Indecision 2024

[Woodchopper]

I think she's one state away from winning the EC as it stands.

Trump beat his polling in both 2016 and 2020.

Show how badly poor Joe was losing.

This year, I'd expect Harris to easily beat her polling.

Pollsters will still not deal with the reality that their samples are highly biased towards people unable to use call screening, and that their likely voter screens ignore the extremely democratic bias of new voters.

They deal with that problem by not calling people and using internet based panels instead. Its possible to compare polls using both methods.

Though using panels isn't a panacea, as the company needs a representative sample of the voting population, and its difficult to predict exactly how different demographics will turn out.

[Woodchopper]

I don't disagree.

But the 2020 polls in Pennsylvania had Biden +3 and he won by +1.

538/ABC for Pennsylvania currently has Harris +1.7.

Can never underestimate the stupidity of American voters.

However, the polling companies are aware of this and are very likely trying to correct the error. Maybe they'll get it right and the polling id accurate. That's possible as the polling for the 2022 mid-terms was more accurate than in 2020.

Or maybe they've overcorrected and are overestimating support for Harris, or perhaps there is another source of bias that we'll only find out about after the election.

[Woodchopper]

One other thing is that the general polls show Harris doing better than Biden with both college and non-college whites, but doing worse with young people and people of color.

But polls specifically of young people and of people of color show Harris doing much better than Biden with those groups.

Those two groups are hard to poll well with general polling.

That's interesting, but people should be careful with that kind of analysis. A survey with circa 1 000 participants should have a margin of error of about +/- 3% (assuming its a representative sample, and we know there are problems there). But a sub-group (eg people aged 18-30) will have far fewer members, perhaps 1-200, and so the margin of error will be all over the place.

[Woodchopper]

To sum up, all the polls can tell us now is that the election is too close to call.

Of course its interesting to debate changes in the polling numbers. But right now the changes are so small they mean very little.

Even the most professionally conducted survey will have a margin of error of +/-3%. If a survey states that Harris has 48% and Trump has 46%, the we should assume that it tells us that Harris has between 45-51% in general population, and Trump 43-49%. This of course ignores systematic bias, but we can't be certain what that is.

Currently, the most recent polls have nine states with less than 5 percentage points difference between Harris and Trump: FL, GA, NC, PA, MI, WI, IA, AZ, NV.

Collectively, they're enough to give either candidate a landslide and such a result is within the margin of error. I doubt that would happen, the point is just that a narrow victory for one of the other is well within what is feasible.

[dyqik]

One other thing is that the general polls show Harris doing better than Biden with both college and non-college whites, but doing worse with young people and people of color.

But polls specifically of young people and of people of color show Harris doing much better than Biden with those groups.

Those two groups are hard to poll well with general polling.

That's interesting, but people should be careful with that kind of analysis. A survey with circa 1 000 participants should have a margin of error of about +/- 3% (assuming its a representative sample, and we know there are problems there). But a sub-group (eg people aged 18-30) will have far fewer members, perhaps 1-200, and so the margin of error will be all over the place.

The important point around this is that e.g. the NY times polls showing Trump drawing level have 18-29 year olds going from +21 Harris in August to -8 Harris in September.

This is, of course, not a real change, but an illustration of the level of both statistical noise due to small samples in sub groups and sample bias noise in a very hard to poll group. And that's with n~2700, not the usual n~1000 for the full sample.

At the same time, polling of easier to sample subgroups, like over 55s, is more useful for understanding changes from the 2020 electorate, and changes within the race.

[Woodchopper]

One other thing is that the general polls show Harris doing better than Biden with both college and non-college whites, but doing worse with young people and people of color.

But polls specifically of young people and of people of color show Harris doing much better than Biden with those groups.

Those two groups are hard to poll well with general polling.

That's interesting, but people should be careful with that kind of analysis. A survey with circa 1 000 participants should have a margin of error of about +/- 3% (assuming its a representative sample, and we know there are problems there). But a sub-group (eg people aged 18-30) will have far fewer members, perhaps 1-200, and so the margin of error will be all over the place.

