Can We Agree That Polls Are Just A Form Of Fake News?


A couple of week before the Mueller Report’s summary was released, one  poll announced that Trump’s approval had finally topped 51%.  Then it dived again because of some  dumb tweet or bad news. After the report came out and there were no smoking impeachment guns, we were told that it was ominous that the President’s approval rating hadn’t moved.  Then it did move–up—in some polls, but after the “I’m fucked!” story, it moved down. Meanwhile, the Biden and Bernie polling race continued like the climax in “Seabiscuit,” but it was Pete Buttigieg’s polls that had everyone talking, since they had increased exponentially, though only into single digits still. Then, as Joe Biden’s entry into the race approached, Drudge announced a shock poll yesterday: Good Ol’ Joe led the President by 8 percentage points in a hypothetical run against Trump. Run, Joe, Run!

Issues & Insights offers a useful perspective on all of these polls:

…In May 1983, a Gallup poll came out showing that then-Sen. John Glenn would do better in a matchup against President Reagan, with a far wider margin than the Biden/Trump split.

As the New York Times reported on May 19 of that year, the survey “found that Senator Glenn led Mr. Reagan 54 percent to 37 percent.”

At that point, the eventual Democratic nominee, Walter Mondale, had a 49 percent to 43 percent edge over Reagan.

Harris survey in July 1983 concluded that “it seems certain that the 1984 presidential election will be close, with neither the President nor his most probable Democratic opponents assured of any solid lead.”

Actual result: Reagan won a massive landslide over Mondale, getting almost 59% of the popular vote and 525 electoral votes — winning every state except Mondale’s home state of Minnesota.

Fast forward to 2003, when President Bush’s sky-high post 9/11 ratings started to collapse….An April 2003 Gallup poll found Sen. Joe Lieberman with a lead over the nine other candidates running at the time. A month before, Rep. Dick Gephardt was leading the pack.

…Bush beat Sen. John “Reporting for Duty” Kerry by a 50.7 percent to 48.3 percent margin. Bush won 31 states, compared to Kerry’s 19.

Nor does Trump’s low approval rating at this point mean anything in terms of his reelection prospects. As Gallup points out:

“There are several examples when presidents who had nationwide approval ratings in the 40% range in the year before the election won a second term, including Ronald Reagan, Bill Clinton and Barack Obama. All managed to get to 50% approval by the time of the presidential election. Trump has yet to attain that level in his presidency, but he will have the next 20 months to get to there, and to make the case to Americans that he deserves a second term.”

What’s more, Trump’s approval rating in the Gallup tracking poll — 45 percent — is identical to Obama’s at this same point in his presidency. It’s higher than Reagan’s, two points below Clinton’s.  And, just to underscore how meaningless approval ratings are this far out from a presidential election, George H.W. Bush’s approval rating at this point in his first term was 77 percent.

OK, I’m stumped: why are these polls even reported? OK, I’m not really stumped: they are reported to mislead and manipulate public opinion by advocates, political parties and news media pushing for public support of their agendas, in an “appeal to authority,” using science–bad science—as the authority involved. They are used to make the public think other people think something, so the “Everybody thinks it!” herd mentality kicks in. This is the theory, even now, after we have a President who was elected without a single poll ever showing that he would win.

The results themselves are easily and frequently skewed by pool selection, polling methods and questions, and yet even when a poll is professional and objective, there is still no reason to believe it. People who say the will vote, won’t. People lie.  Most people are idiots. A depressingly large  proportion of the public will tell pollsters that they believe or feel diametrically contradictory different ways within short periods of time based on dubious information, or no information at all, because they just don’t think very much or hard about the matters at issue, and don’t have adequate critical thinking skills anyway.

I know that the news media, pundits, advocacy groups  and political parties simply cannot resist publicizing these things. However, I propose that any published poll carry this full disclosure warning along with an explanation of its methodology and the period during which the responses were obtained:

“The polling organization does not guarantee the accuracy or reliability of any of the information or conclusion contained in, resulting from, or linked to this poll, and  any reliance upon any of these materials is at the sole judgment and risk of those using it. Such polls are provided for entertainment purposes only.  No warranties are made as to the correctness of the results, methods or tabulations.”

(Oh…I adapted that from an Astrology column.)

34 thoughts on “Can We Agree That Polls Are Just A Form Of Fake News?

