Male Pollsters Shocked — Shocked!! — When a Woman Pollster Discovers Women Voters
On Friday, Nate Silver tweeted and substacked a piece declaring that there must be herding going on in the casino.
His substack analysis didn’t acknowledge, however, that this herding was going on in his casino.
Now granted, our forecast is close too. But it’s based on polling averages: dozens of polls have been released in each of these states over the past month. That greatly increases the sample size. Collectively, they’ve surveyed about 230,000 voters.
By contrast, the median sample size in individual polls in these states is 800 voters. In a 49-49 race in a poll of 800 people — assuming 2 percent goes to third parties — the theoretical margin of error for the difference between Trump and Harris is ±6 points. If that sounds higher than you’re expecting, that’s because the margin of error that’s usually reported in polls is only for one candidate’s vote share. For instance, in a poll of 800 people, Trump’s margin of error is about ±3 points, as is Harris’s. However, basically every vote that isn’t a vote for Trump is a vote for Harris. If Trump gets 52 percent of the vote instead of 49, that implies Harris will receive 46 percent.1 So the margin of error on the difference separating Trump and Harris is ±6.
What this means is that if pollsters are doing honest work, we should see a lot more “outliers” than we do — even if people love to complain about them on Twitter.
In our database as of this afternoon’s model run, there were 249 polls in the seven battleground states that met Silver Bulletin standards and did at least some of their fieldwork in October.2 How many of them showed the race in either direction within 2.5 percentage points3, close enough that you could basically call it a tie?
Well, 193 of them did, or 78 percent. That’s way more than you should get in theory — even if the candidates are actually exactly tied in all seven states, which they almost certainly aren’t.
The reason we’re seeing this herding is because Nate Silver has spent 16 years training pollsters to herd. It probably makes things worse that polling has become far more difficult, far more expensive, and far more important in shitty campaign coverage (not least because of Nate Silver). The herding is happening because, thanks to the early but not more recent success of Nate, political operatives know they can create a reality in poll averages.
Well, ask and you shall receive, Nate, because Ann Selzer doesn’t herd. Last night, the Iowa pollster surprised everyone with a poll showing Kamala Harris ahead in Iowa.
A new Des Moines Register/Mediacom Iowa Poll shows Vice President Harris leading former President Trump 47% to 44% among likely voters just days before a high-stakes election that appears deadlocked in key battleground states.
The results follow a September Iowa Poll that showed Trump with a 4-point lead over Harris and a June Iowa Poll showing him with an 18-point lead over Democratic President Joe Biden, who was the presumed Democratic nominee at the time.
“It’s hard for anybody to say they saw this coming,” said pollster J. Ann Selzer, president of Selzer & Co. “She has clearly leaped into a leading position.”
As Selzer describes, this effect is driven by older women voters.
The poll shows that women — particularly those who are older or who are politically independent — are driving the late shift toward Harris.
“Age and gender are the two most dynamic factors that are explaining these numbers,” Selzer said.
Independent voters, who had consistently supported Trump in the leadup to this election, now break for Harris. That’s driven by the strength of independent women, who back Harris by a 28-point margin, while independent men support Trump, but by a smaller margin.
Similarly, senior voters who are 65 and older favor Harris. But senior women support her by a more than 2-to-1 margin, 63% to 28%, while senior men favor her by just 2 percentage points, 47% to 45%.
Nate promptly set to work describing how brave it was for Selzer to release this because a gambling market Nate has helped to create says she’s wrong.
Releasing this poll took an incredible amount of guts because — let me state this as carefully as I can — if you had to play the odds, this time Selzer will probably be wrong. Harris’s chances of winning Iowa nearly doubled in our model from 9 percent to 17 percent tonight, which isn’t nothing. Polymarket shows a similar trend, moving from 6 percent to 18 percent after the survey. But that still places Harris’s odds at around 5:1 against.
While I don’t pay for Nate Silver’s gambling market, he appears to have gone on to argue that what Selzer is measuring is inflation, not just — as Selzer explained — big movement from women.
Meanwhile, that Other Nate, Nate Cohn, released this column, in what former pollster Adam Carlson described as “rough timing.”
Other Nate then went on to point to things that followed the pandemic but which are either now in the past or which polls always reflected a misunderstanding on the part of voters.
[T]he events that followed the pandemic took a serious toll on the case for liberalism, whatever the precise merits of the arguments. Inflation and high interest rates could be blamed on high government spending stimulating excessive demand. High gas prices could be blamed on suspending drilling permits and the termination of the Keystone pipeline project. A surge of migrants could be blamed on the administration’s looser border policy, which became politically untenable; homelessness, crime and disorder made the case for “law and order.”
From there, Other Nate made some false claims, such as that Harris — whose plan to include home health care in Medicare has been ignored by the press, and who mentions plans to increase the child tax credit and mortgage assistance in every stump speech — had moved away from populism.
The traditional Democratic program to expand the social safety net has played a diminished role in the campaign. In her speech at the Democratic convention, Ms. Harris didn’t mention the unfinished business of Obama-Biden era liberalism: paid family leave, an expanded child tax credit, student loan forgiveness, universal pre-K or free community college.
And Other Nate claimed that polls have shown that Trump does better on the issue that is most important to voters votes.
If there’s any poll question that captures the swing toward a more conservative environment, it’s the question of which party (or candidate) would do the best on the issue that’s most important to your vote. All cycle, polls have shown Republicans and Mr. Trump with an advantage on this measure.
Other Nate didn’t mention that on the one that matters to most people, the economy, Harris has caught up to Trump.
Crazier still, Other Nate didn’t mention that for those who say abortion is the most important, Harris always wins. He simply disappeared abortion in his claim that voters think Trump would do better on all issues that are most important to them.
The other day, I hoped that someone with interns would review the Other Nate’s statements about this race — but hell, do it for First Nate, too — to see whether his mentions of polling adjustments adjust for the Dobbs effect, the recent trend in which, with one exception, polls have always underestimated and often wildly underestimated pro-abortion outcomes.
Other Nate has spent a great deal of time talking about how polls missed Trump in 2016 and 2020. On October 6, Other Nate described how whether pollsters are using voters’ recalled votes in 2020 explains a split in the polls (in none of Other Nate’s discussions of recalled vote have I seen any discussion of whether pollsters are accommodating for the documented greater mortality rates among Trump supporters since 2020 arising from vaccine skepticism). On October 22, Other Nate described the theories for why pollsters missed Trump in 2020. The next day, Other Nate described all the adjustments polls have made to remedy their 2020 Trump miss.
This trend in ignoring the Dobbs effect has continued more recently. Other Nate didn’t mention abortion or Dobbs in this column asking whether we can trust polls.
Other Nate didn’t field any questions that addressed abortion in this column about early voting (he did suggest that pollsters expect the normal amount of women in the electorate, 53%).
This column is particularly remarkable. In a column conceding that Harris may win because her coalition draws on more reliable voters than Trump’s, Other Nate didn’t mention that women are more reliable voters. Other Nate focuses on education, but only speaks of gender when describing that Trump’s bros are less likely to turn out.
But this election seems different. As we’ve reported all cycle, Democrats excel among high-turnout voters, while Donald J. Trump is strong among relatively low-turnout voters. He’s made his biggest gains among low-turnout demographic groups like young men and nonwhite voters.
Other Nate linked to an Anna Greenberg tweet noting that Trump was relying on less reliable voters. He didn’t mention the Greenberg tweet posted five minutes later in the very same thread decrying the focus on young men while ignoring how pissed women are.
Greenberg is among the thousands of people who have RTed this tweet from me.
You get my point.
This entire election has been disproportionately viewed through the lens of polling, polling that even the Two Nates confess exhibit obvious problems.
And best as we can tell, pollsters have contorted their polls to ensure they don’t miss Trump voters (again, with no discussion of how you account for higher COVID-associated death rates among Trump supporters since 2020). But they have (apparently) done absolutely nothing to ensure they don’t miss pissed off women.
