Polling: Unskewed Polls and comments on polling (moved from Harris Thread)

From Robert Hubbell’s Today’s Edition Newsletter on Substack:

The media narrative abruptly shifted post-Labor Day weekend to proclaim that Pennsylvania will decide the 2024 election. Don’t believe it. That narrative is wrong on so many levels that it is insulting to the intelligence of American voters.

The narrative that “it will all come down to Pennsylvania” is wrong and defeatist. It presumes that the state electoral map is frozen in amber and that only one or two states are flippable. Don’t believe it! Kamala Harris and Tim Walz are contesting states that are presumed to be frozen in amber on red vs. blue maps on cable news.

It is a mistake to sell the American people short. The 2024 election will not be decided by Pennsylvania — although the votes of all Pennsylvanians will matter just like the votes of every other American. The election will be decided by hundreds of millions of Americans taking democracy seriously by voting in tens of thousands of elections at a moment in history when one party wants to deny women full citizenship and personhood, deny Black Americans the right to vote, deny LGBTQ Americans their dignity and equality, deny children safe schools, deny all Americans a future free of man-made climate catastrophes, deny workers of a living wage, and deny the peaceful transfer of power every four years.

My money is on the American people to vote on issues that are important to them—not on a media-concocted horse race designed to sell soap.

Gives Nate Silver a little what-for, as well.

That link sounds like an evangelical preacher telling his flock that Jesus is lord and all true believing Americans know it . So how can the devil win, when all of us true believers know (and vote) the right way?

The link says that true-believing women will turn out in droves to vote against the party that will take away their freedom of abortion, that true-believing minorities will turn out in droves to vote against the party that will take away their right to vote, that true-believing Americans will vote against the party that denies climate change, etc.

His logic is undeniable, and leads him to declare, with no evidence:
“My money is on the American people to vote on issues that are important to them”
It’s a nice exhortation of faith.

But he does not give

in fact, the link doesn’t even offer a single hard fact to counter Silver.

Nate Silver, on the other hand, deals in facts.

He then tries to change the facts-- by “weighting” the statistics. with certain assumption. These assumptions may not be accurate.
But it is closer to reality than evangelical preaching.

It’s going to be a very close election.

“People had opinions, probably, and neckties.”
— E. E. Cummings

No they don’t. Not from strangers, anyway.

I really wonder how pollsters get their data these days, and suspect there’s an enormous “didn’t answer” bias. I just don’t know which way that bias leans.

It really is the big issue. You can’t know much about those who don’t answer. Even why they don’t answer. There is a very strong possibility that polls are systematically skewed one way or the other … and no one knows which way if so.

I wish pollsters had a way for you to validate them before doing the poll to establish trust. I’m not answering any questions from someone who calls or texts out of the blue no matter who they say they are. I figure it’s a scammer or marketer who is really trying to gather as much personal info as they can for scamming or marketing purposes under the guise of political polling. The validation would have to be a way I could trust. I’m not just going to take their name at face value. Even if they say they’re from Gallup News, I’m not going to just accept their word. I’d want something like they give me a code and then I go to the Gallup News website to open a chat window with that code. I’d be much more likely to answer polls that way rather than getting a random call or text that asks about my income range, zip code, political views, etc.

2016 was really an aberration; the polls were messed up for a couple of reasons that don’t exist in 2024. Many of those polled refused to admit that they were voting for Trump. That is not the same today when the MAGA crowd is proud to be Trumanistas. Secondly, many of the polls looked at popular vote assuming the the electoral votes would follow. Meanwhile Clinton campaigned hard to win the popular vote while Trump ran hard for 270 electoral votes.

68 years earlier was the famous “Dewey beats Truman” polling but IIRC that was a failure in polling methodology and not the poll-model. But because of those two elections it is now trendy to question what the polls imply although with modern techniques in psephology I believe that the polls are almost always correct.

This really is just a factually incorrect post.

First off, in terms of predicting vote totals, the polls were more off the mark in 2020 than in 2016.

Second the expert consensus is that there was no “shy Trumper” effect. There were voters who did not answer polls but not many who answered but denied their support. (Trying to correct for that by weighting GOP identified respondents more failed so maybe it was specifically Trump supporters who declined to participate, unclear).

Third, Team Clinton was very aware of the EC. They invested resources based on that awareness based on where polling told them was close and didn’t “waste” resources where the lead was solid. One state that polling said they had a comfortable lead, that was unlikely to be a tipping point, was WI. She was consistently around five points up there. But she lost.

FWIW they tried to fix that miss in 2020. Result? A miss of nine points instead. Oy.

I do NOT believe that is NECESSARILY a Trumpward skew this time. Maybe the attempts to reweight work this time. Maybe they overshoot. But a state or two or more being significantly off one way the other is to be expected.

Exactly. I keep explaining this to people.

The difference between 68 and 75 is seven. The difference between 49 and 52 is three. The second instance (for many situations) is more consequential – but it’s smaller than the first.

I’m pretty sure we are back to this failure. We just don’t have any way to poll a representative sample of people.

Agree w everything you said. To which I’d add:

I would be vastly more willing to participate in a poll if I was confident that would not unleash a torrent of future poll requests now that they know they have a live one.

I feel even more strongly that way about fund-raising. I’d send the legal max to a bunch of different causes if I was certain it wouldn’t trigger an endless parade of further pleas for money. In fact under the current reality, any donation doesn’t prevent more solicitations, it guarantees them. Ouch!

Amen to this. I sent a donation to ActBlue immediately after Biden endorsed Harris. Since then, my inbox has been getting slammed with requests for more and more and more.

But, hey, now I’m on a first-name basis with a whole slew of Democratic politicians.

I have participated in polls, and it has not unleashed a torrent of future poll requests. I think the selection is somewhat random.

(I’ve also given money and been inundated with additional requests for money. It feels completely different.)

So here is a gift link to the NYT article about the new Times/Siena Poll findings about the race.

Long story short, there is a large group who say they can’t support her because they don’t have enough info about Harris and what her policies will be.

The poll is basically tied, which I continue to find baffling. But it appears Harris has an opportunity at the debate this week to fill in some of those folks.

This poll continues to increase my anxiety about where this election will go.

The younger folks are getting pay increases to keep up with inflation.

I called that a few months ago- mind you, it is still in the realm of optimism, but not crazy talk anymore.

He sold off his company and his staff that did research. So, I no longer think he is the “go to” guy.

I do not find it baffling. The NYT is not gonna support Harris until she gives them a very special exclusive interview. I dont trust the NYT any more.

As much as I share your deep concern about the NYT coverage of Biden and Harris what I find baffling with this poll is the fact this race is basically tied.

That is something I’m seeing in pretty much ALL the polling.

DJT should be getting buried and he isn’t. It can’t all be laid at the door of the Sulzberger cabal.

Harris -overall- ahead by 3%.

Penn- Harris ahead.

Mich- Harris ahead.

Wisconsin- Harris ahead

Florida- Harris under by only a few points. If Fla goes Dem- it is over.

Yes, the race is close, but Harris is ahead. Not by enough to be optimistic, but no reason to be pessimistic either.

I appreciate your enthusiasm but most, if not all, of those are within the margin of error.

So that means tied. I find it baffling.

ETA: Yes, if FL goes D it is over. I’m not holding my breath.

Yeah, for any one poll. But when all the poll accumulators show Harris ahead, that means Harris is ahead. Not enough to be complacent, of course.

Those polls measure popular votes.

The Electoral College vote is likely to play out differently.

Dems have a talent regarding winning the popular vote but losing the EV.

Thus the polls are not enough to calm my nerves.