General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region Forums"The Blowout No One Sees Coming" from Vantage Data House
Has anyone heard of this group?
Vantage has been tracking seven swing states daily since August. Until recently, the race consistently appeared as a toss-up. Just a few days ago, the numbers started to shift. It's clear the momentum is moving away from Trump. Here's how our nightly tracking data (1,200 interviews per state) stacks up against what FiveThirtyEight (538) and RealClearPolitics (RCP) are reporting in their averages.
https://app.vantagedatahouse.com/analysis/TheBlowoutNoOneSeesComing-1
Wiz Imp
(1,202 posts)Hopefully they are right.
dweller
(24,677 posts) When a campaign releases a poll, its not to inform the public; its to shape public perception in favor of their candidate or agenda.
✌🏻
Fiendish Thingy
(17,879 posts)CaptainTruth
(7,149 posts)The key is to know which polls are included in the average & how they're being weighted, if at all.
Fiendish Thingy
(17,879 posts)538 is excluding a couple of the worst ones, but not all of them.
I dont think RCP is excluding any.
Dennis Donovan
(24,105 posts)Why do our numbers tell a different story than the averages? Simple: public polling is riddled with noise.
Republicans are in serious trouble, though few are willing to acknowledge it. Every major Republican Senate candidate is trailing in swing states according to leaked Senate Leadership Fund polling. Some split-ticket voting still happens, but every major Senate race is down by 5-8 points while Trump leads in the presidential polls. The math just doesnt add up.
LAS14
(14,411 posts)Dennis Donovan
(24,105 posts)Polls are still a good way for the media and the general public to gauge political races, as long as theyre taken in aggregate.
LAS14
(14,411 posts).... unless it is just "stuff that makes a poll inaccurate." The word "noise" was used only in the title and the closing sentence.
cally
(21,680 posts)Long time time statistics days for me but the noise in just normal variation found in surveys. Sometimes the sample taken has responses that differ significantly from other polls or earlier polls or your statistical assumptions in designing polls. Statisticians expect this and refer to this as noise. If other polls have similar responses than these are taken more seriously.
For example, a pollster take s a daily tracking poll. One days results show black women voting overwhelming for Trump. Since all these polls have relatively small sample sizes for various demographics, then a few abnormal answers will change the results. Thats noise.
This is just normal noise and expected. Nothing nefarious.
Bernardo de La Paz
(50,577 posts)We are familiar with noise on a radio when there is loss of signal beginning to occur from distance or weak receiver or poor antenna location or nearby devices generating interference. The noise obscures the signal and makes it harder to perceive and understand.
Ideally (never realized), a poll is an instant snapshot (signal) of all the voters intentions at a particular moment.
Noise comes in (more or less) because of many factors that are dealt with more or less.
1) A poll is not taken in an instant. Typically the data is gathered over several days, or at least several hours. A voter might change their mind after being asked and recorded but before the data collection closes.
2) A poll is only a sample of the voters, not all of them. So the sample might be more or less an accurate representation of the voters as to affiliation, demographics, and geographic location.
3) People responding might lie or make mistakes.
4) People recording responses can make mistakes.
5) Questions may be poorly formed or poorly understood by a voter.
6) When various adjustments are made by pollsters to the output numbers to correct for past biases, known sampling errors, etc., mistakes can be made.
7) Malfeasance can creep in at any stage where participants and pollsters attempt to put a thumb on the scale to skew results for hidden aims.
All of these factors put noise into the results, i.e. they do not perfectly reflect reality.
Farmer-Rick
(11,127 posts)Last edited Thu Oct 24, 2024, 03:45 PM - Edit history (1)
Here's what I read on another post.
It's the best explanation I have seen yet of why the polls are showing a tight race when it's clearly a Harris win.
"This time, because they started earlier and are producing more polls their polls are often only 1-2 points to the right of the polling average or independent polls. They are working the averages more carefully this time, over using the time and volume to move them gradually so the op is not as crude and as easy to see as last time. If Harris leads by 2.5 points and you drop 5 polls showing the race tied or her ahead by 1 or even 2 points the averages moves and it looks like she is losing altitude.