And then Ann Selzer came along.
Differential mortality was likely only outcome determinative in Arizona AG in 2022 where the margin of victory for the Dem <300 was likely attributed to excess mortality among GOP regular voters. Excess differential mortality Might be worth 0.1 points
Thanks for that.
Isn’t the question somewhat different though?
This would be a problem primarily for recalled vote polls. In those, say they end up with a sample finding 360 recalled Trump voters and 440 recalled Biden voters. They will weight each of those 360 Trump supporters somewhat more to get up to 375 recalled Trump voters, making each of their views weigh more. But they SHOULD be only bumping up to — I don’t know the number, but it’s slightly less than 375 — to count for the Trump voters whose conspiracy theorizing has killed them off.
It just makes the possible distortion somewhat greater.
I’d been wondering about exactly this sort of error.
I think it could be even broader. Because of how strong the Harris/Trump split is on education level and the significantly higher rates of mortality for those with less education. Similarly so on gender too.
Basically from 4 years ago the most positive Trump demographic — older less educated white men — would have seen higher substantially higher mortality rate than women of similar education level, better educated men, and especially better educated women. All demographics that favor Harris a lot.
If the gender and education splits weren’t so stark, this wouldn’t be that big a deal. But the splits are pretty stark in virtually every poll every region.
Add to that, “buyers’ remorse” for those who took anti-vax covid advice and suffered horribly for it. That’s a pretty severe personal insult.
Not a large demographic, perhaps, but their loved ones might also share their umbrage — and small margins are still margins.
Since you bring up health, ap reports “ Robert F. Kennedy Jr. says Donald Trump would push to remove fluoride from drinking water”
The 833rd Base Commander failed to mention his obsession with removing fluoride from “ice cream, children’s ice cream, Mandrake.” To preserve their precious bodily fluids.
Next thing you know. Trump will want to give RFK Jr the launch codes.
Putting RFK Jr. in charge of health is like putting Ashli Babbitt in charge of an insurrection.
Putting RFK Jr in charge of anything, let alone the FDA, would be more like putting Gen. Jack D. Ripper in charge of nuclear non-proliferation talks with the Ruskies.
xyxyxyxy: this made me look up the number of dentists in swing states. I believe that most dentists are pro-fluoride (maybe not RFK JR’s though). No. Carolina has 5,400 dentists, Pennsylvania 7,700. For each dentist, there are probably at least 2 dental assistants. I wonder if there are enough dental professionals to outweigh the no. of fluoride-averse voters.
Fascinating. Leaves me wondering: does California have a dentist deficit, or Pennsylvania a surfeit?
30,000 dentists in California. 5,400 North Carolina (swing state) dentists.
The maximum number of right-leaning doctors, dentists, and nursing assistants that are swayable have probably already been swayed over the past 40 years.
[Welcome to emptywheel. Please choose and use a unique username with a minimum of 8 letters. We have adopted this minimum standard to support community security. Because your username is far too short it will be temporarily changed to match the date/time of your first known comment until you have a new compliant username. Thanks. /~Rayne]
Pretty sure both the pandemic and the rolling changes to reproductive health care since 2017 have affected health care provider’s attitudes more than the previous ~30 years.
Make America (fill in) Again
xyxyxyxy says:
November 4, 2024 at 3:19 am
Make America (fill in) Again
I would rephrase that as Making America Ill Again. Very Ill. RFK Jr wants to ban vaccination. Trump apparently is open to that.
Fuck.
Say hello to polio, rubella, zika, etc. This reminds me of Thatcher’s ‘Victorian Values’ shit.
No wonder Elmo wants to get off-planet: they’re just going to burn this one up until they run out of proles to cater to their every need.
‘Herding’ isn’t only a problem in opinion polling. The Nobel physicist R.P. Feynman pointed out that the same thing happened in measurements of physical constants where bias was created by throwing out measurements further from the previously accepted value (see his semi-autobiography Surely You’re Joking Mr. Feynman).
look forward to requesting that book from my library – unless it’s already theirs. thanks !
The Feynman book is hilarious. Feynman played bongos in a band while working on the Manhattan project and took side bets on whether the atomic bomb would destroy the atmosphere. think there was a sequel.
I think that Quake888 was referring to the Milliken oil drop experiment, which measured the charge on the electron.
Yes, that’s one example. But there are many others… the speed of light, etc.
I can’t recall where I saw it, but someone went back to Milliken’s notebooks and assessed why he had thrown out some results. His conclusion was that there were external reasons (weather was one I recall) for not accepting the excluded results.
But, in my view, the comparison is inexact, since — as Professor Wheeler’s post explains — pollsters make their ‘adjustments’ before their measurements rather than, purportedly like Milliken, after.
Anyhow, I think political polling is just witchcraft. Backcasting suggests I am right, too.
I was working at The Des Moines Register in 1980 when ‘the respected Iowa Poll’ completely blew it. Didn’t even get the sign right, still less the margins. But it wasn’t only the Iowa Poll that blew it.
In those days, Iowa had 6 congressional districts, 2 in the west, 2 central and 2 east. Politicians and observers were agreed that the western districts were safely Democrat and the eastern districts were safely Republican. The Republicans won the west and the Democrats the east.
As one-time lieutenant governor of Virginia Henry Howell (who won that post to the amazement of everyone) liked to say: ‘There’s more things going around in the dark than Santa Claus.’
Thank you for the smile, Ebeneezer S. and Quake 888.
You reminded me that I love the commenters here because not only are all y’all “my people” in a political sense, but also as fellow science nerds/scientists.
It warmed this scientist’s nerdy heart, and I needed that today!
Dr. Feynman was a joy to listen to (search Youtube) and to read. Some others: The Pleasure of Finding Things Out, What Do You Care What Other People Think.
Also interesting is Tuva or Bust!, by his good friend Ralph Leighton.
I still have his “Lectures on Physics” textbook about Electricity and Magnetism on the shelf in front of me. The most readable textbook I ever used.
LV, everybodies gold standard, does not include 18-24 year olds who are voting for the first time. There may be a surprise for pollsters with women in this cohort.
I sort of agree with EW–the Selzer poll is significant, herding is almost indisputable, and women may have been underweighted with respect to men. But. The D shift of older voters has been noted many times before, and is probably baked into everybody’s model. And Selzer may be the best in the business, but individual polls can be weird. Two-sigma errors happen. (Remember Bernie in the MI primaries in 2016?)
It’s still a coin toss. We’ll find out in two days.
It might or might not be a coin toss. But definitely vote as if it’s a coin toss, because otherwise you might get a nasty surprise on Wednesday morning!
I am not in a position myself to dispute this statement.
“ The D shift of older voters has been noted many times before, and is probably baked into everybody’s model.”
But it seems to me to run counter to the views of people knowledgeable in the minutiae of the problematics of electoral polling.
Take eg The Bulwark’s Tim Miller and Sarah Longwell describe Ann Selzer as “America’s Best Pollster” and discuss it all here (as mentioned in previous thread but seems relevant here)
https://youtu.be/yTU0ZWFNmzs
And in particular discuss the issues of herding and weighting in polls.
This issue with white women voters is about misweighting rather than underweighting. Comparable to the error in treating the party registration of early voters as predictive/ indicative of the votes the cast:
there are strong grounds to believe that there are few registered D’s voting for Trump and that substantial numbers of independents and Rs voted for Harris.
These are precisely the issues that poll herding may not have adequately captured, as they clung to weighting for Trump derived from having missed the Obama-Trump voter shift.
The mini-pod is worth the watch IMHO
Interesting numbers here:
https://miamioh.edu/news/2024/11/miami-university-students-conduct-survey-of-ohio-voters.html
This poll does not account for Dobbs impact or the attack on Springfield, Ohio or the impact of Red, Wine and Blue. When Red, Wine and Blue did a fundraiser Zoom event for the Springfield, Ohio schools and Haitian Center, they exceeded their $20,000 goal in 15 min and had to extend the fund goal eventually to $200,000 and they exceeded that goal. When abortion rights were on the ballot, it won with 57% of the vote.