  1. Two comments:

    Polls are the “Science” in “political science”, when they use scientific methodology. To dismiss all polls as “Fake News” is unfair.

    I reject the term “Fake News” as it is arbitrary and necessarily reflects the bias of whoever pronounces something as “Fake News.” In current parlance, our President declares things he doesn’t like or agree with as Faje News. That doesn’t mean the next person thinks it’s Fake. The term simply expresses a difference of opinion but implies that one person is right while the other person is wrong.

    • I use polls sometimes, like the annual Gallup poll on which occupations are trusted. But i don’t understand how political polling snapshots that are so routinely undependable should ever be on a front page or even page 50. I know your perspective as a legislative professional, and appreciate that representatives have to have some way to take the pulse of the voters.

      “Can’t live with ’em, can’t live without ’em.”

      • I agree. We saw both pre-election and exit polls declaring Hillary Clinton winner by large margins. The actual results proved all of them to unreliable. To a poll, they all got it wrong.


    • Interestingly in the early 2000’s most democrats/lefties thought we had fake news because only nine corporations owned all the media. Now that Trump says it I guess the notion “reflects the bias of whoever pronounces something as “Fake News.'”

      Btw does anyone know how many corporations own the media in the states now? I heard it was like six.

      • We have fake news because journalists and editors no longer follow their own professional standards, and because the media reports fake news, distorted news, or hyped news for various agendas, including online views and ratings. Simple as that.

        Ethics Alarms is talking about reporting that is objectively and demonstrably false, slanted, or distorted. That has nothing to do with my biases. It has to do with knowing something.

        I’ll use my favorite fake news on election night, 2016, which I have written about several times. We were told by a historian that the same party seldom wins the White House three terms running. It’s one of Hillary’s excuses. I still hear people quote it. The statement is false. It has never been corrected by a news organization. I know it is false because I know my Presidential history. The news media counts on the public being ignorant and gullible.

    • I think this comment makes some good points.

      The problem is not really the polling, generally. Good polling is based on scientific and statistical principles, and when done according to those principles produces broadly reliable results.

      The problem Jack is getting at, I think, has to do with the way the news media uses the polls they commission. I think we can probably all agree that some polls have slanted questions designed to increase the likelihood of a particular result. Those probably qualify as “fake news,” since poor question construction creates the GIGO effect (garbage in, garbage out).

      But assuming the collection instrument is solid, the tendency for the media is to look for results that either fit the narrative they want to pursue, or provide a tasty headline. I’ve seen many times a click-bait headline like the one in the instant case, which was later ameliorated to insignificance far down in the discussion of the actual results. This creates a false narrative because of the collectively sort attention span of the news-consuming public, and that is intentional either as a bias by the reporter of the matter or as a case of economic importance overriding the desirability of accuracy. In common vernacular, the results are often “spun.”

      Also in the instant case, the reporting mentions that the poll is of likely voters, and it highlights two main groups (millenials and women) where Biden maintains a large lead. What they fail to mention, among many other things which Jack has identified, is that those two groups are significantly less likely to actually vote than the ones where Trump enjoys majority support.

      So Jack’s implication that this poll was designed to create a headline is likely correct. It contains many big-picture flaws in execution, but it gets treated as groundbreaking or important. That simply is a form of fake news — No poll at this point is important enough for that treatment, and given the deliberately offhand treatment of the major sources of Biden’s lead, it is clearly designed to be an attention-getter for clicks.

        • He seemed nice to me, and didn’t regurgitate talking points or some kind of deluded nonsense some of our prior commenters on the left have engaged in. I appreciate that.

          Like you, I hope I see more of him.

  2. Great post, Jack. Another one of your, “Man somebody should write about X” arrows into the bull’s eye.

    • By the way, although I’ve stopped watching sports talk shows, I’m still waiting for a regular Monday morning show (or any morning at all show) that goes over the previous day’s prognostications and reports how accurate (or inaccurate) those prognostications were. As I was once told by a statistician, “You can’t predict the future. You might as well flip a coin. And don’t get me started on TV stock pickers!

      • Another nominee for unneeded fake news: Reporting how much money a particular candidate has raised in a particular period of time. Who cares? It’s as lame as reporting how much box office particular movies amassed over over a weekend. Just laziness on the part of reporters. As bad as the now popular headline trick of calculating how many people are potentially effected by a potential storm. Who cares?

