Writes Nate Silver, in "Don’t Blame Polling" (NYT).
I'm blaming polling. What's Silver's reason not to blame polling? First, Republicans are (supposedly) more suspicious and less likely to respond to polls. Pollsters using "weighting" to try to compensate. There's also "herding," which is massaging the numbers to make them more like other pollsters' numbers, but herding isn't a basis for not blaming polling, and Silver is certain that herding occurred in 2024. So what justifies that headline? I see this:
Should we trust polls less? I’ll offer a brave and qualified no, but only because the shift in public sentiment about polls — from viewing them as oracular to seeing them as fake news — has probably overcorrected relative to reality....
Blaming and not trusting are 2 different things! But that's an issue with the headline writer. Silver is talking about trust, and he's only saying don't trust polls any less that you already do. I guess it's like the way I feel about reading the mainstream news, which I do every day. I don't consider it a complete waste of time. I regard it as biased and manipulative, but the alternatives are even worse. (And this blog is not an alternative to MSM. It feeds off MSM.)
Posts
104 comments:
"Empirically, heads win 50.5 percent of the time."
So many questions come to mind about that statement. I call BS. Over a long-enough coin-tossing run -- say 1 trillion flips -- the difference between the number of heads and tails showing up had better be miniscule, like a 5th decimal point percentage.
What happened to the peephole to the propaganda? Navel gazing by outsiders isn’t really I suppose…They have enough data points to account for the shy Republican voter. They choose not to help the other side. Instead the two things pollsters will report are ‘coin toss’ or ‘Democrats lead’…on some NYT podcast Pelosi supposedly let slip their internals knew with Biden it would be a Republican blowout worse than the Kamala loss. How’s about publishing of few of those numbers next time?…one of those Bill Maher post election rants autoplayed on my computer and I heard him say America needs a viable center left party. The thought that popped into my head was ‘We just did!’ Take a good look at these people for chrissakes…
Silver's business, which is selling cope to Democrats, is dependent on polling. He really doesn't care if the polls are correct or not since his models are always going to produce a 'chance' of either candidate winning. (His rant about herding was pretty transparently a plea for organizations to publish more polls favorable to Harris.) If media organizations cut down on the number of polls they create it will have a significant impact on his livelihood so of course he's in favor of their readers and viewers continuing to demand polls.
Also, thank you for curating the MSM for us.
I'm blaming polling.
@Althouse, me too.
…trillions needs to account for the edge result or for when the coin escapes…
If this were twitter, my reply would be a GIF of Daffy Duck tap dancing, because that's all that Silver is doing here.
If they take money for their product and can’t be trusted with what they do, it seems like a fraud.
Well, except for in Iowa.
Pollster J. Ann Selzer was paid by the Des Moines Register to create a fake poll on the eve of the election predicting that Harris would beat Trump and take the state of Iowa 47% to 44%.
In reality, Trump won the state 55%-42%.
The poll made Selzer so much money, she's announced her retirement. It's unknown how much money the Register paid for the poll, but the NY Times recently reported that Town Hall producer Harpo Productions was paid $2.5 million to produce a town hall for Harris that Trump wasn't invited to.
…yes, like carefully packing up the Christmas decorations he’s preserving the propaganda benefit of polls for next time…
I don't do polls. If I have to do a poll, unless I am 100% sure of the anonymity of it I lie to the left. I don't answer ANY online poll of any substance. In fact the only polling I have actively worked to answer correctly was the recent "extended" census that my wife and I were required to submit to.
Most polls are conducted by or for organizations with agenda's and most polls are worded to support that agenda (however subtly). When liberals and/or democrats are involved, your answers to polls can have life affecting ramifications for you and your family.
I try to save MY polling for the anonymity of the ballot.
I am sure I am not the only one. Enough people feel and do as I do and your polls are worthless. Yet more spillage from liberal/dem attitudes and behaviors.
Per Google's AI, there is a slight bias to the side of the coin that was facing up when flipped being the side facing up when it lands so if you always place the coin with heads facing up, heads will come up more often.