As I drove through a central Ohio community last week that is normally red dominant, Harris-Walz signs outnumbered Trump signs 2 to 1. As we take our Sunday drives in the country, the normal “all Trump” very red signs are now split evenly slightly breaking to favor Harris-Walz.
I think pollsters are missing many dynamics in Ohio.
I also think the research farmers did in Iowa about Project 2025 intending to kill the farm bill, gov backed crop insurance and gov backed soil conservation has started to reach Ohio farmers.
I think Ohio has a chance to go like Iowa and be in play.
In these parts of Michigan, the Trump signs outnumber Harris signs very heavily. BUT, between us and our neighbor, we’ve had 12 Harris-Walz signs stolen.
Put up another Harris sign last night! Gotta go check and see if it’s still there this morning.
Signs don’t vote!
While I agree signs don’t vote, this central Ohio community would be the exception to that phrase.
In Ohio, I thought it was one vote for each pick-up truck. Two, if you have the right 8×12 flags mounted in the back.
I’d make a sign to put next to your new sign. “Every time you steal my Harris sign, I give her another donation and get a new sign. More signs=more money for Harris. Thank you for supporting her.
[Welcome to emptywheel. Please use a more differentiated username when you comment next as we have several community members named “Brett” and “Bret.” Please choose and use a unique username with a minimum of 8 letters. We have adopted this minimum standard to support community security. Because your username is both too short common it will be temporarily changed to match the date/time of your first known comment until you have a new compliant username. /~Rayne]
A complete and utter dipshit in my neighborhood nailed his Trump/Vance sign to a tree in his front yard, along with his pathetic homemade “vote yes on amendments 1 & 2” sign made from a used cornhole board.
Yes, you read it right, he nailed campaign signs to a tree.
Can’t fix stupid.
From a small town Ohioan native, who grew up in (he who will not be mentioned’s) Congressional district…4th district…..Go Buckeyes!!!!
Pure theorizing on my part, but I expect Iowa is the state with the second highest Walz effect.
My cousins in Cedar Rapids have voted Trump until this year and are now voting Harris bc of Dobbs!
But will they vote republican on the remaining ballot? That’s the thing that most concerns me. I’ve seen numerous Harris/Walz signs & yard signs in my red stronghold & I have to say that is a first in my nearly 50 years of living here, but will they vote straight democratic? I ask the goddess for every sign I see.
Here’s Kos’ take on it:
https://www.dailykos.com/stories/2024/11/2/2281976/-Harris-leads-in-the-one-Iowa-poll-that-matters-with-major-national-implications
He hasn’t quite reached the level of disregard for Silver that Marcy has, maybe because he hosted Nate in a bygone era. But in that post he does a good job of underlining the basis for Selzer’s cred.
So which poll is counting the 18-24 year olds?
https://circle.tufts.edu/latest-research/youth-voter-registration-major-challenge-2024-election
“Using voter file data aggregated by Catalist, we calculated the number of youth (ages 18-29) registered to vote in each state with data available as of October 28, 2024, and compared it to the number of youth registered on November 3, 2020.
We find that nine states have more youth in that age group registered to vote now than they did in 2020, including major battleground states like Michigan and Nevada, as well as Kansas, Tennessee, Texas, Oklahoma, West Virginia, Colorado, and North Carolina.”
Wow. We could see some red states swing left based on youth vote combined with increased excess death mortality among Trump voters.
That sounds like North Carolina. Plus, it has a Nazi thug for a Lt Gov., who is running for Gov.
Not certain about NC because of its changing demographics. Over 30 years it has become less white, but non-whites had worse mortality during the pandemic.
If the GOP gubernatorial candidate wasn’t such a moron, the odds of going blue down ticket would be iffy. Could be even thanks to Robinson.
Agreed, I think this is a highly underestimated voting bloc. Pretty much anyone in this age bracket is missed by polls – how do you even reach them when they screen unknown callers and delete unsolicited texts? Anecdotally, the rumbles in their safe spaces (hello TikTok) show that they are pissed and ready to vote. Basically half their lives have been dominated by exhausting Trump chaos/global pandemic/major climate disasters – they’re done with it and ready for a positive change.
[Welcome to emptywheel. Please use a more differentiated username when you comment next as we have several community members named “Danielle.” Please choose and use a unique username with a minimum of 8 letters. We have adopted this minimum standard to support community security. Because your username is too common it will be temporarily changed to match the date/time of your first known comment until you have a new compliant username. /~Rayne]
Not surprised at all to see Kansas on that list.
When KS voted to protect reproductive rights and keep it in the KS constitution just a month after Dobbs came down, it was young women who drove the winning campaign. They crafted the messaging that would work in KS (not choice and women’s rights, but “keep the legislature out of my doctor’s office”), they knocked on the doors, and they registered a helluva lot of new voters — who turned out in droves.
The pollster who comes closest to matching the actual vote this year will be the pollster who best figures out how to include new voters in their sample of “likely voters.”
A new “Kansas Speaks” poll from Fort Hays State University shows Harris only 5 points behind Trump in that ruby-red state among “registered voters excluding those who do not plan on voting or can’t vote.”.
https://www.fhsu.edu/docking/Kansas-Speaks/
I’ve been surprised by the “tied” polls because I expected a large Dobbs effect. Polls also do not measure enthusiasm. I’ve posted before that Obama in 2008 had far more energy than McCain. I believe it was Cincinnati that a reporter went to the Republican HQ and there was a phone bank with 20 phones, but only two people were manning them. At Obama HQ, they had more volunteers than they could handle.
Lots of polls had Obama winning a state by <2% and pollsters said this was within the margin of error, but after seeing the enthusiasm gap, I thought Obama would not only win those states, but also states where he was behind by <2%. Missouri was the only state I was wrong about, Obama lost by < 10,000 votes, but he won all other states whether the polls had him ahead or behind by less than two pct.
This year is Obama 2008 redux plus. The enthusiasm is off the charts. I’m 70 and never had a bumper sticker or yard sign before. Now there’s a Harris/Walz sign and a Colin Allred sign as well in the front of my house. I’m introverted, have depression, and anxiety issues that have increased with age, but I was on a phone bank last week in a Texas AFL-CIO outreach for Harris and Allred. I’m a Dobbs granddad with five granddaughters. My wife and I voted blue the first day of early voting. Finally, not everyone who reads emptywheel comes from the political left, if you catch my drift. Kamala will win and it won’t be close.
New registrations reflect enthusiasm. Tom Bonier stated that when a demographic group creates new registrants its a reliable sign that both the new registrants and the previous registrants will vote.
Are there logical explanations for the Iowa poll? There are many. But the Dobbs decision and the example of former Rep Liz Cheney stand out. The pathway has been cleared and paved.
I’m still not sure whether NYT’s polls, which are based on voting rolls, incorporate post-Kamala registrations. If they don’t, it would explain a great deal about their skew.
From your keyboard to God’s ears. FWIW, I’m situated very close to you (also in Texas, with three granddaughters so far, also voting on the first day, before 9 AM – we got there so early that one of the poll workers was struggling with my check-in, and said “I haven’t had my coffee yet”).
also from the FWIW dept:
“’I think Harris is going to win easily. I don’t think it’s going to be particularly close,’” said [ex-GOP strategist Stuart] Stevens.
“’It’s the most stable race I can remember. 47% of the country is either MAGA or open to MAGA and 53% isn’t. So the Harris campaign’s goal, task, challenge has been to get as much of that 53% as they can and get them to vote. So we wake up in a world where our Senator Bernie Sanders and my old friend Liz Cheney are on the same side.'”
Stevens was a top advisor to five Republican presidential campaigns, including Romney, G.W. Bush & McCain, and consultant on dozens of GOP campaigns for governor, house & senate. He is working with the Lincoln Project at this point.
(from The Vermont Conversation, with David Goodman, 10/30/24. vtdigger.org podcast.)
