Pages

Friday, October 23, 2020

The end of the political poll

Another reason to anticipate a Trump victory: when the President wins again, we'll never have to pay any more attention to the pollsters:

Should Trump once again dispatch his Democratic challenger, the polling industry is finished, Republican pollster Frank Luntz told Fox News on Thursday. 

“Well, I hate to acknowledge it, because that’s my industry,” he told Fox anchor Bret Baier. “But the public will have no faith. No confidence. If Donald Trump surprises people… my profession is done.”

Luntz insists that his polling is accurate this time, and that Biden will win. However, undecided voters may be leaning toward Trump. As the two men faced off in the final presidential debate in Tennessee on Thursday night, Luntz organized a focus group of undecided voters. After the showdown, a majority of these voters were leaning toward backing Trump. They described him as “controlled,” “poised,” and “surprisingly presidential,” while Biden was thought of as “vague,” “elusive,” and “defensive.” 

If a Biden-friendly pollster, backed by his latest focus group, is publicly opening the door to a potential Trump victory, the election gurus may not be as confident in their figures as they let on.

FiveThirtyEight’s Nate Silver is another one of these gurus, and his predictions are taken seriously in Washington. Right now, Silver gives Biden an 87 percent chance of winning. Yet the nation’s premier pollster has also been hedging his bets. On the same day that he bumped Biden’s chances up to 87 percent, Silver took to Twitter to remind his followers that “Trump still does have a nontrivial chance.”

“There’s still some time left,” he continued. “Tipping-point state polls are closer than national polls, sometimes polls are wrong… and mail voting and court disputes create some additional uncertainties.”

The Economist’s G. Elliot Morris is similarly all-in on Biden, giving the Democrat a 92 percent chance of winning on November 3. Morris is confident in his prediction, overly so according to Silver. Yet even though Morris has declared the race “very probably” over for Trump, he couldn’t help but write an article explaining exactly how he may be wrong.

I expect both Luntz and Morris know perfectly well what is going to happen on November 3. Silver is the only one who is dumb enough to insist that his model is right and it is reality that has it wrong. Watch as the polls "unexpectedly" begin to "tighten" between now and then.

76 comments:

  1. “ a majority of these voters were leaning toward backing Trump.“ Bullshit. I watched the Luntz focus group stream this morning. 11 raised their hands for Trump, the other one couldn’t pull the trigger between voting for Trump or not voting at all. The women all said Trump reassured them that he can put aside the larger-than-life act and act Presidential if he chooses to.
    Not one of them put their hand up for Biden.

    ReplyDelete
  2. Should Trump once again dispatch his Democratic challenger, the polling industry is finished, Republican pollster Frank Luntz told Fox News on Thursday.

    The death of the poll zombies? Nah, that would be too much to hope for.

    ReplyDelete
  3. The rap star Wacka Flaka Flame is backing Trump. He said that Trump is a better POTUS then Obama!

    ReplyDelete
  4. Wasn’t Frank Luntz implicated in the emails along with Paul Ryan for basically trying to buy poll numbers to sway opinions? The profession is already done now that people are becoming low-trust

    ReplyDelete
  5. Note to Frank Luntz: Your profession was done in 2016. People are slow to catch on sometimes.

    ReplyDelete
  6. I can’t imagine answering a random phone call from out of state, and telling them my name and address, that I’m a nazi, a racist, and I want grandma to die. We’ve all seen flash mobs of violent leftists showing up at people’s houses, and how their mayors and governors are handling it.

    Even if they wanted to conduct a good poll, they can’t.

    ReplyDelete
    Replies
    1. That's one of the best observations on the state of polling and data privacy to date.
      People are going to be a lot more circumspect about what they really think on subjects from here on.
      Your job life property are all potentially at risk.

      Delete
  7. I am expecting that on November 3 around 22:00 EST it will be declared that "unexpectedly" President Trump has won!