It’s the probability models that are the problem. A probability model means never having to say you were wrong…
Three strikes and you're out. Plenty of other pollsters seem to have no problem finding Republican respondents, but those polls are often downgraded by Silver and others as 'biased'. "Weighting" is just another word for manipulation.
I agree the polling was poor again. However, why was it bad? Because tons of people don’t trust it and won’t respond to the surveys. Another reason they don't respond is they have been conditioned to keep their opinions to themselves lest they face the fury of their neighbors and colleagues.
So it turns out if you force people to go underground, you don’t have a clue what they really think, you can’t represent that in a poll, and you can’t win elections without that missing knowledge. The left’s wonderment of what went wrong is surreal - again thinking the reality is what they want it to be rather than what it is.
Republican students under attack since Trump win…I wanna know why the school shelled out for coloring books for the losers but nothing for the lids that wanted to celebrate?
Public opinion changes every day and is just an opinion open to change when given. A vote is final and involves a moment of madness corresponding to that finality, when the decision is made. Those are different circumstances.
"but the alternatives are even worse. "
There are two kinds of journalists, the ones who try their best to get things right, and the ones trying to manipulate the perceptions of the moment, and figure that for their audience, reality is like an Etch-a-Sketch, and don't worry that next week, or next month, people paying close attention will have them pegged as a liar.
That Iowa poll was a perfect example of a journalist whose main object was to manipulate the perception of the present. She actually did create a moment where even the betting lines started to move towards Harris.
Silver fancies himself one of the good guys, but reading him you know that he is like that tailor, or whatever he was in The Godfather, who asked a favor, and promised a service, if that day should ever come. Well, the day comes for Silver *a lot* and he knows that he will lose his cushy gig if he doesn't pull on his oar to the beat of the DNC drum when it matters most.
Y’all remember this when the GOP Congress says ‘we took a show of hands in the cloak room and we didn’t have the votes so we’re not voting’. Not how it’s supposed to work…
"The poll made Selzer so much money...."
I don't know if that's true, but I thought Silver's analysis should have covered the topic of making money, including for himself. The polls still have value because pollsters get paid, and if we the consumers of polls turn away, that value will change. He's self-interested, instructing us not to keep consuming. It's the best we've got, he assures us. Submit to the efforts at manipulating OUR minds.
After 3 misses in a row going in the same direction for the same candidate with different outcomes in each race, I think it is time to retire the 'Republicans/Trump voters don't respond to polls' excuse. All polls have extremely low response rates. Clearly there are polls like Rasmussen and Trafalgar that have always been able to find people willing to respond favorably when asked if they support Trump. The difference between those polls and others more favorable to Democrats is the assumptions they make about how the raw data they gather from surveys is translated via the demographics they are assuming about the population. It's hard to escape the conclusion that polls consistently favoring Democrats are using demographic assumptions (I suspect particularly party identification but quite possibly ethnographic data) that don't match the actual makeup of the country.
What I meant to say with this comment is that picking a "guide" to help with the interpretation of world events, for example, requires one to separate the manipulators from the "trying to be honest" reporters. You just have to pay attention to who gets stuff right consistently, as judged by seeing how closely events played out as they reported them at the time, and who thinks of his audience as if they had the mind of an Etch-a-Sketch, and as long as he gives them the emotions that they are looking for, they won't bother checking to see if they have been lied to, in fact, this type of journalist knows, and banks on, the fact that his audience will actively avoid any evidence that they have been lied to.
Russiagate proves this. FOIA showed that the FBI knew within a week that the "dossier," which was the basis of the Mueller "investigation" was made up by the Clinton campaign out of whole cloth, to counter the emails thing. Handwritten note by the head of the CIA shows that they knew that Hillary's people had made it all up within a day.
Yet still people here believe it, and call those of us who follow the actual evidence over time, "reprobates." Well, that's what gadfly called us in his parting shot, anyway.
"Submit to the efforts at manipulating OUR minds."
The bias goes against conservatives, and so pollsters have more trouble getting conservatives to take the poll. They keep trying to "weight" the poll results to compensate for this distortion, but the distortion is something they caused, it seems, by being biased against conservatives. It keeps getting worse, so they can never weight enough. They can't guess how quickly and asymmetrically the trust is collapsing.
