The polls have gone cancerous


The news is hideous and diseased — every new poll inspires a frenzy of speculation, and gets promoted heavily on the networks, depending on whether the results meet their biases or not. Unfortunately, that provides an incentive to the Poll Industry to do more polls and to found more polling companies, further undermining their credibility. The latest flurry of rabid prognostication was triggered by a poll that suggested Iowans might favor Harris.

I honestly don’t give a fuck. I have come to despise both polls and the media’s reliance on them to generate “news”.

I am pleased to see that Bob agrees with me.

P.S. Especially Nate Silver, you can just fuck off to the moon.

Comments

  1. Rich Woods says

    We’re getting the same in the UK news web sites. It’s tedious. It’s also lazy journalism, but somebody must be clicking on it else the reptiles wouldn’t bother.

  2. billseymour says

    I just saw a report on NBC’s Today in which the polling “expert” showed an 18% advantage for Trump among men and a 16% advantage for Harris among women.  He then added those two numbers together and said that there’s a 34% gender gap.

    How can these clowns seriously claim to have even one significant digit of information when they don’t even pay any attention to the sign?

  3. seversky says

    The polls are basically reporting a dead heat so all we can infer is that there is likely to be a legal furball over the results. This means a Trump victory since he’s rigged SCOTUS.

  4. stuffin says

    I followed the 2016 election very close, read every poll that came out especially ones saying Hillary would win. I knew what Trump was, have been reading about him since the 80s. Scumbag is term I would use to describe him back then. A few more adjectives are now available for us to describe Trump. Anyway, the polls from 2016 were all wrong, I then followed the 2020 polls moderately and they were fucked up also. So, my mantra this election is to ignore the polls, no matter how tempting the headline. I have been successful with only one more day to go.

  5. robro says

    I hate polls. They’re a scam. Both sides use them as a scare tactic to pump people for donations. The media uses them to gin up eyeballs for their lame “news” reporting to sell ad space. The only poll that matters is the one that will start getting counted tomorrow evening.

    Nate Silver is a serial failure. You didn’t specify which moon, so I’ll add: he can fuck off all the way to the moons of Pluto.

  6. raven says

    I followed the 2016 election very close, read every poll that came out especially ones saying Hillary would win.

    That was what I was going to write. Exactly.

    I followed the 2016 election close, donated money to the Hillary campaign, and thought Hillary would win, without a lot of confidence.
    Stayed up watching the election results and by 11:00 PM, knew that Trump would win the electoral college vote.
    Disappointed and at that time, I knew the USA was in for a bad 4 years, which is what happened.

    Ever since then, I don’t believe the polls or pay any attention to them. Why should I when they are wrong a lot?

    So, who will win tomorrow?
    I have no idea. Will try not to even watch the results until they mean something.

  7. raven says

    Trump and the GOP have weaponized lies. There is a point to their incessant lies. It helps them win elections.

    AFAICT, most or all of the polls have been weaponized. They are slanted in various ways for various reasons.

    What they aren’t is accurate.
    Which is why I won’t pay much attention to them or believe them.

    If there is a 10 point difference, they might be accurate. A 5 point difference, probably not.

  8. says

    @#1 Rich Woods
    Exactly, lazy journalism, more concerned with clicks and likes than with what anything actually means.
    Real news happens on its own timeline, takes place when and where it wants to, is messy, and has implications and ramifications for the future.
    Polls are scheduled events, come in neat packages, everything is quantified, broken down by demographics and don’t require anyone to leave the office. Lazy journalism.
    Throw in the well-known fact that results can be skewed on any question depending on how it’s phrased, and the fact that in today’s world, you’re lucky if one per cent of the people called actually answer the poll, and what do you have? A non-random sample answering slanted questions that they heard asked over the phone and may or may not have understood, outsourced to a call center somewhere.
    But it makes a good headline and can be scheduled conveniently, and will reliably garner the clicks. Journalism is all but dead.

  9. raven says

    …and the fact that in today’s world, you’re lucky if one per cent of the people called actually answer the poll.