Great post. Ann Selzer is getting a lot of attention. Given Selzer’s accuracy in recent elections, even if Vice President Harris doesn’t win in Iowa, if she overperforms by a lot that bodes well across the PA-Michigan wall. I found it noteworthy that Other Nate used the passive voice “blamed on” rather than “caused by” when supporting his baseless assertion, leaving out who is doing the blaming (spoiler alert, lazy media).
This concept of “herding” has crossed with an old tune running through my head these last few days – “The In Crowd” (Dobie Gray, 1964)
I’m in with the in crowd
I go where the in crowd goes
I’m in with the in crowd
And I know what the in crowd knows
Any time of the year, don’t you hear?
Dressin’ fine, makin’ time
We breeze up and down the street
We get respect from the people we meet
They make way day or night
They know the in crowd is out of sight,
Just think back to high school, or grade school. There’s always been an in crowd. Don’t you want to be in? Sounds like Nate sure does.
The only crowd Nate will be in with when Kamala blows this out of the water will be the Trump crowd claiming that a fix was in.
Which brings to mind exit polling, which only samples non-absentee voters (which this year may account for only ~55% of the vote total). Is anybody still doing exit polling, and if so, why?
https://election.lab.ufl.edu/early-vote/2024-early-voting/#us
If you interview only individual voters leaving the polling station on election day, you would miss early voters, mail-in voters, and drop-box voters. I saw no one interviewing early voters at my precinct, for example. So how do pollsters choose which stations at which to conduct interviews?
Moreover, people seem to be as reluctant to be interviewed leaving the polling station as they are by phone on any day. If so, that would also skew the poll.
They choose them in a manner dictated by their representative model. They’ll identify precincts with various demographic characteristics (race/ethnicity, age, wealth, etc.), various political characteristics (heavy Dem, heavy GOP, mix, etc.) then blend the results to get their overall representative sample.
High quality polls have folks out there talking to early voters right now, but the only way to reach mail-in or drop-box voters is via a standard telephone poll. I suspect that at this point, folks making calls like these ask early on if the person has already voted, and if the answer is yes, they hold that data to mix into their traditional exit polls.
I remember that song, but had no idea it was by Dobie Gray. I only thought of “Drift Away” when I heard his name. Thanks!
Pleasure.
First, this post and the comments are most informative. I drove my 98 year-old mom to early vote last week. She left her walker at the house, didn’t need the small flashlight I gave her for reading the ballot, either (we read it beforehand). She just didn’t want to wait for ED to vote Harris.
Second, I wasn’t familiar with Dobie Gray and listened to his version. My parents saw The Ramsey Lewis Trio perform The In Crowd in a club in the 60s and subsequently bought the live album. For years, my dad would blast it so loud the windows shook! Impossible to be sad listening to it (https://www.youtube.com/watch?v=jsFST-7Hx-Y).
Had to reply to the In Crowd/Ramsey Lewis mention. Same with my parents in 1960s St Louis! Its a part of my DNA! Go Harris/Walz!
I remember that song from my elementary school years but I’ve been deaf most of my adult life and can’t remember the tune. Hate it when that happens. I’ve been trying to remember parts of the tune “We’ll Meet Again” that ends the film Dr. Strangelove, too. Love that song and I’d love to have all of that wonderful sound back in my head.
I realize this strays further from the post, but since you were able to hear as a young person, you may be able to “pull” the sound memory. There are musicians (notably Beethoven) who learned to “hear” music by touch after hearing loss. I just played Vera Lynn’s “We’ll Meet Again” (via YouTube) and laid my hand on the speaker cone. At first, I thought I was only picking up the base, which may not be helpful, but then noticed that other parts of the orchestration could be felt and were identifiable. I tried it a couple of times to be sure. Just found Vera Lynn’s isolated vocals for “We’ll Meet Again” on YouTube: https://www.youtube.com/watch?v=nLcl2iWAEig
Ever hear of Evelyn Glennie, deaf Scottish percussionist? Amazing story:
https://en.wikipedia.org/wiki/Evelyn_Glennie
Hey there! Remember me? The thread?
Come back to me and my topic, thanks!
Except the gender split in the 71 million early votes is already 53% women to 44% men and 3% undisclosed.
Right. Rather than writing the Jack Smith post I’ve got started, I’ve been simply staring at the Siena cross-tabs, bc they’re all over the map on what they do with gender of those who’ve already voted. As if in some states, they think the early vote cannibalized most of the ED vote, but in others they expect the current rates to continue. It’s baffling.
A few days ago, Tom Bonier was on Joe Trippi’s podcast and said that GOP early vote is cannabilizing the Election Day voters with something like 45% (IIRC) who shifted their vote to early voting. For Kamala’s, the rate is only 14%. He said that bodes well for Kamala on Election Day.
youtube(dot) com/watch?v=EZvQCIsu8Zs
EW:
This data might lend some Dobbs insight:
https://news.gallup.com/poll/245618/abortion-trends-gender.aspx
Thank you for this post btw!
It’s also very different from locality to locality.
Last week, I heard an interview (don’t recall where) with the head of the Democratic party in NC. After the interviewer noted the unexpectedly high number of GOP early voters in a couple of rural NC counties, the party leader chuckled, then said this (paraphrasing here): “In those counties, many local — city or county — races are decided in GOP primaries, so if folks want their votes to count in those local races, they register as Republicans. I guarantee you that a sizable number of the Republicans who voted early in those counties cast their votes for Harris.”
The gender split in early voting has favored women for a long time. For example, it was 55-45 in Georgia in 2016, which Trump won.
Mallory McMorrow, State Senator from MI, posted a reminder 16 hours ago on IG and TikTok, that 160,000 White Women for Kamala zoom call lasted 4 hours. She had a male friend text her, “Trump is screwed,” when he learned the number of participants and how long the call lasted with an engaged audience. She noted suburban women have been late to the call for justice but that many suburban woman are finding their voice to stand on the side of justice and human rights.
I cannot find her clip posted on any other social platforms to link here. If you have IG, look her up.
As for Iowa.
Then there’s this which I wonder if it might be motivating more women to vote in Iowa? 🤔. https://www.nbcnews.com/health/womens-health/iowa-abortion-law-to-go-into-effect-rcna162810
Yes, absolutely. It’s primal for many women, including those over 65 who experienced pre-ROE America.
Helen Reddy – I Am Woman (Official Lyric Video) 1972
(https://www.youtube.com/watch?v=5WiED7UxcRw)
I am woman, hear me roar
In numbers too big to ignore
And I know too much to go back an’ pretend
‘Cause I’ve heard it all before
And I’ve been down there on the floor
And no one’s ever gonna keep me down again
Yes, I am wise
But it’s wisdom born of pain
Yes, I’ve paid the price
But look how much I’ve gained
If I have to, I can do anything
I am strong (strong)
I am invincible (invincible)
I am woman
After the polls close on Tuesday, I think it will be:
“I am woman, hear me vote!”
Put a woman in charge
https://youtu.be/SYocuL6Rd-w
Is Kamala ‘brat’ or ‘boss’?
https://www.youtube.com/watch?v=xfjgTKQd3SI
I would love to see Brian Setzer and Gwen Stefani perform this live at the Harris inauguration!
“And when it comes to knowin’ which way the wind is blowin’ – aw Baby – you sure are wise…”
‘Nuff said.
What did Nuff say?
According to a new YouGov poll, 1 in 8 women voted differently from their partners in an election without telling them.
x.(dot)com/briantylercohen/status/1852838248290341057
salon(dot)com/2024/11/01/yougov-poll-1-in-8-women-say-they-have-voted-differently-than-their-without-telling-them/
My late neighbor told me that whenever she voted, she cancelled out the vote of her husband, and she was brave enough to tell him so. He was the Chairman of the local Republican Party. She was that generation of women who were born in the early 1900’s; the generation which saw women enter the work place, seriously, during the Depression and the War. They passed that grit and determination on to their Silent Generation children, who have passed it on, and on and on. Change comes: it moves in its inexcorable fashion.