  3. Polls are useless, and are never motivated by an intention to get “real’ mass opinion. Every pollster knows that the way questions are asked has an effect on the responses. They count on that, and don’t care one whit about truth. Your examples of how off-the-mark polls can be tell the tale.

    • Beat me to it. Polls are, and are probably designed to be, inaccurate and, as a predictive instrument a waste of time and money.

  4. I’m with E2. Poll results conveyed by most media today are void of science and packed with wishful thinking and propaganda to reinforce the premeditatedly sought-for results. I don’t trust a single one of them. Why should I? Trust them, I mean. I can scarcely trust the “final tallies” of any election anymore. Especially whenever Democrats are “voting.”

  5. I always liked this bit of Trivia: Walter Mondale lost 49 states in 1984 as you note, only winning Minnesota. In 2002 he ran for Senate, and lost. Walter Mondale is the only person to lose a statewide election in every one of the 50 states.

  6. I don’t disagree with the premise of this post – that polls shouldn’t be reported as news – but I DO disagree with the arguments supporting it. My career involves working with news media, which means that I have to understand it. More importantly, my work also occasionally involves use of commissioned polling data in order to guide and measure the work.

    Here are the realities, as I see them – and again, I do NOT disagree with the conclusion that polls shouldn’t be considered news:

    – Some polls are methodologically worthless. This includes any online poll rooted in a media organization or entity with a self-selecting audience. As such, an online poll splashed on an organizaton’s website has no value because there are no controls on who responds and no attempts made to normalize for the large population. The only people who respond to such polls are those who already have an interest in that organization’s activities, good or bad, and social media makes it extremely easy to screw with the results.

    – With this in mind, it’s important to understand the effect that do-not-call registries, cell phone usage and other changes in the way the population communicates have had on the polling industry. A generation ago, a conscientious polling organization – yes, they do exist – could randomly select a few thousand phone numbers and, assuming reasonably unbiased questions, generate a reasonably predictive result. No longer. Competent and ethical polling and market research firms are deeply concerned about this,

    – There absolutely ARE polling organizations with built-in biases. It’s not hard to figure out who they are. The worst of these are push-polling firms hired by political campaigns. Push polling isn’t a measure of public opinion; its a propaganda technique.

    – The PUBLIC is receptive to news items that center upon opinion polling for one primary reason: confirmation bias. It likes this stuff.

    – The news media is well aware of this fact. News organizations know what plays and what doesn’t; sometime within the past twenty or thirty years some sharp editor recognized that a stroy that reported the results of an opinion poll, however obliquely, was playing well. When a news outlet sees a story doing well, it will invariably create more stories just like it.

    – This is even more important today, as conventional sources of revenue for news outlets dry up. Every Facebook link or tweet promoting a poll-based news item generates literally fractions of a penny in revenue. Given the sorry financial state of much of the news bidness, those fractions are vital.

    – Closely related to that last point, the news media touts poll results as stories because they’re cheap and easy. There’s no real reporting involved, no shoe leather, no stories that take weeks or months to develop. No hours-long meetings between reporters and editors to tick and tie details. And the public responds favorably.

    Yes, much of the news media is politically biased, and yes, biased news organizations are more likely to publish poll results that align towards their bias. But we must remember that the news outlets tend to be biased towards the interest of their audiences. Thus, the REAL bias to poll-based stories is a bias towards the bottom line. Nothing more nefarious than that – though that’s bad enough.

  7. To this day I fail to understand how statistics guru Nate Silver got it so right in 2012 and so dead wrong in 2016. If he was working with data that was no good to begin with, then that would explain why he fell flat in 2016, but that still doesn’t explain how he got it so right in 2012. Was there a huge Bradley effect in 2016?

    • Nate Silver didn’t do so bad in 2016. I was following him at the time and he gave Trump a 29% chance on election day, and a 10% chance of losing the popular vote and winning. I remember thinking he thought much more highly of Trump’s chances than most others. I don’t recall anybody other than him ascribing a significant possibility that Trump could be elected without winning the popular vote. He is the only reason I wasn’t totally shocked by the result.