    I had a polling organization call me on my land line a few weeks ago.

    I listened until I heard the word “poll” and then hung up on them.
    A few days later, they called me again with the same result.

    A lot of phone polls are fake push polls, not really polling but to lead you to vote for their candidate.

    Questions: Would you vote for a Black baby killing, commie woman for president? The one endorsed by the Peoples Republic of China.
    Would you vote for the current Democratic party that caused the major Depression we are now in?

  10. awomanofnoimportance says

    I question whether it’s possible to get a representative sample at this point. Before cell phones, pollsters would call names randomly picked from the phone book, and as long as the sample size was large enough their results were usually pretty good. However, these days there are no phone books, most people screen their calls, and certain demographic groups are far easier to get to if you’re a pollster. So getting a fair sample may no longer be possible.

    All that said, I am reminded of the old joke about a new medication being tested for a disease that affected chickens. One-third of the chickens who were given the medication got better, one-third of them died, and the third chicken ran away so no results could be obtained.

  11. jack lecou says

    I just saw a report on NBC’s Today in which the polling “expert” showed an 18% advantage for Trump among men and a 16% advantage for Harris amaong women. He then added those two numbers together and said that there’s a 34% gender gap.

    For better or worse, that’s not necessarily an error on his part. Not an isolated one, anyway. That particular mathematical shortcut seems to have become the norm in reporting on the subject (e.g., here’s ABC doing the same thing). I guess at this point, it’d probably be more confusing if anyone actually actually used the correct calculation.

    If I’m doing the math right myself, I think adding the gaps will be proportional to the actual number (2x), at least as long as all the signs remain the same. So while it overstates the absolute difference, it is workable-ish for historical comparisons (“this is the largest gender gap in history”, etc.), which I guess is why tv-pollster-pundits and their ilk consider it ‘good enough’.

    It does kind of underscore the overall laziness with that ecosystem though…

  12. charles says

    I wish I could believe Iowa might go for Harris. I am becoming more dissatisfied with politics in Iowa, to the point of considering moving.

  13. jack lecou says

    awomanofnoimportance @11: I question whether it’s possible to get a representative sample at this point.

    Yeah. Especially problems like savvier people tending to just hang up (like raven @10), or, in our hyperpartisan times, maybe even stay on the line and give strategically false responses.

    I know there are theoretically ways to fix it all up in the stats: weighting, control questions to detect falsehoods. But that’s still an awful lot of wrenches in the gears of what was a pretty fragile machine to begin with.

    I myself am kind of hoping the whole political polling process continues to melt down to the point of uselessness. It’s not actually a vital building block of democracy or anything. Before the 1930s, IIRC, national presidential polling wasn’t even a thing. And when it did start happening, it was somewhat controversial — people warned of some of the very problems we’re dealing with now.

    So I wonder if we wouldn’t have somewhat healthier elections without it. Maybe it tamper down some of the irritating horse-race style reporting, and news organizations could actually cover — gasp — the substance of the issues.

  14. stevewatson says

    Some people I know (Canadians!) are having a fucking watch party tomorrow night. Seriously? I’d rather have dental surgery sans anesthetic. I’ll check the internet about midnight to see if there’s any result we can trust yet.

  15. raven says

    I am becoming more dissatisfied with politics in Iowa, to the point of considering moving.

    That can happen.

    That people move from the state they are in, to another one based on politics and the related idea of “culture”.

    I saw that with Utah.
    I’ve had relatives there from long before I was born. I used to visit Utah often as it is one of my favorite places. For the outdoors, the geography, and river running. The Mormons aren’t its best feature.

    For that and other reasons, I had a lot of friends there. Had, past tense.
    Over the years they all moved away from Utah. And it was the dominant culture and politics. Gentiles as they are called are a minority and treated as such by the Mormons. They own the Utah state government which is a theocracy in all but name.
    As a minority, your vote will never count for anything.

    They liked being there and they liked being gone. Some of them left when they had kids and didn’t want them being discriminated against in the schools, which will happen invariably in the public schools.
    Some of them left when they retired and moved to the coast.