I mean one reason why i read this blog is just to know what’s happening in the world of the mainstream, and the articles chosen to blog are usually pretty good indicator of what’s going on.
“This blog is not an alternative to MSM”
AA doing the work so we don’t have to (good or bad, it certainly seems like a thing).
blame "polling"
remember The Joy? back when they anointed Kammy?
SUDDENLY, she was up 4 points.. JUST LIKE THAT..
then, the "pollsters" casually admitted that:
because of "democrat enthusiasm", they were weighting the Kammy vote.. by 4 points
Remember?
They tilted their numbers by enhancing the Kammy vote.. Because of "democrat enthusiasm"
Of Course.. The ONLY 'enthusiasm' was in the "pollsters"
now.. remember the IOWA POLL? that showed Kammy WINNING iowa?
the poll where they Refused to admit, that they'd ONLY polled democrats, and that in reality, Kammy wasn't Even going to get a majority of iowa registered democrats?
remember
“the distortion is something they caused”, along with MSM.
More and more people recognized all the media gaslighting and lying this time around. Pollsters were both perpetrators as well as victims of it. Lose-lose.
The first time I heard the phrase "Shy Trump Voters" was 2016 and the estimate was 10%. 2020 the most common estimate was 5%. So I confidently figured 2.5% for 2024. Bingo!
"why was it bad?"
Very simply most public polls are bought by media companies. They don't get advertising if the actual poll results were released. Why should Trump by ads when he's leading 55-42 in Iowa? The Des Moines Register needs the race to be close in order to trick both sides into buying advertising from the Des Moines Register.
So ... fake poll.
It's literally that simple and for the record, I have worked for and owned newspapers. I know exactly how they work, why they do what they do, and who is making the money and how they are doing it.
And because of the 1st Amendment, there's nothing society can do about this. They are legally free to print just about whatever they want - yes even fake news - so long as they don't libel someone.
Most polls are in fact weighted towards Democrats. By that, I mean they deliberately slant the poll by asking more Democrats than Republicans who they are voting for. They claim all kinds of "scientific" reasons for doing this, but we all know the polls aren't "scientific" or they'd occasionally be correct. And they NEVER are.
I say most because nowadays, the populace is dumbed down enough that a pollster can get away with not releasing their internals (the way the poll is rigged for one side or the other) and people will still believe in them.
Remember ... in the 2008 Financial Crisis ... the only way that the banks could sell crap mortgage securities was to get the ratings agencies to give them AAA ratings. So they paid them to do that - regardless of the truth of it.
Same with pollsters.
Who is getting the $$$$$$. That is all you need to know.
If you make the candidate out to be Hitler, then ask people if they're voting for Hitler, you pretty much guarantee a Shy Trump Voter.
“Why should Trump by ads when he's leading 55-42 in Iowa?”
Great point.
(And this blog is not an alternative to MSM. It feeds off MSM.)
Good eatin', too.
Yes, the Republicans are center left in the traditional sense of the word, well since the Trump takeover. The Democrats could be enjoying a massive victory right now, if they had just gone with Trump, and ignored the donors, whose money meant less than nothing, and rebuffed the neocon takeover of Democratic foreign policy think, and yes, Trump has his billionaire buddy, but the vast majority of the billionaires went with the Democrats, for reasons of economic self interest and protection from the kind of lawfare that Democrats wield at the drop of a hat.
A model is just a washing machine for one's assumptions, once you get outside of closed systems in the physical world, anyway.
Rasmussen smiles.
What about 2022? Did anyone predict swing states not going Red?
Empty calories tho.
If you blame the polling then that means you believe the election is on the up and up. Maybe the problem was the Democrats didn't steal enough votes that's upholsterers had included in one of their fudge factors.
Math is Hard
One thing I noticed about the polling this cycle was that, with few exceptions, it was hard if not impossible to get to the cross tabs. Makes me think they were trying to hide something. Like the pro Harris bias in most of their polling.