    ReplyDelete
  8. Just as many indistries that flourished in the last century have become irrelevant, political polling using the same methods of yesteryear is doomed. I started my online business 1996 (dial up anyone?) and have had to adapt to a non-stop ever changing landscape that moves faster and faster to obsolesce the advertising, web design, and sales metrix tracking methods I use today. I should probably go into the polling business because we have next generation, proprietary methods of discovering what people want and tracking them to make sure our data is correct. Numbers don't lie, the methods to collect them is usually flawed. The only political pollster who comes close to our methodology is the Trafalagr group, and they would considered a dinasaur in our business circles.

    ReplyDelete
  9. The latest MN Senate poll has the Dem Smith only +1. KSTP/SurveyUSA: Smith, Lewis Senate race now a dead heat

    Trump can win the state if that poll is accurate.

    ReplyDelete
  10. They will be given outs as the election approaches. "We could have never factored in that information!". Less Americans will pay attention to them, just as less Americans pay attention to them after each election, but they aren't going anywhere.

    ReplyDelete
    Replies
    1. @jeroth. Their only usefulness now is propaganda. Emotional factors.

      Delete
  11. Vox and others:

    What levels of voter fraud and voting machine hacking and other shenanigans can we expect? Does Team Trump have a plan and tactics and strategy in place to prevent fraud, and to litigate provable fraud that is successful? What kind of post-election battle can we expect? Will there be violence and a campaign of destabilization?

    I've gamed it out in my mind, and I have no insights into what to expect. I do expect the Deep State to go down fighting pretty viciously.

    ReplyDelete
    Replies
    1. The fraud is already happening. Videos surfaced on the chans of people writing on votes during the count. And their plan is already known. They plan to start litigation on counties with Dem judges, refuse to concede, and delay until it's left (pun intended) for the courts to decide the victor, which will be given to Biden.

      Delete
  12. OT: Was looking at the Book List 2020 on the right hand margin and I saw the entry "Blindsight, Peter Hamilton." It was indeed an interesting read, but the author is Peter Watts.

    ReplyDelete
  13. I won't pretend to be a statistical genius, but I can't understand why the pollsters insert themselves into the polls by gatekeeping how many "Democrats", "Republicans", and "Undecideds/Independents" there should be in a poll? Particularly in an era where nice, neat lines of political affiliation are hopelessly blurred. Wouldn't just sampling 1000 random Americans tell you more? But then, I errantly make the assumption the pollsters are trying to give honest information...

    ReplyDelete
  14. When Obama showed up in Philly earlier this week for a car rally for Joe, it was a disaster. A wide angle shot showed only a handful of spaced apart cars in a huge empty parking lot. A very large white tent was set up next to the stand for Obama, but there were hardly any people standing around.

    Where are the hard-core democrats? They aren't showing up. They've thrown in the towel weeks ago. And remember, Trump said last night during the debate, "we may even win the House". I've got a feeling that this will be an historic loss outdoing Reagan/Mondull in 1984.

    ReplyDelete
  15. Nate Silver is saying "Biden's going to win, prepare the guillotines my non-white minions!" as he low-key sells off his immobile assets and prepares to make aliyah. Meanwhile, Luntz is planning a "I'm Republican like you, fellow white civil nationalists!" I don't think it's going to work out very well.

    ReplyDelete
  16. Undecided voters matter, but the real problem, as usual, is undersampling the white working class. Pollsters talk to one spinster librarian living out in the sticks and go "welp, that's rural Pennsylvania covered". They won't even look at voter registration as a sanity check on their data.

    Some will say that it's on purpose, and I can't say I entirely disagree. But accidental or intentional, he's right - the polling industry is done. Fair elections are also done, so it's kind of a moot point.

    ReplyDelete
  17. It's not a conspiracy theory. It's not even smart Luntz. The reasons polls are the way they are for at min 3 decades now is obvious, but not clarified as usual in the media for the masses to understand for obvious reasons.

    The 3 main reasons polls are way off are:
    1) Demoralize the enemy and give comfort to your own. "Oh but there are some right leanibg polls..." be serious

    2)It provides cover for any potential voter fraud they might be caught engaging in. " we was winnin anywayz"...

    3) if they lose, they can say, without saying "it was rigged" or "people were confused at the last minute" or "racists came out" deligitimizing the opposition from the off.

    They don't care that they are shamans with rocks. News outlets and pols will pay them even for wrong data.

