What’s wrong with the polls on Trump
Most of the time I don’t credit the idea that polls are so very far from the truth, because they’ve often predicted results quite well. But in the last couple of years, polls in the West have certainly have been wrong in more than a few key races, including of course our very own 2016 presidential battle.
So at this point I’m inclined to agree with this article by pollster Mark Penn suggesting that the current polls on Trump are underestimating his approval, and to also agree with his suggestions as to why:
The major network polls all now report “U.S adults” as the sampling frame, not people who voted in the last election or expect to vote in the next one. The non-voters include 11 million undocumented aliens and a lot of folks who liked neither candidate and stayed home, as well as younger people who have lower rates of participation. These polls should not be confused with the views of the American electorate.
Penn adds that all indications are that Trump is holding his base. I agree. What’s more:
…[T]he media echo chamber has, I think, made it more difficult for people to express their political views, especially to live interviewers. With the growing gender gap, I’m not sure most men are even telling their spouses or partners what their real views are on the president. In a recent Harvard Harris poll we did, only about 60 percent in the country now feel free enough to express their views to friends and family.
That’s a very sad finding, but it’s one I sense is correct. My own experience reflects it. I’ve been expressing my own views on politics to friends and family for a long time, but not only have I gotten a certain amount of flak for those views, but sometimes among groups of friends people have come up to me privately and said they agree with me but don’t want to voice those views in front of the others.
That sort of thing has been going on in my life ever since my political change (which would make it around 2003), and I don’t think it’s primarily “the media [MSM] echo chamber” that has caused it. The media echo chamber certainly reflects it, and also amplifies it. But the deeper cause is the social echo chamber—both online and in real life—and that echo chamber’s increasing demonization of the other side, which has also been greatly increased in the last few years through the force and power and amplification effect of social media.
Penn adds that pollsters don’t seem to be doing much polling on the issues that are Trump’s strength:
You will find plenty of polling on what a bad idea Americans think the wall is and on the “Muslim ban” (often without even mentioning security), but where is the polling on the rest of his themes and messages? On the power of “Buy American, Hire American”? On tax cuts to stimulate jobs?
More and more people have come to distrust polls and to find them meaningless at best and/or misleading (sometimes purposely misleading) at worst. More and more people refuse to respond to them, and that skews the results as well. As far as Trump’s support goes, I don’t think polls are much of a guide to the absolute numbers—which I doubt have changed much since the election—although polls might be accurately reflecting up-or-down trends.
And if the MSM doesn’t watch its step, its relentlessly negative coverage will increase Trump’s support, if only for the sympathy vote. After all, if the MSM is crediting its own polls, it might do well to note the results of this recent one that found the public more trusting of Trump than of the media covering him. Ouch.
“. . . the deeper cause is the social echo chamber–both online and in real life–and that echo chamber’s increasing demonization of the other side, . . .” [Neo]
and because we are, at heart, herd animals. We do what we need to to be accepted into the herd or we, at least, don’t do what will exclude us from the herd.
Don Surber has a habit of asking why anyone would give any credit to polls (or a media) which have been so consistently wrong about Trump and his administration. I tend to agree with that. They have showcased their inabillty to adapt to changing circumstances, why would the polls they sponsor be any less rigid? They are like the academicians and prelates of the 16th century who still insist that the solar system is geocentric—Copernicus and Galileo be damned.
Party and partisan sample sizes make it easy for these jokers to game it.
vanderleun:
Actually, pollsters usually try to get sample size and party affiliation right. The problems lie in the fact that it’s impossible to do even if they try (I’ve written about this topic many many times before). Penn’s article lists other problems, such as surveying Americans rather than voters or likely voters. That’s one of the ways the samples get messed up, not because of size or party affiliation per se.
The polls are rubbish, and have been for many decades. This is not an opinion, it’s mathematical certainty. Anyone up on Gaussian statistics can see how the published numbers are derived, and can also see why they’re completely useless for the quantities supposedly being measured.
Sure, a poll may be right on the nose, once in a while. That doesn’t mean the methodology is sound; a dart-throwing monkey will, on rare occasions, be right on the nose, too.
There are further systematic errors introduced by analysts, as well. Recall during the last election that a poll would come out showing that Trump was doing far better than the anteceding pools had implied, and Press and Democratic panic would ensue. Then the next day the numbers would look better, as the analysts found superficially plausible ways to doctor the results to favor Hillary. The day after that, they’d have found even more creative ways to lie, and Hillary would look better and better. And soon we’d have the conventional wisdom, that Hillary would just waltz into the Oval Office. All rubbish then, and all rubbish now.