The important point around this is that e.g. the NY times polls showing Trump drawing level have 18-29 year olds going from +21 Harris in August to -8 Harris in September.

This is, of course, not a real change, but an illustration of the level of both statistical noise due to small samples in sub groups and sample bias noise in a very hard to poll group. And that's with n~2700, not the usual n~1000 for the full sample.

Yes, certainly, a swing like that in a sub-group is likely just to be noise. Journalists may feel that they have to write about it but they're just doing a version of this.

If something similar were to be found across many polls it might be worth looking at further.

[dyqik]

Also, there are several right-aligned pollsters delivering many biased polls right now, to try and drive the averages and narrative to Trump having been cheated out of a win. This includes Rasmussen (no longer any relation to the pollster Scott Rasmussen), who admit having colluded with Trump's campaign.

[lpm]

It's fundamentally unknowable.

Unknown factors that I worry about

1. Racist/sexist Democrats who will be the shy Tories of the election - not voting, or voting Trump in secret

2. Gaza/Israel - Democrats/young voters thinking they're "sending a message" by not voting

[Woodchopper]

It's fundamentally unknowable.

Unknown factors that I worry about

1. Racist/sexist Democrats who will be the shy Tories of the election - not voting, or voting Trump in secret

2. Gaza/Israel - Democrats/young voters thinking they're "sending a message" by not voting

Gaza may not affect things very much. According to the Harvard youth poll very few young people in the US see it as the most important issue. They’re far more concerned about domestic problems like inflation, jobs, healthcare etc.

See here, point 4 https://iop.harvard.edu/youth-poll/47th ... pring-2024

Also, among those whose vote will be swayed by Gaza, how many are in swing states. If they’re in California or New York they won’t make a difference to how the state votes overall.

Sexism and racism are another matter.

[El Pollo Diablo]

Polls from the last week show Harris ahead by an average of 4.1 points, the week before an average of 3.3 points. US polls are a little harder to decipher because polling companies will occasionally issue six polls at the same time, but that's focusing on likely voters rather than registered or adults.

There's also been a recent surge in the US public wanting Democrats in congress and Harris now has a net positive approval rating for the first time since the early Biden presidency. Trump remains down on nearly net ten points disapproval.

So, the recent signs nationally are good. But (a) everything can change, (b) polls are inaccurate, (c) the issue is popularity in individual states and (d) the US has the stupidest presidential election system in the world - no one else does it like they do.

The key state remains Pennsylvania. Still on a knife edge, still mystifying.

[TopBadger]

In the US on business, was eating at the bar last night and conversation turned to politics...

I said I didnt understand the support that Trump gets from Christians given his convictions, sexual assaults. This guy says he'd never heard it put like that, but he didn't care about that, he could forgive Trump those sins, so long as he was against transexuals who are a distortion of nature... and did I know that they were operating on babies to alter their genitals... also, f.ck all these migrants who keep coming and shouldn't be here.

Later turned out he's a catholic 2nd Gen mexican immigrant. But you know, sinning is ok, f.ck tolerance and immigration bad... mkay.

[shpalman]

and did I know that they were operating on babies to alter their genitals...

Cutting off their foreskins?

[TopBadger]

and did I know that they were operating on babies to alter their genitals...

Cutting off their foreskins?

Nope, he meant gender reassignment on kids.

[Woodchopper]

and did I know that they were operating on babies to alter their genitals...

Cutting off their foreskins?

Nope, he meant gender reassignment on kids.

Yes, one consequence of the enshittification of Twitter has been an inkling into the volume of reactionary posts asserting that it’s widespread that children are being pressured or forced to have gender reassignment surgery.

[Fishnut]

... and did I know that they were operating on babies to alter their genitals...

I take it he's against circumcision then...

[hakwright]

Polls from the last week show Harris ahead by an average of 4.1 points, the week before an average of 3.3 points. US polls are a little harder to decipher because polling companies will occasionally issue six polls at the same time, but that's focusing on likely voters rather than registered or adults.

There's also been a recent surge in the US public wanting Democrats in congress and Harris now has a net positive approval rating for the first time since the early Biden presidency. Trump remains down on nearly net ten points disapproval.