I think Dixcus is correct but for a slightly different reason which is why I am insisting that "Republicans/Trump voters don't respond to polls" is a bogus excuse for what's happening now that we have multiple iterations of the same error.
Let's say you are polling an extremely blue area with a closed primary. You get responses from 6 self-identified Democrats and 10 self-identified Republicans. However, if you look at the voter registration data for that area you see 9 Democrats registered and only 5 Republicans because a portion of the Republican identifiers actually register as Democrats so they can vote in the Democrat primary which is the real election. Now you have a conundrum. You know that your survey is unlikely to have just randomly selected the exact mix of voters in each party so how do you go about adjusting the responses? Do you assume the registration data is correct? If not, how do you adjust it and by what proportion, and then apply that to your survey data? In almost any case you are likely going to wind up overweighting the Democrat responses unless you ignore party identification completely.
The same thing could happen with Hispanic identification. Many people who would be counted by current government guidelines as Hispanic often self-identify as white. For the area you surveyed, what do you use as the 'correct' proportion of Hispanics, and do the people who could identify as Hispanic who self-identify as white have consistently different political leanings than people who don't?
The real fakes are opinion polls - not election polls. take all the statistical problems inherent in a rather simple "Are you voting for A or B" and multiply it by x10 and you have opinion polls. Voting polls can use history to "Weight" their responses. Opinion polls ususally don't.
Btw. the reason you have these fake polls exaggerating the lead of Democrat candidates is because people are stupid. They don't want to vote for "Losers".
The entire question is ludicrous. It is analagous to asking: "Should we trust the media."
Now I ask you ... how many instances of media manipulation of the news - printing of overtly fake news - has to occur before we stop even asking that question?
The media do polling. Everything ELSE the media does is corrupt and false, including the sports, and the weather.
This was from David Axelrod in 2016
Axelrod told Ingraham that the emails were unlikely to disqualify Clinton with voters and she was in strong position to win the presidency. "I can't recall any candidate who had a lead like this this close to the election and didn't win," he said. Recent polls show the race even tightening in deep red Texas, but Axelrod doesn't expect Clinton to capture the Lone Star state. - Buzzfeed
One of the leaked emails was from Axelrod to the media telling them how to shade their polls to help Hillary. For instance, where to get working class whites in Florida who actually leaned towards Hillary, the Orlando area. So basically his lips are moving, and he is feeding Buzzfeed lies. This was Biden's real gift, that he could do this so effectively, say the opposite of what he was thinking. Taking that into account, his gaffes were pretty minimal.
…those ‘right track/wrong direction’ questions are quite manipulative. They are summing agreement of people at opposite ends of the political spectrum, then the media uses it to support the left’s position, either consensus to throw out Republicans or a call for their own party to do more or be more radical. It’s a scam…
I like how he became a guru for predicting that Obama would beat McCain. As if even casual observers didn't see that coming. Has he ever predicted a republican presidential victory? Ever? 🤔 I question the man's methods.
“I'm blaming polling.”
What exactly are you blaming polling for?
I don’t see how you can blame polling for failing to predict the results of the election when you yourself used polling to correctly predict the outcome (although you had the “over” and Trump didn’t beat the point spread).
I don’t see how you can blame polling for Trump winning and Harris losing the election (although labeling it blame would be a tell as to who you really wanted to win).
I read a story about the French "whale", the reclusive financier who bet and won millions of dollars by betting that Trump would win the presidency: Early speculation was that he was trying to manipulate the betting markets, but it seems he commissioned his own polling and based his bets on that. This poll was by association. They would ask people who they thought their neighbors would be voting for. People may be shy about admitting who they personally like, but have no reservations about guessing who the people across the street will vote for.
Well nate if you keep getting it wrong
What Rasmussen does right is automate their polls, so you just punch a number on the phone and you don't have to say Tr*mp's name out loud.
Moreover, they're short and to the point. Most the polls use a live caller asking long and tedious questions.
It's a real good thing if you enjoy countless posts about Nate Silver. But does anyone? Tedium in a cup, IMO.
90% of all statistics are made up anyway.