  16. drew says

    It’s not polling that bothers me, but the lack of (understanding of) error bars in the results.

    My weather forecasts tell me likelihood of events. Why can’t other predictive news?

  17. drew says

    It’s not polling that bothers me, but the lack of (understanding of) error bars in the results.

    My weather forecasts tell me likelihood of events. Why can’t other predictive news?

  18. says

    Who’d want to play a drinking game like that? I kina suspect I’d either be totally sober at the end, or drink myself to death by midnight.

  19. Tethys says

    I’m cautiously hopeful about that Iowa poll, as it is based largely on exit polls from actual voters rather than the shoddy methodology of calling random people or conducting an online poll.

    It’s not just Iowa. Ohio and Kansas and Oklahoma are all experiencing the same trend of enormous voter turnout in early voting, with women and young voters accounting for a huge percentage of those voters.

    Combine that with the fact that approximately 20% of registered Republicans voters are never Trumpers.

    Those are excellent signs that Harris is going to win.
    I hope it’s by large enough margins that no recounts are triggered and election officials are ready for the inevitable whinging and frivolous lawsuits of the magats.

  20. jack lecou says

    robro @6: Nate Silver is a serial failure. You didn’t specify which moon, so I’ll add: he can fuck off all the way to the moons of Pluto.

    So, he can definitely fuck right off. But I do think credit where it’s due. Silver’s approach of actually running a monte carlo simulation over all the possible variations is objectively the right way to aggregate polls and do election predictions — as opposed to just averaging all the polls together (or worse, just reporting on whichever poll happens to be the latest). Obviously he didn’t invent the technique, but AFAIK, he was the first to manage to get that approach into anything like the mainstream of political horse race analysis.

    And again in all fairness to Silver, I don’t think he actually got 2016 wrong. Despite the polls being a bit tilted, his model gave Trump about a 25% shot, IIRC. Many people took that 75-25 number and ran away with “Oh, so that’s saying the 75 side is going to win,” but…that’s not how odds work. It was never a prediction that Clinton would win. It was a prediction that Trump wins 1 time in 4. And that’s the same chance as flipping a coin twice and getting all heads. IOW: pretty good odds. Certainly a much better shot than almost any other prediction was giving Trump. Silver could have been entirely right, and it’s just our bad luck to end up in this universe rather than one of the other 3.

    Now, I think the polls are even more thoroughly cooked now than in 2016 (sounds like Silver sort of thinks so too), but back in, e.g., 2012, Silver’s approach worked pretty well. So the interesting thing to me about that was that despite the polls not being super-cooked yet, and despite 538 — or any non-superficial analysis of the polls — very accurately predicting pretty much a snowball’s chance in hell for Romney, the mainstream horse-race pundits still insisted on reporting on the race as if it was a nail biter right down to the very last precinct counted. (I actually had a chance to ask George Stephanopoulos for a justification of that nonsense at a private event in early 2013. He actually seemed surprised by the question.)

    So coming back to 2024 and beyond, I think the bottom line is that whether accurate polls (or aggregate predictions) are theoretically possible or not these days is kind of a moot point. Both the media and the campaigns themselves have converged into a position of identifying genuinely predictive polling (anything that doesn’t say 50-50) as anathema. Probably correctly, from their point of view. The media wants to be able to maintain the pretext of 24/7 down-to-the-wire horse-race coverage. Meanwhile, the campaigns want their supporters to not feel discouraged or confident — why vote if your guy is too far up OR down.

    On top of all that, there’s the Trump phenomenon, where it’s very important to have polls (however garbage) to later cite in a court case calling the vote into question. (“A recent Two-MAGA-Chuds-In-A-Barn poll had our client UP by 5 points in Philadelphia. How do you explain that, Mr. Secretary of State?”)

  21. jack lecou says

    robro @6: Nate Silver is a serial failure. You didn’t specify which moon, so I’ll add: he can fuck off all the way to the moons of Pluto.