    For 15 years I was an exit pollster, then regional supervisor. And by supervisor I mean election day I drove around naking sure the exit poll people like me were present and not sleeping or drunk. Seriously. What you learn is despite the magic of statistics, older people and men of the right don't like talking to strangers about their personal sh7t and who they voted for. Especially to some 25 year old hipster with a cardboard box and ipad posted by a door outside the voting area.

    But you know who loves to talk to pollsters??? Liberal women. Especially 50 plus. You can account all you like with misses and refusals to offset the bias, but you ll always be a bit off.

    But this is known from pre Trump and pre cancel culture. Now it is even more pronounced. 25 years ago even "liberal" was a word most avoided. Now communists and black racists say anything openly. Those who say they like America, or God, or simply basic republican stuff are quiet in front of others. Almost shamefully so.

    Tl;dr Basically with the popular vote glut from the west coast being irrelevant on electoral college voting, any poll that isn't showing Biden miles better than hillary was at this time in 16 are basically telling you Trump is going to win easily.

    ReplyDelete
  18. Luntz actually found some inconsistency in his methodology (and probably in the industry, not only his). And if Trump wins the industry will be challenged not because of its faults but because of Trump's triumph. Silly, stupid.

    ReplyDelete
  19. Their demise was inevitable once they shifted their focus from predicting how people will vote to influencing how people will vote.

    There may still be push polls which nothing but campaign ads anyway and there will still be polls candidates pay for to determine which issues to focus on (i.e. what does the electorate want?), but even those will be limited by the public's distrust of polls and/or desire to mislead pollsters.

    Pollsters will likely be replaced by computer programs that try to predict votes based on whatever data big tech collects. Of course, that will be subject to the human foilbles and failings that have killed polling (programs don't write themselves, after all), but it may have a good run. In fact, I am surprised we have not seen such predictions already such as the "Google election algorithym" or the "Amazon voter index" or somesuch.

    ReplyDelete
    Replies
    1. I'm sure Gulag tried AI polling, but when it kept returning with the "wrong" results the cord was quietly pulled out, and Skynet was aborted once more.

      Delete
  20. I really felt Trump did well. He learned from the disastrous style of his first debate - was calm, direct, and presidential. And he was very effective in rebutting Biden's claims. I only wish Trump had acted like this during the first debate.

    ReplyDelete
    Replies
    1. "I really felt Trump did well. He learned from the disastrous style of his first debate - was calm, direct, and presidential. And he was very effective in rebutting Biden's claims. I only wish Trump had acted like this during the first debate."

      Biden was essentially a cyborg in the first debate and had been rehearsing his answers for weeks. The constant chattering in Joe's ear, which I'm sure Trump noticed, prevented Biden from being able to respond well. Then there was the Teleprompter contacts he was wearing. Distracting ol' Joe while he was reading his script was the best way to throw Biden off. That's why Trump talked out of turn and interrupted him. This is also why they muted the mics last night. It still didn't help Biden, since his adderall drip was running out halfway through. Trump insisting on a 90 minute debate was genius. Biden's batteries go dead after 45 minutes.

      Delete
    2. I thought you were joking or fantasizing, but then I found out about "MOJO lens" they are quoted as saying that their first priority is to help those who need it most, which is of course Joe Biden. "https://singularityhub.com/2020/01/17/mojo-visions-augmented-reality-contact-lenses-kick-off-a-race-to-ar-in-your-eye/

      Delete
  21. Yeah in the "Cancelled" era people just love to give straight answers to strangers on the phone, sure. Now if the rhetoric of Cultural Marxism was cancelled then the only poll question worth asking would be, "What should the bag limit on Leftists be?"

    ReplyDelete
  22. Wait. Are you guys saying that Frank (((Luntz))) might be deceitful and duplicitous?

    I'm gonna go drink myself into a stupor

    ReplyDelete
  23. “But the public will have no faith. No confidence. If Donald Trump surprises people… my profession is done.”

    The public lost faith in pollsters a long time ago. To most right-wing folks, "pollsters" and the MSM are one and the same. Even if pollsters hadn't been so obviously wrong for so long, I think The Right would've discounted them for mental health reasons alone.