      • His method of giving odds rather than predictions is clever, because he can always say, as he did in 2016, “I never said X wouldn’t win, I just said the odds were against it. But basically, he and his site predicted that Hillary would win, and didn’t see the flips in the Rust Belt coming. If a sportswriter said that, say, the Mariners had a less than 30% chance to win the AL West, and they won, nobody would let him say, “See?” Nate understands that the long-shot wins much more often than most people think. Bill James once did a peice about how it was ridiculous for the odds against a team winning a championship to ever be 1000 to 1.

      • To this day, I remain mystified by the idea of a college major called “Political Science.” I guess I assume it’s “The Science of Getting Yourself or Someone Else Elected to Public Office for Fun and Profit?”

  8. There is a significant aspect of this issue missing in the commentary, so as a loyal subject of the, er, subject, I will pinch hit.

    Many people in the lead up to 2016 were confronted about their political views in a harsh, intimidating way. This in turn led to resentment, and people refusing to engage about what they really thought, preferring to get ‘revenge at the ballot box.’

    Pollsters (I assume push pollsters) began to confront and intimidate DURING the poll call. I had many friends who discussed this phenomenon. As a result, people either refused to answer, or gave a false answer, which skewed the results. (This was also noted with exit polls: people did not want to be accosted and thus gave whatever answer got them away from the questioner.)

    I personally had this happen several times. One pollster started arguing about my answers, insisting I was wrong. One from a major polling firm openly scoffed at my answers, and attempted to intimidate me by insinuating that he knew where I live as he had my phone number. (I invited him to a buckshot reception: he did not realize his autodialer had him calling Texas, evidently) Many others simply hung up after the first few questions determined I was not ‘woke.’

    So I understand why polling data is off to begin with: the politics of the pollsters has soured the public on answering them.

    • That’s really interesting, sw. I’ve probably only ever been called by three or four pollsters in my entire life, none ever until a few years ago. Virtually none recently. I used to wonder who the heck pollsters talked to.

      • I have a theory: not having social media accounts and not being registered Republican mean I am on the short list of ‘unengaged/independent’ voters whom they might intimidate or propagandize.

        Just a theory.

  9. Google in one of its rare bursts of irony says: “The main difference between the two is the fact that a poll is small, simple and quick. Whereas, a survey can be slightly long, open-ended and time-consuming.” I only do fun polls. The most recent one was administered by two children at my doorstep last Hallowe’en, asking me for my “ickiest” choice – among slug, cockroach, or dead squished frog – pick none, one, two, or all. Turned out it was a bet among family members but they wouldn’t say who had what (I was pretty sure the youngest, about eight? made a face at “slug” – my choice as well, having had to ruin my entire first backyard garden to get rid of them). The girls refused the candy — “mom can smell chocolate when we breathe” — accepted the grapes.

    About a week ago I just finished filling out the only (and ongoing) Survey of my latter years. Full Disclosure: $20 for filling out each annually updated version; subject is the individual. Hey, why not!

    Google, to my amazement, is wrong: the only substantive difference between the two is that the latter is (questionably) legitimate raw material for drawing hypotheses. The problem to my chagrin, is that Google is also correct — technically, a poll could cover everything in the survey if written so simple answers sufficed. The problem is that neither the polls nor the surveys cover or count the responses the polled or surveyed don’t want to answer at all, know they haven’t sufficient information to answer, do not apply in all cases, or – most often and most frustrating – are not being given a valid alternative for. (The latter is also the main problem with multiple-choice questions.)

    Fortunately, I thought, I had the ideal offer (“as much space as you require”) to add my own choices and add them at the end. That was not, as I hoped, “extra credit.”. Since some of the same “bad” (unanswerable) situations came up again and again, I called and asked whether these alternatives might be used (“unable or unwilling to answer” was one, the old Perry Mason trio: incompetent, irrelevant and immaterial was another). Instead of a clear “no” or “yes, but,” I was referred back to the template from the first survey ten years ago, originated by a dual academic/community cohort expressing a tiresome tangle of form + content. So I said I would then, in all honesty, have to skip certain questions and was told that “might invalidate the survey.” I finished a long, last rewrite of the entire survey last week, put my own postage on the expanded envelope out of plain stubborn pride. Sigh. I really could have used twenty bucks.

    Now I don’t do polls OR surveys. I will have to decide soon about the Census.

    Unless those kids come back. I could always split them a clove of garlic to hide the criminal choc fumes.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.