Conclusion: women and the current generation young women voting for the first time will make a huge difference. Their Grandmothers and Great-Grandmothers lit the pathway.
Don’t ever underestimate the women who are married to authoritarian males. The ballot box is the greatest vehicle for equalization in this country.
The only way legislation that could make it more so would be to create more polling places.
Needless to say, Republicans have put in a lot of effort to close polling stations, in Democratic leaning districts, of course.
My mother used to say that, too – but she was registered R…until 1980, when she officially became a Dem (she’d been voting that way for years. She *hated* Nixon for what he did when running against Helen Gahagan Douglas.)
I guess the Nate Nates didn’t pay attention to the seniors at The Villages. Unlike the Nates, the people there don’t live in silos. They are networked in multiple ways.
And one thing they are concerned about is Project 2025 and Medicare. They also are concerned for the future of their children and grandchildren. And they are concerned about social issues, including women’s health issues and their right to autonomy over their own bodies. Like I’ve said before, we seniors remember the struggles we faced in the 1970s.
https://www.youtube.com/watch?v=vG2rU76HvIE
“Kamala Harris Golf Cart Rally in The Villages, Florida”
“What was in it for them?” Donald Trump to John Kelly, referring to the war dead in Arlington National Cemetery.
That line and the I/Me/Mine focus (to the exclusion of all others, including children and grandchildren) consistently displayed by Trump came to mind immediately upon reading your account about the Villages, SL. I’m sure he considers the inhabitants “his” by divine right. He would find their true feelings incomprehensible.
The “October Surprise” is that male pollsters didn’t account for pissed-off woman. Sounds about right. Its not like the 2022 election results and every special election thereafter wasn’t a clue for these guys. (It was).
A picture worth a thousand words: how many older women angry enough to march in spite of disability also voted? How many voted for the first time, like 70-year-old singer Stevie Nicks?
This would explain the support that VP Harris has in The Villages in Florida. Normally a consistent red voting block.
Yes, older women remember when.
“A picture worth a thousand words.” Indeed. Great photo, Rayne. And thanks for sharing your graphic and visual arts skills in previous posts. You have your finger on the pulse. One of your many talents that enrich us. Much appreciated.
Thank you Rayne. Screams Volumes!
“Oh honey. Don’t worry your pretty little head. Vote as I vote because I am the king of the castle and a control freak.”
This election with Vice President Harris running has exposed the misogyny Trump-Republican right wing religious radicals. Always been there, however hidden with the lack of respect, devaluing of women and suppression. Now unearthed. If they have their way, they want to go back to women being pregnant, barefoot and in the kitchen which would be the Immaculate Conception for myself and many of my mature and wiser girlfriends.
Looking forward to saying, “Madam President.”
Thanks, Marcy for this.
Last week I saw this interview (H/T Hopium), and I got rid of most of my anxiety. I quit looking at polls. https://www.msnbc.com/inside-with-jen-psaki/watch/-we-re-going-to-win-this-harris-campaign-chair-jen-o-malley-dillon-on-why-she-s-very-confident-222797893682
I started looking at voter turn out, which is broken down by age, gender, type of ballot, party registration here. https://www.nbcnews.com/politics/2024-elections/early-vote
Until yesterday, Iowa’s turnout was astonishingly, heavily Democratic. I referred to the prediction which said Iowa was solid red. The Rs caught up yesterday. Women have been leading the men in voting in every state I watch, which is not for the most part red states. Iowa was looking like an outlier. I was not exactly surprised to hear the news/polling about Iowa all of a sudden.
Women voters lead men by increasing numbers across the country with the split about 10% consistently. The number has grown. https://targetearly.targetsmart.com/g2024 The number of women who have already voted nationwide is 12 MILLION more than men. I understand the tsunami of votes are expected on Tuesday. Abortion is on the ballot in 10 states with FL being the only state requiring 60% to pass.
Republican women are voting for Harris and perhaps other pro-choice legislators. I don’t know those numbers, but certainly it is happening. MO provides no data. TX and AZ look difficult, but Colin Allred is doing well, and I think Ruben Gallego is fine.
I have been giving money to turnout campaigns: Native Americans, youth, MoveOn, and others, as well as to Senate campaigns. I am still nervous, but I think Harris is going to win. Her appearance on SNL last night was a joyful moment. Friday for the first time I heard “trifecta.” I can only hope.
I am a little bit of a data nerd. I door knock and have been a volunteer for many campaigns. I’ve been involved with turnout for a Congressional district that has flipped several times. If we win this one, it will be the first time for Ds to hold it more than 1 term. I get data, adjust for the party lean of the county with the undeclared votes, and sometimes track the daily reports. It has been a reliable indicator in the past. Our current campaign in this district has targeted 40,000 voters with multiple mailings. I believe the targets have been “unlikely” voters. The campaign has had EV turnout of over 20%. The data on this campaign has been tracked and is reported 3X daily. The last time the race was won with about 1300 votes.
A friend who drives for Uber has asked every young voter if they are registered and have voted. They are all registered and only one had not yet voted.(One young woman: “My father thinks I voted for Trump. I voted for her.”) We are an abortion destination. A lot of women and often their partners come for care to our clinics, mostly they are from TX. (Also Uber info) Abortion is on the ballot in TX, though not with a voter initiative. The deaths of women from the loss of healthcare in TX are broadly known.
Everyone asks how it is possible the election is this close. I think it is not that close.
My 95 year old father has said for months, “Women will put Kamala in the White House.” I agree.
Reading the specious arguments of the Two Nates, I’m reminded of Twain/possibly Disraeli: “There are three kinds of lies in the world: Lies, Damn Lies, and Statistics.” – Polling seems to be the Nates’ “wishful” statistics.
[Welcome to emptywheel. Please choose and use a unique username with a minimum of 8 letters. We have adopted this minimum standard to support community security. Because your username is far too short it will be temporarily changed to match the date/time of your first known comment until you have a new compliant username. Thanks. /~Rayne]
a lifelong West Coast resident, Presidential polling has always been something i watch play out from afar. here in WA State, with Kamala the likely winner, the attention is focused more on our resident Proud Boys/Militia Maga chuds and the havoc they may get up to. already the fires in Drop Boxes (thing i learned: they put fire suppression devices inside the boxes for this scenario. one deployed and saved a lot of ballots, another did not). our outgoing Governor, Jay Inslee, has already called up our National Guard to be on call for any pre or post election violence.
All my female friends are fired up and voting, voting, voting. And one more Kamala California vote is my 96-year old aunt. Sibling helped her fill out her absentee ballot and into the mail.
1. Ann Selzer, excellent though she is, isn’t going to be much more accurate than her data, which, after all, is a (carefully constructed) sample of 808 people with a 95% confidence interval of +/- 3.4%. As a professional statistician with decades of experience, some of it in sample surveying, my take is that if my prior belief were that Trump had a 100-1 chance of winning Iowa, my posterior belief would be about 5-1. Yes, there is some math behind that. Great news! And even better news for Wisconsin, Michigan, and, dare I say it, Pennsylvania!
2. Herding is hardly Nate Silver’s fault, and he doesn’t do it himself – he’s not a pollster but an aggregator whose results depend to a large extent on the (herded) polls. Blaming him for this is like blaming a sports statistician for Aaron Rodger’s lousy season to date. If he didn’t exist, the market forces that compel this unfortunate practice would be the same and just as strong – no one wants to be significantly wronger than everyone else because of the reputational hit and consequent loss of business.
Always one to find the black thread in a silver cloud. Mary’s discussion about Nate Silver illustrates the impact of bias on any data set. We’ll also have to disagree about the role of Nate Silver and your position that with or without him, we’d see the same thing.