"I don't have any (other) marketable skills. Please don't take polling money away from me".
- Probable Nate (within the margin of error) Silver
Took the words right out of my mouth. We’re talking about SERIOUS money.
MattTaibbi's take.
The corrupt left are horrible lying liars who lie. Lying is the left's charter.
Polling is wrong, but don't blame the Pollsters? Who should get the blame? Voters? Nate Silver is trying to secure his relevance.
from Matt T's link above:
"Every major aggregate, that is, but RCP. McIntyre’s site was removed on October 11th, after Wikipedia editors decided it had a “strong Republican bias” that made it “suspect,” ven though it didn’t conduct any polls itself, merely listing surveys and averaging them. One editor snootily insisted, “Pollsters should have a pretty spotless reputation. I say leave them out.” After last week’s election, when RCP for the third presidential cycle in a row proved among the most accurate of the averages, Wikipedia quietly restored RCP. "
RCP doesn't do any polling - but the corrupt hack delusional Soviet left had to cancel RCP.
wiki has proven itself to be complete garbage.
I thought the bias is in weight distribution.. If you slice a coin vertically, the halves do not weigh the same due to the difference in engravings. Since the details on Tails are traditionally more ornate, that side is a small fraction heavier hence minutely more likely to wind up face down.
That's how an engineer explained to me.
Conservatives naturally lean to practices that mitigate authoritarian progress.
He became a guru also because of his flashy marketing of himself as a data "scientist" to an impressionable audience.
I think the betting markets are proving to be a superior indicator, not surprisingly. Like Silver, the bettors look at all the polling data, but they have their own money on the line, and therefore adjust for the biases. That's why they had Trump with a 60-40 likelihood of winning while Silver was stuck at 50-50.
The fundamental (and unfixable) problem with polling is the inherent inaccuracy of sampling. No matter how scientifically and rigorously they try to sample, there's no way that the typical hundreds or thousands of people in their samples will accurately reflect millions or tens of millions who end up actually voting.
I don't even answer my phone if I don't know who is calling. If they leave a message and are a party I want to respond to, I call them back. I don't even trust the census and only answer the minimum, and I wouldn't respond to their extended version.
Public polling has always been used to shape rather than reflect opinion. That is why most media will note that some pollsters are Republican or Democrat. If they were all just doing accurate public opinion you would not be able to identify how they lean politically. On the TwoWay and Real Clear Politics podcasts they openly discuss how the candidates private polls are much better. Why do the candidates pay for private polling - because they know the public polls are bs.
"Their internals must be really terrible."
We hear this over and over. Is there a reason why we should think that various campaigns' internal polling always hold the ultimate truth of the matter - truth that the public polls just don't for some reason have the wherewithal to capture? You know, a reason to think that the public pollsters just can't figure out how the internal pollsters always achieves their mythically more accurate results? Do we ever get the opportunity to see the internal polling results after the election to learn whether or not our suspicions were true? Hell, (slightly tongue-in-cheek) are "internals" even actually a thing?
Polling should be illegal.
Just as it is illegal to "electioneer" within 100 feet of a polling place, it should be illegal for a person to ask a member of the public who he or she will vote for.
While it makes sense for politicians to commission honest polls to determine where resources should be directed, in real life the polls that are released (not the "internal polls" that we are told about but never get to see) are 100% about manipulating public opinion. They are never described as such by their conductors; they are the quintessential definition of "misinformation."
Let candidates state their positions, let the voters learn these positions and cast their votes.
Stop the misinformation, stop the manipulation.
What makes you think they're not trusted to provide an outcome?
Polling is sold as a device measuring the sentiments of the pollee's, and that is where the flaw is. The big Party Pollers don't strive to measure voting sentiment, they strive to create it, and are paid for just that.
Ann Selzer didn't retire because her Iowa poll was so far off, she retired because she has squandered her credibility in the marketplace, and will no longer be able to trade on her reputation. It's not embarrassment, IMO - this kind of person is immune to feeling such pressures. Somebody called in a huge marker, to make her do something so stupidly self-destructive.
I see that Trump has now had his lead eroded down to less than 50%, while the votes still rack up. Interesting, that.