    Perhaps what Luntz really means is, the Left (and their cheerleaders in the MSM) won't believe their crap any more. A Trump victory doesn't represent a failure to predict the results so much as a failure to influence them.

    ReplyDelete
    Replies
    1. @Jack Amok- Even when the posters turn out to be accurate, or more accurate than we thought, it doesn’t slow down our ever-growing distrust.

      I mean, I was surprised that Romney did worse than McCain, given the anti-Obama backlash in 2010. Not sure what happened over the course of two years. Too many grew too comfortable with Obama, I guess, and Mittens didn’t try very hard (unless he was trying not to try, if you know what I mean).

      Still, come 2016 I never once considered the pollsters might be right because they were ahead of me on Romney. No. I just assumed they were wrong again.

      Delete
    2. Tublecane romney did worse because of the incumbent bump that all the polls seem to neglect. And he sucked and was the obvious can't win patsy don't ruin jeb in 2016 guy.

      When i was an exit pollster you always heard, "i don't like the guy but i just don't know the other dude well and it's not burning to the ground. So I just voted for xyz who was the incumbent.."

      Unless we are truly massively fukked demographically, a challenger has to overcome the incumbent bias with extreme popularity and optimism. Do you see anyone thrilked to vote for Joe? You can't win just running as I'm not him without gross incompetence and we all know that aint the GE

      Delete
  24. When the polls have Joe Biden ahead by double digits for weeks, that's a good indication that they aren't accurate. On what planet does any candidate for President ever have a double digit lead over his opponent in reality?

    ReplyDelete
    Replies
    1. In a Reagan decade,...but i take your point

      Delete
    2. Trump does now. They all know it too. The plan isn't to win, but to convince their minions that Trump cheated. Paid rioters are only going to show up if they are sure they won't do any real time for the damage they're doing. Look for a similar sweep of rioters like what happened at Lafayette Park a couple of months ago. Their phones were jammed just as they were being mass arrested. Very cool.

      Delete
  25. Their modeling methods are so dumb that for a long time I thought I must be missing something. If state-by-state polls exist, then national polls should be irrelevant, since the popular vote is irrelevant. Just pick a winner in each state based on its polls, add up the electoral votes, and there's your prediction.

    But they don't do that. They factor the national polls into the model, so if Biden's national number goes up because he gained in California, he gains in Michigan too. That's idiotic. I've heard claims that they have to do that because there isn't enough state polling to be accurate. If that's the case, then stop wasting money on national polling and spend it on more state polling. There are millions of dollars spent on polling. It shouldn't cost much more to poll X% of each state's population than it does to poll X% of the country. Which means they don't do it that way because they don't want to. It wouldn't give them enough flexibility to make up their own narratives.

    ReplyDelete
  26. I won't pretend to be a statistical genius, but I can't understand why the pollsters insert themselves into the polls by gatekeeping how many "Democrats", "Republicans", and "Undecideds/Independents" there should be in a poll?

    The surface answer: it saves money, because you can poll fewer people and "correct" the numbers based on control questions. If you know a particular population has 50% gun owners, for instance, but your sample has 70% gun owners, you probably just got a biased sample, like flipping a coin 10 times and getting heads 7 times, so you weight the score accordingly. If you do it right, that can make a smaller sample as accurate as if you polled a whole lot more people and stuck to the raw results.

    But one problem is where the weighting assumptions come from, which is usually either the past or their ass. They *can't* weight for something that hasn't happened before. So if a bunch of voters somewhere like Michigan are buying guns for the first time, the pollsters will keep under-weighting them.

    One way to avoid falling for that would be to pay attention to whether the sample keeps having the same bias. If you keep flipping that coin and getting 70% heads, at some point you realize your coin is unbalanced. The pollsters don't seem capable of doing that, so they just keep correcting away anything they aren't aware of or don't want to see.

    That was the surface answer. The underlying answer is that the more modeling they do and the more number games they play, the easier it is to convince people they have The Answers. Any customer service drone working for minimum wage could call people and add up the results. You need a million-dollar polling analyst to turn that into Science.

    ReplyDelete
  27. It is interesting the lengths people go to convince themselves that polls are accurate.Does anyone really believe that undecideds suddenly break right at high margins before an election? Especially when, on the rare occasions an undecided is trotted out,they always sound like lunatic communist SJWs?