So, the recent signs nationally are good. But (a) everything can change, (b) polls are inaccurate, (c) the issue is popularity in individual states and (d) the US has the stupidest presidential election system in the world - no one else does it like they do.

The five thirty eight polling averages show Harris has had a consistent lead on Trump by between 2.5 and 3% for the last month. It's nudging up and down by a fraction of a percent here and there, but not really moving much. I think they do a decent job of combining polls, taking into account known polling biases and quality issues.

But it's electoral college votes that count. Their forecast for who actually wins has shown it to be more or less a toss-up since August. Harris maintains a slight edge, but it's been between 50-50 and 60-40 all that time. I doubt it will change much.

Because of the voting system, really tiny shifts at local level could have significant effects overall, so lots of things are on a knife-edge. But it will likely be a coin toss as to who wins the presidency (though Harris is much more likely to win the popular vote, but who cares about that?!?)

[lpm]

Because of the voting system, really tiny shifts at local level could have significant effects overall

I got annoyed at Twitter stupidity celebrating that there are 20,000 Haitian-Americans in Pennsylvania and that could make all the difference.

There are 150,000 Muslim Americans in Pennsylvania who get distressed every time Biden backs the latest Israeli event. I worry this could be where the election is lost.

But who the hell knows.

[El Pollo Diablo]

Pennsylvania looks to have moved towards Trump in the last few days. Who knows if Melania's anti-Trump intervention will change things.

[headshot]

Because of the voting system, really tiny shifts at local level could have significant effects overall

I got annoyed at Twitter stupidity celebrating that there are 20,000 Haitian-Americans in Pennsylvania and that could make all the difference.

There are 150,000 Muslim Americans in Pennsylvania who get distressed every time Biden backs the latest Israeli event. I worry this could be where the election is lost.

But who the hell knows.

It won’t lead to more votes for Trump because he’s even more supportive of Israel than Harris.

It might just mean Harris loses some of the Muslim vote.

[El Pollo Diablo]

It won’t lead to more votes for Trump because he’s even more supportive of Israel than Harris.

The people who make decisions like this are barely clinging on to reality, things like this won't put them off voting for Trump.

[dyqik]

Because of the voting system, really tiny shifts at local level could have significant effects overall

I got annoyed at Twitter stupidity celebrating that there are 20,000 Haitian-Americans in Pennsylvania and that could make all the difference.

There are 150,000 Muslim Americans in Pennsylvania who get distressed every time Biden backs the latest Israeli event. I worry this could be where the election is lost.

But who the hell knows.

There are also 800,000 Polish Americans in Pennsylvania who get distressed every time Trump or Vance push back on supporting Ukraine against Russia. A significant number of these are also Jewish.

[lpm]

There are also 6,500,000 car driving voters in Pennsylvania who seem prepared to throw away democracy due to the price of gas.

[dyqik]

There are also 6,500,000 car driving voters in Pennsylvania who seem prepared to throw away democracy due to the price of gas.

Gas prices have been dropping in the US. Under $2.75 a gallon last week for me. Obviously this week it's back up a bit, but the US is self-sufficient in gas at the moment.

[lpm]

There are also 6,500,000 car driving voters in Pennsylvania who seem prepared to throw away democracy due to the price of gas.

Gas prices have been dropping in the US. Under $2.75 a gallon last week for me. Obviously this week it's back up a bit, but the US is self-sufficient in gas at the moment.

I've heard gas prices are $30 a gallon under Biden, compared to 10 cents when Trump was in power. They seem pretty confident about the facts.

Reality has no place in Trump's America.

[headshot]

Vance just outright said that Trump won the 2020 election…

https://www.independent.co.uk/news/worl ... 23517.html

“Who won the 2020 election? Could you just answer? Did Donald Trump win?” Selvig asked.

“Yes,” Vance replied.

After asking Vance to confirm that he was saying Trump won in 2020, Vance replied “Yep.”

The comedian then pressed Vance, asking if he would concede if he and Trump lose in the 2020 election.

Vance did not answer directly, instead just saying “I really feel bad for you man.”

Selvig continued to push, asking “if your opponent gets more votes, will you concede,” after which Vance walked away without answering.