Pollsters have agendas- it is that simple. When the pollsters were showing Biden losing in a rout to Trump in the national and state level polling in the late Winter and early Spring, I warned everyone that you couldn't trust that result because a lot of Democrats didn't want Biden running again because he had become toxic to down ticket races. I don't think the state of the race in regards to voters had changed much at all since the two New York state trials involving Trump issued their verdicts. I think after June 99.5% of the voters had decided how they were going to vote and the outcome always rested on how well the two sides got their supporters to the polls. Trump's team did significantly better than I had thought they would do in embracing and encouraging their voters to vote early in person or by mail. If the Republican Party can continue to eat into the Democrats advantage in absentee ballots, I suspect the Democrats will start advocating for in-person voting. Maybe in a few elections when turnout hits 120% on both sides we can agree that absentee balloting is infested with fraud.
There is no verification of poll results about candidates until the election, and maybe not even then in parts of PA. So the polls really are worth exactly nothing until the actual votes are counted, at which time the pollas are worth exactly nothing, except to pollsters, who got paid for creating them.
Did you hear the story about the pollster that got it right? Polymarket got the numbers right. You'd think that would get some lauding by the media. What it got them was an early morning raid by the FBI, seizing all of their electronics.
They can correct for conservatives who won't respond. What they're missing are outward Dems who are lying.
"And this blog is not an alternative to MSM. It feeds off MSM."
You discount the comment section! I learn a lot here, for which I am grateful.
Inga weeps.
I still want to know what Silver thinks he's doing. 80,000 simulations? What is he simulating? I think he's trying to pretend he's doing Monte Carlo simulations, to make it seem scientific, but I don't see how Monte Carlo analysis is applicable to this problem. What are his random variables?
The betting markets were more accurate
Flop sweat from Nate Silver. Who was supposed to be better.
The media takes its marching orders from media departments, poly-sci, legal, and “fact-checking” institutions embedded in academia. Let’s talk about that once in a while. No offense. But you must have seen this. Why stay quiet?
Who makes you take polls?
The People's Pundit, Rich Baris, was pretty much spot on, too. Polymarket is a betting site whose 20-something CEO was just raided by the FBI to seize his electronic devices. The "B" apparently stands for 'Beria'; no charges have been filed, no accusations leveled - owing to the demands for secrecy in an 'ongoing investigation' (although I don't think this has been admitted, either).
Dixus, I’m very sure I know who you are now. Why not just come out?
Thanks Aggie.
Silver is quoting a study by Diaconis. The study did not find a "heads" bias, it found a dynamic bias: the side starting face up tended to be the side that landed face up, at a rate of about 50.8%.
Also, weight imbalances don't matter if you catch the flipped coin.
That is always a better than average poll question. Smart whale. Even if he his French.
…you’re not obligated to drink the whole thing.
Meanwhile, Trump now below 50% of popular vote as counting continues.
"But it’s no mystery. The polling problem in America looks like good-old-fashioned lying, mixed with dollops of censorship and manipulation:"
Wow, this is a really interesting read. It discusses in detail the argument Inga was feeding us about RCP being a Republican site because they didn't weight polls and the NYT being the go to, authoritative site because they put their smart thumbs on the scale.
"Every major aggregate, that is, but RCP. McIntyre’s site was removed on October 11th, after Wikipedia editors decided it had a “strong Republican bias” that made it “suspect,” even though it didn’t conduct any polls itself, merely listing surveys and averaging them. One editor snootily insisted, “Pollsters should have a pretty spotless reputation. I say leave them out.” After last week’s election, when RCP for the third presidential cycle in a row proved among the most accurate of the averages, Wikipedia quietly restored RCP.
“They just whitewashed us away three weeks before the election because we were a point or a point and a half more favorable to Trump, which as it turns out still underestimated him,” McIntyre said."
What is your fixation with 50%?
It puts the lie to Trump's claim of unprecedented mandate.
Every presidential winner claims they have a mandate. Big whoop.
I mean, if you're going to get your panties in a wad over Trump's hyperbole, the next four years are going to be hell. Let it go.