Consider this Tweet from one of the top analysts at 538. It is part of a thread referring to what was a very junky PA poll by TIPP.
https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F16f8f964-36e2-4522-a702-7bbb12522595_898x328.png
(I couldn't copy the image but the link will take you to the tweet he's referring to.)
So here is 538 admitting that even something they consider a bad poll could move the average in PA by a tenth of a point. OK, red wave pollsters say, got it. Appreciate the tip! So I just need to produce more polls to move the average by a meaningful amount. Which is what theyve done. If each of these polls moves the average by a tenth of a point then 16 of them in October in PA could have moved the average by 1.6 points - and poof a Harris lead becomes a tied race.
Elon Musk and primary political patron of JD Vance. Polymarket is buying product placement on sites and with influencers for their 2024 American election results even though NO AMERICAN CAN LEGALLY PARTICIPATE IN PROCESS THAT DETERMINES THE DATA. Polymarket is everywhere. Harry Enten higlights their data on CNN. Right wing influencers pump Polymarket maps showing Trump winning everyday. Nate Silver is an advisor to Polymarket. Elon and Twitter of course are pumping this stuff, hard."
https://www.hopiumchronicles.com/p/vp-harris-and-her-campaign-are-working
They are creating a Red Mirage with these manipulated polls. They use-to NOT do sooooo many. Now they do them daily. So the sites that aggregate polling data are very likely very wrong.
If you have a pollster you think is truly honest with no pressure to rig a poll or 2 for the filthy-rich oligarchs, then that is probably your best bet of getting good polling results. I'm sure the Harris campaign has them.
The aggregate sites are too easily manipulated, no matter how well they claim to weigh the polls they put in the mix. That weighing itself is put a slight bias into the polling results.
And ignore the betting sites. If you're rich, dumping money into a betting site to create an appearance of Trump winning is easier than buying pollsters.
LymphocyteLover
(6,460 posts)dweller
(24,677 posts)One particularly confusing case is in North Carolina where Lt. Governor Mark Robinsons disastrous gubernatorial campaign is sinking every race on the ticket. Robinson is trailing by 22 points overall and a staggering 41 points among women. Yet, Trump is leading by 0.4-1.2 in the averages. Its hard to imagine a Republican losing by 41 points among women while Trump is supposedly running a close race. Even without the gender gap, the idea that Robinson is down 22 points while Trump is ahead defies logic. This would be a 23% split ticket margin, which would be astonishing.
✌🏻
Doodley
(10,117 posts)Praying Vantage Data is right.
SunSeeker
(53,404 posts)It is a cynical attempt to hide Kamala's momentum and make it appear that Trump has momentum. The polling aggregators like RCP and 538 are being used in this scheme.
wolfie001
(3,400 posts)I'm hoping it's just Red Wave BS hyping they're so adept at using. Just a tidal wave of misinfo.
calimary
(83,835 posts)lees1975
(5,598 posts)They favor the red leaning polls over the neutral ones.
LauraInLA
(1,200 posts)he bucks the trend and Republican voters stick with Trump but cant bear to support Robinson.
LymphocyteLover
(6,460 posts)running stronger now than in 2016 and 2020 and it makes no fucking sense at all, given all that's happened since the 2020 election (Jan 6th, the various criminal indictments and convictions and his general mental deterioration).
The only thing I can figure is either the polls are just fundamentally FUBARed or inflation and GOP nonsense about the border scrambled a lot of people's brains.
GusBob
(7,484 posts)I understand, (kinda) this point:
" This poll says its a dead heat, so you better vote" instead of "relax this poll says we got it in the bag" The media of course prefers the horse race, blow outs are boring. But surprise blowouts? Now there's some copy!
LAS14
(14,411 posts)I wish I were technically savvy enough to read this article with ease.
Would some of you smarter DUers comment on it?
LauraInLA
(1,200 posts)Authenticated Predictive Projection Model. Like you, Id like to hear other more informed opinions .