    ReplyDelete
  28. TargetEarly TargetSmart has 22.8 million Democrat ballots in so far vs. 18.0 million Republican. Sure, Democrats are ahead, but that was expected, and Republicans are keeping up. I do wonder how many TDS sufferers will be waiting to vote on Election Day. I'd venture to guess not many, especially since nearly all of them are literal covidphobes.

    ReplyDelete
    Replies
    1. I voted yesterday, but merely because I was 1 street away from the polling center. It was also much easier than the regular set up. Best part was you gotta have an ID!!

      Delete
  29. @20 Stilicho:

    Pollsters will likely be replaced by computer programs that try to predict votes based on whatever data big tech collects. Of course, that will be subject to the human foilbles and failings that have killed polling (programs don't write themselves, after all), but it may have a good run.

    I think that will be hard, unless big tech has already collected enough of the data that's needed. Two major problems: when it comes to using Machine Learning on people, the whole field has become pozzed because neutral data + programs tend to tell the truth, which is career ending. Second, it's not like you can run your program over a body of pictures and see what comes up for gorillas, the only real test is elections, which are infrequent. Thus I'm positing a need for retrospective data, but that also brings in biases to get your system to give the same results as the election, but they may be the wrong ones, some likely will be. The whole field is accomplishing major stuff, but there's still a lot of voodoo involved. As well as data scientists who don't know how to program and use cookbook recipe style techniques that are strictly voodoo to them for their Python code.

    @28 Damelon Brinn:

    I've heard claims that they have to [depend on polling at the national level] because there isn't enough state polling to be accurate.

    It's more like public state polls ask 500 people at most and have huge error bars. Perhaps internal polls for campaigns put more effort into them, especially for what they think are "battleground" states. Even then, it's pretty common for a non-battleground state to surprise the public pollsters, perhaps almost everyone.

    ReplyDelete
    Replies
    1. I don't think it will be very accurate, but I think it will be pushed hard. Judging by the ads I see when surfing the web, their algorityms suck. Same for Amazon which has hard data about what I buy.

      Delete
  30. Nate Silver is saying "Biden's going to win, prepare the guillotines my non-white minions!" as he low-key sells off his immobile assets and prepares to make aliyah.

    This is why I think naming the Jew is so important, even though all Jews are not guilty for the few. This is the first time I've consciously realized Nate Silver is a Jew. The shape shifting exploits the European predisposition for altruistic punishment. People have to realize these people are Jews. If they were Chinese, black, etc, they simply would, and the subversion would have never been so successful.

    ReplyDelete
  31. Pollsters are speculators. Find me a good one and he is keeping to himself and making money on everyone else's ignorance.

    ReplyDelete
  32. I don't answer my personal phone if I don't know the number. The secretary blocks everything on the office phone.

    I am going to vote in person because I don't trust the clerks and I want to see the counter on the scanner go up. I would tell anyone who asked me as I was leaving how I voted, to go screw themselves.

    Not sure who they are polling.

    ReplyDelete
    Replies
    1. This is exactly in line with what I said in my earlier posts. People over a certain age on the right don't talk to anyome about their business. Especially men.

      Delete
  33. Who are you going to believe? Our polls or your own lying eyes?

    Just a simple look at the handful of bodies at a Biden rally compared to Trump. FFS - Joe and the Ho posted to an event last week and no one showed. OK Luntz, tell us when that has ever happened.

    ReplyDelete
  34. @11 "Does Team Trump have a plan and tactics and strategy in place to prevent fraud,"

    I point out to folks: remember back in 2017 Trump set up an elections fraud committee that was to pull state data and look at fraud? "May 11, 2017 — President Donald Trump signed an executive order Thursday creating a commission to examine voter fraud and voter suppression,"

    And a couple of big corrupt states refused to give up their data? And so Trump closed the committee down? "Trump signs executive order disbanding voter fraud commission. Updated on: January 3, 2018."

    He has also signed; Executive Order on Imposing Certain Sanctions in the Event of Foreign Interference in a United States Election September 12, 2018.