It's not like Biden has been a fount of veracity.
One way to look at it. Kinda like every winner says they will appoint the best people to their cabinet....
Well, now, that's in the eye of the beholder, init?
Didn't Wikipedia deliberately exclude RCP just before the election? Why should we trust the polling industry, when Wiki, like other liberals, deliberately hides 'misinformation'?
Yes, read the article I just linked to at 1:41pm.
I actually wasn't aware that Wikipedia even "did" polling. Can you think of a less reliable source? Hell, I'd trust the NYT before I'd trust Wiipedia.
You switching to polling?
It’s not great that the polls missed low on Mr. Trump for the third and final time, even in a year when survey companies adopted all sorts of novel strategies to avoid this exact outcome."
1: The Pravda coming out before the election was "pollsters never get it wrong 3x in a row because they over-compensate to fix the previous problem.
So that was a lie
2: Pollsters had ways to correct for the problem, but they didn't use them
A: The Selzer poll asked their poll respondents how they voted in 2020. The result was Trump up by 1. Trump won in IA in 2020 by over 8 pts. This should have told them they were off by at least 7 pts. After which adjustment Trump would be winning IA in teh poll.
She didn't do that
Her poll was crap, not because it was an "outlier", but because she chose to produce a dishonest poll
B: It would have been easy to produce an honest poll in NV.
1: Use the latest voter data for your poll population
2: Make sure at least 50% of the people you try to contact are people recorded as having already voted
3: Report your results twice, once as you would do w/ normal polling principles, once where you weight and adjust based on the known demographics of the early voting voters.
The Clark voter file was good for age, number of times voted in the past, and declared Party. Only about 10% of the records had a sex.
But if they'd weighted on that and geographic location, they could have reported how ind voters were breaking, and what teh crossover votes were.
The reason why neither of these obvious adjustments were made was because they would make Trump look better.
This tells us the reason for the polling failure was deliberate choice
Should we trust polls less? I’ll offer a brave and qualified no, but only because the shift in public sentiment about polls — from viewing them as oracular to seeing them as fake news — has probably overcorrected relative to reality....
No, it hasn't. IA and NV show that
I regard it as biased and manipulative, but the alternatives are even worse. (And this blog is not an alternative to MSM. It feeds off MSM.)
Which alternatives are worse than being lied to by people who are working to block all access to the primary sources that would let you know they are lying?
"Mr. Trump beat his polling numbers by about 2.5 points nationally... and 2.1 points in the average swing state."
This tells us the upper limit on the value of the Dem "get out the vote" organization: 0.4% of turnout
Neither Party believed the poll results, as can be see by the fact that neither side upped their presence in IA
So if that was the goal, they failed
Howard, according to the polling the election was close enough that teh Dems could steal it
Happily, it wasn't, at least not at the Presidential level
NV and WI Senate were both definitely stolen by the Dems
Point is, if elections are a "coin flip," Silver's model *always* picks "heads." And, when it is heads he is a genius or something.
Meanwhile, Trump now below 50% of popular vote as counting continues.
So Harris will take office in January? No?
Did Hillary Clinton? Her husband never made it to 50 per cent either. Missing the point.
That counting occurring in California, where voters were not ID'd, so anyone could vote.
Keep digging for dat pony, Readering!
Whenever we say that a decision is a "coin-flip", we mean it's equally likely to come out either way (we even use the term "coin-flip" when there's more than two possible outcomes -- I suppose we imagine a coin with more than two sides). I wouldn't bet real money on the outcome of the flip, so I'm happy to treat it as 50-50.
The MAIN POINT is: This is the first close presidential election that I recall in which pollsters, or commenters about polls, actually acknowledged that the polls didn't really say anything meaningful about the outcome. Of course that happened ONLY at the very end of the campaign -- and I assume after the pollsters had cashed their checks.
I wonder if we'll remember that in four years, in the run-up to the Vance-Newsome election. 60 days before election day, the polls will have Newsome at 53% and Vance at 46%. And Vance will win. Save this, and if I'm wrong, let me know.
Post a Comment