All right, I’m somewhat of a party pooper, I admit. But regarding the role of Nate Silver – I was in this business when Nate was in elementary school, and not being the wrong outlier was a known economic influencer even then. On rethinking what I wrote, though, it may be that the major pollsters are more afraid of being the wrong outlier now than in the past because of Nate’s presence in the marketplace and his track record. Maybe…
Typo: Marcy. My “c” key is a little stickley.
Earl of – at least it’s the ‘c’ and not the ‘g’, otherwise you may be up a bit too high and waste time diddling with a stray pimple or a collar button.
*showing myself out now…*
1. Harris/Walz doesn’t need to win Iowa
2. The Republican never had 100/1 odds of winning Iowa. Candidate and then President Obama won in that state in 2008 and 2012. No poll presents material facts without reference to elections.
3. If the poll information appears to be infected with herding behavior the non poll information becomes more persuasive.
1. Never said or implied they did,
2. “If my prior belief were…” was illustrative of a calculation, I should have made that clearer. If my prior belief was, say, 20-1 in favor of Trump, I’d revise those odds to about 50-50, myself.
3. Completely agree.
Got it. Thanks for the reply.
These points are all well and good, but only if one chooses to ignore the track record of this pollster being an outlier from the mean and capturing something significant in her data which other herd inclined pollsters missed. And in the previous cycles, particularly 2016 what was significant was something which helped explain Trump’s success.
That latter point, and the Trump campaign’s need for good news had them salivating in advance of the Selzer poll dropping so convinced were that that this gold standard poll would deliver the good news they were desperate for from Iowa.
Their hearts and brains have been broken by the actual poll numbers, and are furious.
See Pondering Politics
https://youtu.be/5tTDBBFmoB0
I suspect their fury is because they know that this poll (whether wholly accurate or not) says something true and revealing about the disgust with Trump’s politics and the ineptitude of their campaign. It also reaveals the extent to which they rely on herd polls to create a false perceptions of the campaign’s success as a basis for later claiming that a loss is due to theft.
Oh, sure. I’m just emphasizing the survey’s imprecision, not discounting it entirely. My bad for giving that impression.
Personally, given the survey and my prior beliefs, I think there’s a reasonable chance that Harris ends up within 3% of Trump in Iowa and a smallish chance she wins it. (Trump won Iowa by about 8 points in 2020.) Of course, it points to what is likely happening in other states under the adjusted polls’ radars, which is probably even more critical to a Harris win.
I have to say, it cheered me up immensely, although Madison Square Garden throwing off the MAGA mask was also pretty cheering, especially given how the Puerto Rican community reacted. I feel good about this election in a way that I didn’t in 2020.
I’m definitely with you that Selzer’s results don’t convince me Harris is going to win IA. I’ll take those 5-1 odds.
Selzer describes herself as shocked by her own data .
Interview with Tim Millar. Emergency Bulwark Podcast
https://youtu.be/P-ysKh_Gyd0
Two Iowa districts in Congressional races are tilting to Ds including
53:37 tilt to D in 1st Congressional district and abortion in particular is the driver in that district
The abortion issue is driving the politics of Iowa since their 6 week ban abortion law going into force in the summer
The Presidential campaigns spent little or no money in Iowa, so it maybe the congressional races are driving the politics.
Gender gap older women 63:28 D
Evangelicals still 3:1 for Trump
From show notes links to segments
0:00 – Intro
1:14 – Iowa Poll Results
5:56 – Accuracy of Selzer’s Poll
9:52 – Method of Releasing Polling Data
14:05 – Congressional Election Data
15:49 – Selzer’s Reaction to the Polls
17:35 – Selzer’s Poll in ’04
19:25 – What This Means for Other Swing States
As someone who uses her mind in a more intuitive way, polls, and numbers, odds, and horse betting amount to sheer gobbly-de-gook. Such conversations leave my head spinning, especially concerning such serious activity as voting. Voting is not a horse race, a card game, or any other game. Voting is an act of emotion, grievances, education level, research, self-interest, prejudices, etc. That is something impossible to quantify. It’s like trying to define what the Human Spirit is.
The only things quantifiable are the number of people voting in general, the age groups, and the political party affiliations. The latter is a moveable feast.
I’ve crossed over in voting in past years. Even political identification for voting purposes is not a given. In the end, it’s how many people are willing to follow the leader over the cliff or up the mountain. My intuition in this election is that more people are headed for the mountain.
I’m with you on this because of all your cleareyed reasoning.
The personal reasons for me is that I majored Math, minored statistics, MBA Finance, 30 years on the street.
All of my previous Comments on this election have been concerned with observations. Women, Youth, Elders. (and their observers – bartenders, taxi drivers, etc).
Also on debunking Polls.
I, like everybody else, have a Big Bet riding on this. So far commenters here have remained levelheaded, so are beginning to see the first positive signs of our Vice President’s accomplishment in her election.
Seltzer’s recent poll may be indicating systemic polling error, especially with regards to MOE of other Iowa polls, and perhaps many other non Iowa polls too.
I see error from two sources: 1) the polling itself and 2) the forecast composition of the voting electorate
To me Seltzer’s results might be indicating a much bigger than usual systemic error in 2 in other polls. That would explain a lot, imho.
If Seltzer’s electorate model is correct, Harris winning by 6% is just a likely as Trump winning Iowa. MOE cuts both ways.
It’s tempting to think otherwise, and surmise ‘I guess Iowa is closer than I previously thought’ but Trump still probably wins. But that really is just tantamount to mentally aggregating Seltzer with other prior polls, which in turn discounts the magnitude of potential systemic error that might be present.
She has an impressive track record of making correct adjustments to her model late. The systemic error in other projections could be very real.
We will know soon enough.
I understand pollsters for trying to get it right by conforming near other polls through weighting.
But I do blame them, and Silver especially, for not looking deeper much earlier, and recognizing the lack of variance.
Even if they didn’t do a deep dive into each poll they should have recognized sooner that the combination of a greater % of women voters, them being motivated, rising uneducated Reagan voters dying, Covid disparity deaths, and finally Trump’s terrible non-inclusive campaign – should have been setting off alarm bells long ago.
The concerted effect of horse-race results may also inspire hard to quantify anti-Trump voting.
1. The herding effect wasn’t nearly as great much earlier. It ramped up as we approach the election.
2. Do you really think none of the pollsters did all that stuff you mentioned? The problem is threefold – the last normal Presidential election we had was eight years ago, there aren’t many of them and they are all significantly different, and how do you quantify the effects of something like “Trump’s terrible non-inclusive campaign” when the last such was George Wallace in a three-way race 50 years ago? Prediction is *hard*, especially about the future.
2016 was a normal election? Certainly fooled me then. ¯\_(ツ)_/¯
Nate Silver goes on to say that he thinks that Trump will beat Harris 55-45 in the popular vote. Which is utterly delusional. Trump should be absolutely ecstatic if he was somehow able to barely crack 50%. And no, I don’t think that will happen. More likely 53% Harris 45% Trump, 2% protest.
I need to go pull the exact quote from the book, but Terry Pratchett (RIP) had a line in Equal Rites, where the main character grew up in a smithy. Paraphrase: “There are different kinds of fire. There’s the kind when you start the forge that’s all great orange flames that look big but do no work. Then there’s the small white hot flame that cuts steel.” Guess where women are right now.
Oh, my. Nate Silver thinks Trump will beat Harris 55-45 in the popular vote? Delusional doesn’t quite capture it. If that were true, even his polls would say the race is not neck-and-neck. Silver is so in the tank for his billionaire patron and the gambling syndicate. No one should publish his work.
Right. Who did his sponsors bet on and why would they want to drive up bets on the Republican candidate?
I have read that Terry Pratchett disavowed being a feminist. I’m not sure why, perhaps he just didn’t want to take a hit on sales, but he sure reads like a feminist. Particularly that book, that introduced the character Granny Weatherwax and her posse, along with its followons.
I think he meant there was a 55% chance he would beat Harris in the popular vote, not that he thought Trump would have a 10% edge in the popular vote.