MontanaFarmer
(710 posts)not from 1-off polls but tracking polls. To me the fact that it's a set of numbers from tracking polls should allow it to pick up the movement from independent voters pretty accurately. Certainly interesting.
Martin68
(24,379 posts)LauraInLA
(1,200 posts)So, first of all, it's not a blowout no one sees coming. The polling data clearly shows that there are enough electoral votes in play that either candidate could have an overwhelming electoral college advantage.
Their analysis about the correlation between Senate polling vs. presidential polling is kind of silly. It's like saying, "look at the national polls, because those have correlated with outcomes the vast majority of the time". many of these polls ask about the Senate and the presidential race at the same time, so if you get different numbers, it literally means a respondent is telling you they are not intending to vote along party lines. Regardless of how unusual that may or may not be, the poll is telling you that isn't following historical trends, it's because reality isn't following historical trends.
LauraInLA
(1,200 posts)GusBob
(7,484 posts)(Without doing any research aka pulling an opinion out of my backside)
It sounds to me like they are shooting to make a "splash play" prediction
wolfie001
(3,400 posts).....I would hope this is the logical outcome.
dsc
(52,550 posts)Stein is ahead by 13.8 in the average on RCP 52 to 38.2. If anything like that margin holds, the ticket splitting for a Trump win would have to be way bigger than it was in either 2016 or 2020. In both of those years we elected a Dem gov but Trump carried the state. Trump did just under 4% better than the GOP result in governor (.2% loss vs 3.5% win) and he did about 6% better in 2020 (4.5 % loss vs 1.5% win). He would require a nearly 20% differential to win, I would guess it is doable but not likely.
Ray Bruns
(4,492 posts)Farmer-Rick
(11,127 posts)That will win the race for Trump.
Not so much the "They're eating the cats, they're eating the pets." That will win it. And now he's going on about geese. Can't lose with those huge national concerns.
PatrickforB
(15,051 posts)PatrickforB
(15,051 posts)jgmiller
(437 posts)I like that they were very open in their analysis and methodolgy, there's some weird things in here though. Just on a gut feeling GA, FL and NC seem way too skewed to Harris. PA has that feel too but they are a hard one to peg down so it might be realistic yet AZ,MI and NV seem logical. Let's hope they are right though!
bucolic_frolic
(46,405 posts)Data tells us a lot about everything if we do enough analysis
Orange Buffoon
(194 posts)It shows a tremendous shift toward Kamala from September to October in all the swing states EXCEPT North Carolina, where it shows her support lessening from + 24 to +10. Why should N.C. be the mirror opposite of all the other swing states?
kerry-is-my-prez
(8,932 posts)lees1975
(5,598 posts)Seven swing states, 1,200 responses daily. That's catching any trend for sure. And their methodology of tracking voters on more than one race, and candidate, seems a lot more sound than the separate responses and sometimes no other responses, of other polls. They've provided evidence that shows ticket splitting is not anywhere near what it would have to be if the numbers the polls put out are accurate. I mean, in Arizona, for example, how could it be that a full 10% of Trump voters would not vote for Kari Lake, or that 10% of Gallego's voters would not vote at all for the top of the ticket? That would be historically different from past patterns.
Back in 2000, I subscribed to a Gallup daily tracker. It was frustrating at times, because it was just raw data, no factoring or any guesses on "voter enthusiasm," but it also turned out to be remarkably accurate. One day, it would show an 11 point Gore lead, the next day, Bush was up by 10. But at the end, they were within a single point of the final count, which was close and which resulted in a Gore popular vote victory.
Katcat
(337 posts)WV and i have never seen as many voters as I saw today. Red WV so that might be bad but Im glad they voted
shotten99
(660 posts)I have an unhealthy obsession with polling this time of year and reading something like this definitely calms my nerves a bit.
Regardless, bad polling news makes me nervous so I donate when I see it.
Good polling news makes me think my donations are working so I donate more.
PortTack
(34,279 posts)Doesnt really appear to be a blow out.
Warpy
(113,025 posts)instrumental in keeping the "horse race" going for the media. I can only assume they're reading the writing that has always been on the wall and trying to salvage their reputation at the last ,minute.