    So, he can definitely fuck right off. But I do think credit where it’s due. Silver’s approach of actually running a monte carlo simulation over all the possible variations is objectively the right way to aggregate polls and do election predictions — as opposed to just averaging all the polls together (or worse, just reporting on whichever poll happens to be the latest). Obviously he didn’t invent the technique, but AFAIK, he was the first to manage to get that approach into anything like the mainstream of political horse race analysis.

    And again in all fairness to Silver, I don’t think he actually got 2016 wrong. Despite the polls being a bit tilted, his model gave Trump about a 25% shot, IIRC. Many people took that 75-25 number and ran away with “Oh, so that’s saying the 75 side is going to win,” but…that’s not how odds work. It was never a prediction that Clinton would win. It was a prediction that Trump wins 1 time in 4. And that’s the same chance as flipping a coin twice and getting all heads. IOW: pretty good odds. Certainly a much better shot than almost any other prediction was giving Trump. Silver could have been entirely right, and it’s just our bad luck to end up in this universe rather than one of the other 3.

    Now, I think the polls are even more thoroughly cooked now than in 2016 (sounds like Silver sort of thinks so too), but back in, e.g., 2012, Silver’s approach worked pretty well. So the interesting thing to me about that was that despite the polls not being super-cooked yet, and despite 538 — or any non-superficial analysis of the polls — very accurately predicting pretty much a snowball’s chance in hell for Romney, the mainstream horse-race pundits still insisted on reporting on the race as if it was a nail biter right down to the very last precinct counted. (I actually had a chance to ask George Stephanopoulos for a justification of that nonsense at a private event in early 2013. He actually seemed surprised by the question.)

    So coming back to 2024 and beyond, I think the bottom line is that whether accurate polls (or aggregate predictions) are theoretically possible or not these days is kind of a moot point. Both the media and the campaigns themselves have converged into a position of identifying genuinely predictive polling (anything that doesn’t say 50-50) as anathema. Probably correctly, from their point of view. The media wants to be able to maintain the pretext of 24/7 down-to-the-wire horse-race coverage. Meanwhile, the campaigns want their supporters to not feel discouraged or confident — why vote if your guy is too far up OR down.

    On top of all that, there’s the Trump phenomenon, where it’s very important to have polls (however garbage) to later cite in a court case calling the vote into question. (“A recent Two-MAGA-Chuds-In-A-Barn poll had our client UP by 5 points in Philadelphia. How do you explain that, Mr. Secretary of State?”)

  22. billseymour says

    I think that jack lecou probably gives a more sober account of what 538 actually does; and Silver is very clear that he is not making a prediction, but just giving probabilities.  He also publishes his methodology and says what polls he’s including, and how they’re weighted for historical accuracy, so that others can see what he does and maybe find fault or suggest improvements.  He’s not giving you a magic number that you should believe because Nate Silver said it.

    I haven’t been following 538 this election, but I’ll probably have their live blog up tomorrow night.  In the past, they’ve had a handful of reporters assigned to cover particular races, or county-by-county results in the various swing states.  All will be clear that they’re reporting trends that could change, and so they’re not predicting any outcome (unless it’s obvious).

    Also this year, atarting around 10pm central time (IIRC), Robert Reich and Heather Lofthouse will have a live podcast following the election results.  I’ll probably be tripple-tasking, including having PBS’ News Hour coverage on the TV.  I’m not very good at multi-tasking, so that might not go well…we’ll see.

  23. DanDare says

    Polls are displacing actual discussion of issues.
    Human rights, the environment, the economy, the structure of our government systems.

  24. chrislawson says

    @26– This is a progression of a very old problem in journalism that I started to see abot 20 years ago even in supposedly good news outlets where the reporting became increasingly about the fight between two candidates or parties. Closer to a report on a boxing match. The actual issue the people were fighting about was just a label, and never examined. No discussion of the policy differences, what the policies would mean to people, or what the state of the evidence was. As if saying a fight was about ‘the environment’ or ‘government debt’ was all the reader needed to know.

Leave a Reply