538 has been correcting people’s perception on the 55% vs 45% thing for many years. It is not even vaguely related to a 10-point lead in vote totals, but to the near-tie between a 55% and a 45% chance to win.
In NYT terms, it’s the Needle quivering near the center point, versus being nearly pegged toward the side with the 10-point lead.
538’s efforts to correct people’s perceptions do not seem to be remotely effective.
These are Silver’s words, as captured at The Daily Beast:
“Silver has maintained his prediction that Trump will overtake Harris, 55 percent and 45 percent respectively, to return to the White House.”
Maybe he means that Trump has a 55% chance to win. If he does, he’s using extraordinarily imprecise language to say that.
https://www.thedailybeast.com/nate-silver-cheating-pollsters-are-putting-finger-on-the-scale/
Imprecision. Not a characteristic normally associated with statisticians.
Nate Silver also said today or yesterday, in his hurried response to the Iowa poll he disagreed with, “Please forgive the typos.” So, maybe imprecision is part of his brand, which means he should be unemployed sooner rather than later.
The state polls and the Harvard Youth poll contradict the national aggregate polls. Polls not influenced by herding behavior provide better information.
Maybe this election will be the inverse of 2016. Eight years ago, polling missed the extent to which Trump was likely to win, and perhaps some Democratic voters, relying on overly sanguine polling, felt it was ok to sit out and watch Clinton coast into the White House. This year, polling may be missing the extent to which Harris could win. This less-sanguine polling might fuel Democratic efforts to GOTV. With a “50-50 race” and the lessons of 2016—to say nothing of the effects of Dobbs and an authoritarian insurrectionist on the ballot—Democrats will take voting very seriously. One would hope anyway.
I have enormous respect for Marcy Wheeler and believe that she is very right about pollsters herding in a way that helps Trump.
However, she is paying more attention to the public health study that, after the vaccine became available, there was a measurable difference in death rates between Biden and Trump majority counties, attributed to antivax propaganda. I concur that this difference is real but I believe that it is much smaller than the 2 larger mortality trends, which will also affect polling, via recalled voting and other devices used to adjust raw polling data.
Am linking to a CDC report on 2023 versus 2022 mortality which shows the larger differentials.
https://www.cdc.gov/mmwr/volumes/73/wr/mm7331a1.htm#T1_down
The early part of COVID before vaccines, manly in 2020, was deadly to our black and Hispanic sisters and brothers. This is when most of the 3 million deaths occurred. Whammies in all directions: they started with more chronic illness and less access to high quality treatment and disproportionately worked in professions that required them coming to work. The death rate among people of color is always materially higher and this difference spiked during the pandemic. (See NYC hospitals being more overwhelmed).
Older white people also vote more republican and older white people had a higher mortality rate during COVID. The die off of white people being replaced by more diverse young people turning 18 already drops the white percentage of the voting population by about 200 basis points every 4 years. This was exaggerated during COVID because old white people had a higher number of absolute deaths, even though their spike in mortality was not as high as that among people of color.
I believe that Marcy’s point is correct: pollsters probably did not adjust enough for COVID related death rates when they modeled pre-pandemic populations. However, the horrible and very high mortality among people of color in the pre vaccine days will offset the smaller effect of republicans not getting the vaccine, and likely also mutes any differential pandemic related losses in the white population.
The comparison is not pre-COVID to now. It is 2020 election, when a number of pollsters are taking recalled vote, to now.
After the election, after the vaccine roll-out, those death rates shifted dramatically.
You could even argue that the differential mortality wasn’t properly factored in in 2020 on Biden’s electorate, and now it is not in Trump’s electorate.
[corrected]
“When a measure becomes a target, it ceases to be a good measure” Goodhart’s law.
https://en.wikipedia.org/wiki/Goodhart%27s_law
Polls and Betting Markets have become extremely newsworthy because at one point in time they showed they were unbiased estimates of popular opinion. Some pollsters have adjusted the weights to correct for measurement errors in the past (thereby overfitting current data) while others have learned to game those measurements so that the poling average can reflect what they want it to reflect.
The media needs to rethink how they report on polling but the media as a whole has not shown much introspection, so I will not hold my breath.
[Welcome to emptywheel. Please choose and use a unique username with a minimum of 8 letters. We have adopted this minimum standard to support community security. Because your username is far too short it will be temporarily changed to match the date/time of your first known comment until you have a new compliant username. Thanks. /~Rayne]
Tim Miller interview with Ann Selzer. Approx 1:00pm CST Nov 3rd.
.youtube.com/watch?v=P-ysKh_Gyd0
Content description in comment here
https://www.emptywheel.net/2024/11/03/male-pollsters-shocked-shocked-when-a-woman-pollster-discovers-women-voters/#comment-1076781
That was Miller and Sarah Longwell yesterday (20 hours ago). The clip I posted is from today, about 2 hours ago and it is Miller interviewing Selzer herself, not chatting about her with Longwell.
Nope it was the interview I referenced in that specific comment up thread ie #1076781 to which I provided the link,
Which begins
“ Selzer describes herself as shocked by her own data”
It was an entirely different comment in which I referred to the Miller/Longwell discussion ie #1076695
https://www.emptywheel.net/2024/11/03/male-pollsters-shocked-shocked-when-a-woman-pollster-discovers-women-voters/#comment-1076695
Hope that dispels the confusion in your mind.
To SteveBev: I thought Marcy included that link in her post, which was prior to the interview. I didn’t realize it was in the comments section. My anti-depressant medication is changing and I have trouble keeping things straight :)
Thanks for clarifying.
@Alan
I sympathise with you re changes in meds.
I have similar issues from time to time.
Hope you are able to find the optimal balance soon.
Documentary – Geraldine Ferraro: Paving the Way
https://www.pbs.org/show/geraldine-ferraro-paving-the-way/
The trailblazing Vice-Presidential nominee helped change the face of American politics when she became the first female nominee for national office by a major party. Geraldine Ferraro’s journey from an impoverished childhood to her great accomplishments are captured in interviews with political giants and contemporaries, and her last interview before she died at age 75.
I just finished working 16 days of early voting. We had over 12,500 voters come through just our voting center in that time. Huge numbers of women, black voters, and young voters. We don’t register by party so can’t tell that way, but I saw lots of Democratic sample ballots in their hands, way more than Republican ones. Lots and lots of first-time voters too. We are very blue, but the Dem enthusiasm as compared to the other side was palpable.
Better Nates than lever her.
In Nate ill fate showed
(French proverb, ;-) )
hmm, this one isn’t obvious to a francophone:
something something il fait chaud?
En été il fait chaud
Poor effort making mock of stereotypically British frightful French
pronunciation
… I’ll leave my coat and collect it when nobody is looking
@SteveBev
got it!
An anecdote historians love to repeat is that the one uncrackable code the Brits came up with in WWI was to have its officers speak to each other over field phones using English public schoolboy French. Incomprehensible, especially to the French.
They’ve got the poll whirled
in their brands
They’ve got the poll whirled
in their brands
They’ve got a little bit of maybe
in their brands
They’ve got a little bit of maybe
in their brands
https://www.youtube.com/watch?v=if4X17L4J3s
“He’s Got the Whole World in His Hands Piano | Alfred’s 1”
(chuckle)
All smells that trends spell.
I really, truly hope that this election brings about a genuine reckoning with, and rejection of, the special brand of toxic masculinity that’s only surged since 2016. I think for women, 2016 was a wake-up call; 2022 was the five-alarm fire. The post-it note movement, ads, and more we’re seeing encouraging straight women to vote their conscience, whether they disclose it to their male partners or not, is already sparking a backlash from fragile (conservative) men that will probably get worse if Harris wins before it gets better. But hopefully it prompts these women to question why they accept the kind of relationship where they have to hide something like this, and prompts bigger societal questions. (Sorry if I’m not articulating this well; still figuring out my thoughts on it, but something I keep coming back to this past week.)
