there should be some excitement to vote for Harris, who is a much better candidate than Clinton, as the first female POTUS. There is also fear of another Trump presidency, which drove Biden votes, who was a pretty generic candidate. Could Harris get the double boost in turnout? Who is actually excited by Trump these days?
Sadly, there are people who are excited by Trump, and lots of them. I refuse to vote for Harris, but I also don’t like Trump. I’ll probably end up writing in Nikki Haley.
They aren’t. In statistical learning, we often write the total error of a model as the sum of the reducible error that comes from your model plus the irreducible error that comes form the intrinsic randomness of your data, with the goal being to make models whose reducible error is smaller and thus overall error is less. In polling, the fact that it is a random sample is the irreducible error, and that is what is reported for the margin of error because it can be easily computed. The size of the reducible error much impossible to compute and comes from the flaws in the assumptions of the models.
Set the +/- term to the prespecified error margin which is generally .03. The z comes from 95% probability that the interval contains the true proportion which is 1.96. To achieve the desired level of confidence for all cases use .5 which maximizes the required sample size. Solve for n and round up which gives 1068.
Right - that is my understanding, and most polls target n~1000 to get to +/-3%. But that assumes they are 100% accurate on their determiniation of likely voters, which is usually based on some question like “did you vote in each of the last 2 elections” if you were old enough to vote or “do you plan to definitely vote.”
But when turnout can vary from 55% to 68% in a couple of cycles, and the rates of turnout between two candidates is somewhat independant, that margin of error becomes meaningless.