    Is there ANYONE who thinks Trump just gave up on stopping election fraud, sent those commission guys back home. and did NOT continue to work on it?

    I know this is another 'trust the plan' -- but how likely is it the God Emperor quit trying to prevent voter fraud, knowing there was a 2020 election to come?

    In the meantime Judicial Watch has been doing a bang-up job suing and usually winning against states that have never cleaned up their rolls.


    ReplyDelete
    Replies
    1. I told the wife, the GE knows that if he loses the Dems will hunt his family into extinction, let alone putting all the right into re-education camps.

      Delete
  35. Jeroth wrote:

    This is why I think naming the Jew is so important, even though all Jews are not guilty for the few. This is the first time I've consciously realized Nate Silver is a Jew. The shape shifting exploits the European predisposition for altruistic punishment. People have to realize these people are Jews. If they were Chinese, black, etc, they simply would, and the subversion would have never been so successful.


    If Nate Silver was an articulate and intelligent black man, the media would be drooling over him even more than they are. But absolutely- the idea that a substantial slice of America relies on two Jews to lie is why they have to be named. When Trump wins again, their spell will be broken among all but the most soft-brained Americans.

    ReplyDelete
  36. Hopefully people will start to understand that all models are fake and show only what people creating them want.
    This includes the magazine kind.
    At least this is what my model tells me.

    ReplyDelete
  37. Interesting info out today on people searching Google on whether or not they can change their early vote. (https://trends.google.com/trends/explore?geo=US&q=can%20I%20change%20my%20vote) Seems like some people are having buyer's remorse. It is approaching 100(99 at the moment and almost vertical) placing it at the top for the day.

    ReplyDelete
  38. Polls are the sociology of the statistics field.


    Their demise was inevitable once they shifted their focus from predicting how people will vote to influencing how people will vote.

    What other purpose did the polls ever serve than attempt to dissuade or bully voters? For the candidates, they have uses for know where and what to focus on, but for the public, the polls are nothing more than social media fight fodder and even less valuable than the predictions about the outcome of sporting events or the political polls equally idiotic twin, economic forecasts.

    ReplyDelete
  39. Believe the bookies. They've got SITG.

    ReplyDelete
  40. It's a win win.
    If by some dark providence or really good cheating Trump loses, then the lesson is going to be "no matter how much money and gibs you throw at brown people, they will never vote for you".
    And maybe that will be the end of the cuckery.

    ReplyDelete
  41. 31. VFM #7634

    "TargetEarly TargetSmart has 22.8 million Democrat ballots in so far vs. 18.0 million Republican. Sure, Democrats are ahead, but that was expected, and Republicans are keeping up. "

    Yes, Breitbart is saying over half the Democrats have voted in person or by mail (53%) as of today, yet only 25% of Republicans have voted so far. About 94% of Republicans say they will certainly vote. Add to that a large number of Independents (well over half) will likely vote Trump.

    ReplyDelete
  42. I don't know how much polling is still done by telephone, but I had an interview with the prominent Canadian polling firm Ipsos Reid years ago, and the man there admitted that 70% of the people they called refused to talk to them. He didn't say anything about the remaining 30%, but I suspect they were people with an ax to grind.

    Humanistic psychologist Abraham Maslow noticed that normal people were reluctant to talk about their sexual practices, while weirdos weren't. He warned Alfred Kinsey about the "volunteer bias."

    People have long been known to lie on sex surveys, religious surveys, and political polls.

    "Smart politicians trust the voters, not the polls."
    --Batman, after defeating the heavily favoured Penguin for mayor of Gotham City.

    ReplyDelete
  43. TargetEarly TargetSmart has 22.8 million Democrat ballots in so far vs. 18.0 million Republican.

    That looks like a disaster for Democrats, considering something like 50% of Democrats and 10% of Republicans said they intended to vote early. Either a bunch of people changed their minds about that, or there's been a huge difference in turnout so far, with many Democrats deciding not to vote.

    ReplyDelete
  44. Oh, I don't think the polling industry is finished, for the same reason that self-righteous fools think that racism is the worst sin possible and is rampant in White society. Some people will believe it if someone on "their team" says the sky is green. Two sayings come to mind. "There's a sucker born every minute" and "You can fool all of the people some of the time,and some of the people all the time, but you can't fall all the people all the time". There are enough of the "some people" category that polls will still have a use for the (((elite))).