I’ve been saying for a number of weeks to friends, etc. that Harris will win because of women – that she’ll win the pop. vote by 10 – 15 mil and will win most of the swing states (MI is a crap shoot as well is GA) – and even might win IA and even be competitive in shit places like TX and FL.
I think this despite the oligarchs and corporate MSM (who love Trump for the ratings, anger and ad revenue) pushing heavily for Trump, and the disinformation and complete gaslighting generated ubiquitously by everyone from trolls in this country and elsewhere.
If I am wrong – o well – but I think I am right.
“o well”?
How about where is the exit?
Ta, Marcy. I was pleasantly surprised – but not shocked – by the results of the recent Iowa poll. Why? Well, I grew up there, and my parents – who are in their early 80’s and have lived on the same piece of rural farmland for 50+ years – called me last Wednesday to let me know that they’d early-voted for Harris/Walz. They know I’ve been extremely anxious about this election, and have been telling them that every vote matters.
I also think it’s worth acknowledging that Obama won Iowa in both 2008 and 2012 (thanks in part to people like my parents). I hate the idea of red states and blue states, because it can change at any time based on the candidates who are up for election (see also Georgia). This will be the 1st presidential election post J6 and Dobbs, so I don’t think anyone should take anything for granted.
Couple of Observations:
1. You can tell by Trump’s response how his campaign views the Selzer poll and he is not happy.
2. Biden has been a very good president and if Harris wins, he will have been a great president.
I don’t see how manipulating poll averages in Trump’s favor really helps him. If anything, isn’t it better for Harris — the candidate who is less intensely disliked by and unacceptable to half of voters — if the public believes the race is a toss up as opposed to believing she has a comfortable lead?
I guess poll averages could add fodder for Trump to falsely claim the election was rigged if he loses. But he will claim that anyway and it’s unlikely that he will reverse the result on that basis.
I think it’s more likely the herders are sincerely afraid of the business repercussions of potentially understating Trump’s support 3 cycles in a row.
Trump wants to energize his supporters, and demoralize Kamala’s voters. Energized people are more likely to vote, demoralized/depressed voters less likely.
The perception of a tied vote in spite of everything is indeed demoralizing, so it works in Trump’s favor. Around this time in an election cycle, we see sea lions coming into comment sections with premature declarations of triumph. This is meant to demoralize people. It’s all psyops.
And the perception of a lead in the polls for Trump (even a narrow one) makes claims of theft appear more plausible to those inclined to believe that the election was rigged against him.
See also wash trading on Polymarket
https://fortune.com/crypto/2024/10/30/polymarket-trump-election-crypto-wash-trading-researchers/
Trump doesn’t need slanted polls to convince his supporters. He already showed in 2020 that he could convince his supporters the election was stolen even when polls showed him trailing substantially.
Slanted polls may be good for Trump’s ego & narcisism, and that alone may be enough to motivate their existence, but ultimately, the end game of that doesn’t make much sense. To reverse the election he needs the near-unanimous complicity of GOP Congressmen and/or judges.
See this discussion esp Simon Rosenberg of Hopium from 10:47
1 Re purposes of fake Republican polls
2 The malign influences of betting markets
The multiple incentives to bad faith political actors to use such devices for profit propaganda and political gain
An expert in the field making convincing arguments on all these matters, which I see no reason not to adopt.
https://youtu.be/gRi4Cenw1Kg
Trump voters want their vote to be for the winner. If there’s signs that he won’t win,or even might not win, they might withhold their vote. They want to avoid voting for the loser and then have to deal with the cognitive dissonance afterwards.
When Hillary was presumed to be a lock in 2016, it didn’t demoralize Trump supporters. What it did seem to do is make Democrats more complacent and encourage 3rd party voting.
And Hillary did not get the massive surge of volunteers to support a ground game that Harris is getting now, thanks to the perception that the race is a toss up.
Among the things Trump hopes to do here, is to set up his post-election loss income stream. He needs to keep the base motivated enough to keep sending him money. For one thing, if he loses, his legal bills go up, not down. His tired but outrageous claims will keep him in the news.
What Trump doesn’t care about is that his continued notoriety and grift will be the natural segue for his ultimate replacement.
Three points.
First, Nate Silver’s invocation of a collective sample size of 230,000 is irrelevant. The binomial/multinomial random sampling error reduced by large sample sizes is not what drives differences among polls. The major differences among polls are how they generate their list of potential voters to sample from, and how they weight their samples. The issue is non-response: with less than 1% of people drawn from that sample frame able to be contacted and then willing to answer the poll questions, large biases in participation rates among groups can have huge effects on the predicted outcomes. The current standard practice is to break down the sample into categories or strata: by one or more of party registration, gender, age, education, spatial region, then self-assessed likelihood of voting, and whether they voted in the past several elections. [They also ask other “explanatory” questions like what is your top issue, but those are not used in the stratification and weightings, only in the post hoc narratives.]
Given a need for at least 50-100 respondents in each bin and reasonable total sample sizes (~1000), they can’t break down by combinations of all of these factors, only by 2 or 3. Then, the expected turnout for the election is generated for those bins, and used as survey weights for the poll results by bins. The expected turnout (“likely voters”) is usually based on the previous presidential election, because Democrats have tended to have lower turnout in off year elections. Differences among polls in their weightings, their models of the population who end up actually voting, are generally larger than the variation due to moderate sample sizes. Rasmussen is famous for having several percent higher registered Republicans in their model of the electorate than other major polls.
Second, what Nate Silver calls “herding” is honest statistical work with solid statistical methods behind it (I’m not suggesting pollsters actually follow that solid statistical work). This is a standard Bayesian shrinkage estimator (I’ve used Stein shrinkage estimators, but there are others), which is good for individual polls, but bad for averages or ensembles of polls (like Nate’s). The classic example is estimating the full season batting average of baseball players from the results of their first 50 at bats of the season. If you use just their average after 50 at bats, that will lead to outliers in both directions. But if you shrink those 50 at bat averages back toward each players lifetime average, or the league’s average, you can make better predictions of what their full season averages will be. There’s a lot of theory on how much to shrink, which is what I doubt most pollsters follow. [There are also pernicious applications of Stein shrinkage estimators in things like property assessments for taxes.] In this case pollsters look at the LV crosstabs from other polls and shrink their model back toward that average.
Third, from what I can read in https://www.desmoinesregister.com/story/news/politics/iowa-poll/2021/11/18/iowa-poll-results-questions-how-works-elections-issues-approval-politics/6291360001/ Dr. Selzer and the Iowa poll don’t model “LV” turnout based on previous elections, not this time, but not in previous polls either. This Iowa poll only used responses from individuals who either said they had already voted or definitely would vote (appropriate for a poll when half the votes have already voted, not so appropriate for last June). A lot has changed since those previous elections, including Dobbs, so this approach is very different than “likely voters” modeled from 2020 or 2022, and gets this poll a higher fraction of women in their model of the actual votes than the other polls have. And, as some of those explanatory crosstabs show: women are breaking strongly for Harris. That makes a lot of sense to me because Dobbs has clearly motivated many more women to vote than voted in 2020.
I hope that the low propensity to vote people don’t magically all decide to vote, and mostly decide to vote for Trump. I don’t think that’s likely, as the tracking polls of undecided votes allegedly has them breaking for Harris.
I haven’t followed reporting and analysis on polling very closely, but what I’ve read recently has shocked me with its lack of attention to the post-Dobbs polling misses. 538 published the other day an article headlined, “Trump and Harris are both a normal polling error away from a blowout,” that entirely ignores it, aside from a passing comment that “plenty of polls have overestimated Republicans in the past.” How the hell do you ignore that issue in an article about polling errors? It’s baffling.
On another topic: the reporting that over-65s are breaking tor Harris also shocked me (pleasantly). Is it just a simplistic stereotype in my own head that that age cohort has been reliably red for generations? I certainly never expected them to side with a Democrat woman of color for President.
Just saw a shout-out to this post from Michael Tomasky at New Republic:
https://newrepublic.com/article/187847/donald-trump-fellatio-microphone-rally