I suggest a new way of polling.
Hire people to go out in secret and listen to conversations overheard in barber shops, beauty salons, buses, coffee shops, diners, airports and so on.
When they hear a pro-candidate or con-candidate statement, they mark it accordingly for or against.
: )
“Scientific” polls are anything but. The scientific method: guess, predict, compare.
Here it is in one minute: https://www.youtube.com/watch?v=OL6-x0modwY#aid=P20kDjskZwA
There is no way to determine whether the poll is anywhere near accurate.
I agree polls have underestimated Trump’s support. If I were to speculate why, it would read much like the reasoning above.
That said, there seems some whistling past the graveyard here. Trump won by less than 100,000 votes in three Rust Belt states. In a nation of 300,000,000 people that’s miniscule.
However defective Hillary was as a candidate she could have easily won. And now we would be looking, conceivably, at Bernardine Dohrn — ex-Weather Underground terrorist, Bill Ayers’ wife, and tenured law professor at U Chicago — on the Supreme Court.
Fortunately, Hillary was a bad candidate with a dysfunctional staff which stopped internal polling for the last three weeks of the campaign. Hillary never set foot in Wisconsin, one of the three crucial states she lost.
Nate Silver got 2016 wrong, but not ridiculously so. His last call was Hillary 71.6% vs Trump 28.4%. That’s roughly 2.5:1 odds.
By Silver’s numbers, it was mildly surprising Trump won but hardly death by astonishment.
If polls show you what you want then they are good. If polls don’t show you what you want then they are bad. It is as simple as that.
Polls are actually fairly accurate so long as the methodology is explained and understood.
The polling for the presidential election were not wrong. They were being READ wrong. The media tends to read NATIONAL polls not state by state polls. And the fact is when they were making the Electoral college projections they were reading state by state polls – many of which were outdated because no one thought PA, WI or MI were in play. But nationally the polls were pretty spot on. Clinton won the popular vote. Nice. Except the election is not based on popular vote. Maybe the media and the voters will realize that now.
That said, Trump is not that popular. Although those who like him really like him.
“death by astonishment” What a great phrase, huxley!
Mrs Whatsit: I’d love to take credit but that’s from an old friend, Terence McKenna. Not that I knew him personally but I read him so voraciously he seems so.
McKenna was speaking of smoking DMT, the most powerful and short-acting psychedelic known thus far.
Sometimes people say, “Is DMT dangerous? It sounds so crazy.” The answer is only if you fear death by astonishment.
Remember how you laughed when this possibility was raised. A moment will come which will wipe the smile right off your face.
https://www.youtube.com/watch?v=mQ8t9o9d2Zg
Montage,
Not a big djt fan, although I judge him case by case, and in some cases I give a thumbs up. In hindsight, hrc ran a terrible campaign based upon hubris and a smug sense of entitlement. Her campaign was focused on the popular vote. Wasted energy and greenbacks campaigning in CA and NY, and Seatle, ect. She had little or no ground game in the states that gave djt his victory.
Oh that pesty electoral college; keeping CA, NY, and metro areas from stomping on the necks of we racist, homophobic, sexist, islamophobic, xenophobic, confused people who fail to understand chromosomes (real science) have no meaning.
By the way, the final polls came very close to getting the popular vote predication exactly right. Hillary won the popular vote by a margin quite similar to what the late polls were saying. (I wrote a post about that after the election, but I don’t have time to find it right now).
So the idea that the polls were way way off is incorrect. They were off in predicting the state by state electoral vote, however, by a significant amount. But a lot of polling wasn’t done in all the states towards the end of the contest.
What’s wrong with polls is that there are polls. Or specifically, that people worship them as idols, gods, and astrological soothsayers.
Polls are designed to manipulate humans. They are only the masters of our destiny, to those who haven’t figured this out yet.
“By the way, the final polls came very close to getting the popular vote predication exactly right.
[snip]
So the idea that the polls were way way off is incorrect. They were off in predicting the state by state electoral vote,” [Neo @ 9:34]
So the polls accurately measured the wrong thing. That’s like saying that a wrench turns a nut with precision when the tool needed for the repair is a screwdriver. That doesn’t render the wrench or the poll any more useful under those current circumstances.
Years ago, a Brit writing for National Review said he’d called a UK election for Conservatives despite contrary polling.
Nobody wanted to answer the nice middle class lady calling with the nice middle class accent that you were going go vote for those widely mocked and reviled Conservatives.
See also the Bradley Effect.
And who would believe the most fervent promises of confidentiality, even if the promiser actually meant it and claimed to have mechanisms in place to guarantee it?