    ReplyDelete
  45. Nate Bronze says if the polls move toward Trump as the election nears, which he gives a 50/50 chance of happening, and there's a polling error in Biden's favor now, which he also gives a 50/50 chance, then there's a 50/50 chance of Trump winning, which is how they currently have him at 1-in-8.

    So to parse out that steaming pile of nonsense: If the first two things happen, that doesn't make his chances 50/50. It would depend on the size of the shift and the size of the polling error. But as close as it already is in the battleground states, it would almost certainly put Trump's odds higher than 50%. On the first two things, we know that the polls will narrow; they always do. So that one's 100%, not 50%. And the second thing is really just a restatement of the first: the polls narrow because they take out some of the error. So that one's 100% too. So it translates to a 100% chance of Trump having at least a 50% chance.

    Of course, what Silver is really doing is sticking 50/50 in a tweet so that after Trump wins, his innumerate midwit fans can say, "Well, he was the only one who gave Trump a 50/50 chance."

    ReplyDelete
    Replies
    1. I came to post the same thing, lol.
      I'll repeat what I said on another thread-I did push polling for years, back in the 80s-90s.
      I've NEVER trusted polls and never will.
      People are shockingly stupid and gullible. More than you really can imagine, unless you've had widespread contact with the GP.

      Delete
  46. Luntz's focus group had 12 people? Are there that many in the whole country who are genuinely undecided? I'd be hard put to find that many who'd change their minds if their candidate strangled a live kitten on national TV.

    ReplyDelete
    Replies
    1. That's why you're wrong and they still do and say all these idiotic things. A good 25 percent of the electorate can barely put on foot in front of the other. It's why the media endlessky says biden or hillary has it in the bag. Because they want to demoralize and dampen the vote of the opposituon while getting bandwagon morons who like to be with the winner.

      Delete
  47. > It shouldn't cost much more to poll X% of each state's population than it does to poll X% of the country.

    As a matter of fact it does. The issue is not how many voters in a state are polled, but which ones. It's very easy to poll in the urban jungles, but response rates are absolutely terrible in the suburban and rural areas of many states. I've heard it said that you might have to call 50 people to get a single response, and to increase the response rate you have to spend days or even weeks chasing people down.

    Phone polls have been broken for a long time. In 2016, because the right was dominating social media, internet polls were actually OK - but because of the mass deplatforming, internet polls now have exactly the same problems as phone polls.

    Compared to the cost of getting boots on the ground and soliciting homeowners for days upon days, it's a tiny fraction of the cost to collect another 10 billion responses from California and New York and add a footnote saying that you oversampled Democrats.

    I don't doubt that a lot of the polling is politically motivated, but they'd also have to do a lot more work and spend a lot more money to get accurate numbers. So even if they aren't intentionally trying to fake the results, the incentives are totally against them working a lot harder just to find out that Trump is doing much better. Most of them have no skin in the game, and the ones who do have skin in the game, like IBD/TIPP, unsurprisingly come out with numbers that are vastly more favorable to Trump and Republicans.

    ReplyDelete
  48. I have a suspicion that the polls will continue being wrong until the election day this time. We will see... it can only mean that the Trumpslide will be that much bigger.

    ReplyDelete
  49. @CecilHenry
    No blackpilling. Especially if you're not able to make your own argument and instead link to a flaky "news" site.

    ReplyDelete
  50. Who is paying pollsters? Same organizations paying the economists and the security consultants. The dying dinosaur media.

    Tick tock.

    ReplyDelete
  51. If Trump wins, the media will simply claim that Trump is a unique phenomenon of outsider and charismatic firebrand that only temporarily defeats the polling accuracy.

    Unfortunately, there would be much truth to this assertion.
    I don't see any Republican I would have confidence in for 2024.

    As of now, I believe Trump will win.

    There's no way to accurately quantify the righteous hatred Trump supporters have for the media - including pollsters - or the reluctance of undecided voters to express an "unpopular" opinion.

    While Democrat samples are deep blue, the Republican samples are pink. It's the deep red that will come out in droves in the swing states on election day.
    Also, I expect undecided voters to vote heavily for Trump.

    ReplyDelete
  52. Nate gave Trump 28.6% chance in 2016.
    He gives Trump 13% chance in 2020.
    Trump wins both election, in a 28.6% x 13% = 3.7% chance.
    Hence, Nate's accuracy is only 3.7%. That sounds high, did I make any math mistake?

    ReplyDelete
  53. Phone polls have been broken for a long time.

    It's kinda sad. D.C. is the ground zero of polling, and also the seat of Diversity and Inclusion. For anyone that's had to take a Diversity training session for work, you come away from it realizing the only shred of validity to any of it is the idea that if you want to sell to a particular group, it helps to have some of them on board to tell you what their peers really think about things.

    A White woman really is going to have some trouble understanding what a Black woman needs in a hair product.

    But these beltway dorks, the last thing in the world they could ever conceive of is having rural, White, right-wing folks actually part of the policy process. So they do polling to find out what those fly-over rubes want.

    They can't even follow the one single part of their diversity mantra that actually works. All the non-sense, none of the benefit.

    ReplyDelete
  54. Pollsters are already pre-emptively preparing their excuses for why they were even further off this time.

    ReplyDelete
  55. Lance E wrote:It's very easy to poll in the urban jungles, but response rates are absolutely terrible in the suburban and rural areas of many states. I've heard it said that you might have to call 50 people to get a single response, and to increase the response rate you have to spend days or even weeks chasing people down.
    My mother told me that in the late '40s, she worked briefly for Gallup. Phones still weren't ubiquitous, and she drove out in the country to poll, to places like an indian reservation. That sounds as if Gallop actually wanted accurate measurements back then.

    If pollsters still wanted accurate polls, they could still hire people to drive around and ask questions. They were once willing to pay what that cost. That they aren't now suggests that they and their customers just don't care.

    ReplyDelete
  56. Actually, Romney did better than McCain in the elections. Romney won all of the McCain states plus NC & IN (and one e.v. in NB).

    McCain 173 e.v.
    Romney 206 e.v.

    Popular Vote Margin of defeat

    McCain 9.5 Million votes
    Romney 5 Million votes

    ReplyDelete
  57. It's arguable that pollsters should have gone out of business in either 1936, when The Literary Digest's poll was widely off -- and, of course, in the 1948 election when Truman defeated Dewey, and not the other way around as the famously incorrect Chicago Tribune headline said.

    Dennis DeTurck at Penn -- an old school math prof, not your wokey diversity oriented math prof -- had done a lot of work in this area. See his case studies:

    https://www.math.upenn.edu/~deturck/m170/wk4/lecture/case1.html

    https://www.math.upenn.edu/~deturck/m170/wk4/lecture/case2.html

    As a result, I have to throw cold water on the pleasant notion that the polling business will collapse if a Trumpslide occureth.

    Polling is too much inbred into the body politic. Cable news talking heads and mainstream news outlets need the pollsters for content. University professors need the publishing opportunities and outside money polling brings them. Organizations like Gallup and Pew depend upon the revenue they get from either corporate consulting or grants to stay in business.

    So, yeah, they will slink off into the corner for a couple of years, but by the next cycle they will be back again with new models or new approaches or some stupid math alchemy.

    The industry is too entrenched. And, as we have seen, being really really wrong is NOT a career or business killer, though maybe jerking off on a Zoom call might be.

    Look on the bright side.

    Pollsters are here to stay, but their inevitable failures will be really entertaining.

    ReplyDelete
  58. KPKinSunnyPhiladelphia wrote:Cable news talking heads and mainstream news outlets need the pollsters for content. ... Organizations like Gallup and Pew depend upon the revenue they get from either corporate consulting or grants to stay in business.
    Yes, as long as there is demand, and there will be, somebody will fill it.
    KPKinSunnyPhiladelphia wrote:So, yeah, they will slink off into the corner for a couple of years, but by the next cycle they will be back again with new models or new approaches or some stupid math alchemy.
    I don't think it will be that respectable. They'll rely on Gell-Mann amnesia, short memories and big lies.

    ReplyDelete

Rules of the blog