Friday, November 06, 2020

What the Polls Got Most Wrong

I've just been reading a very perceptive piece by Andrew Sullivan, a left of center writer generally skeptical of left-wing orthodoxy. One part of it struck me as especially interesting:

Eric Kaufmann, one of the most astute political scientists writing today, notes that the segment of the Trump vote the polling missed was educated white voters. He suspects they were afraid to say out loud to pollsters how they were really going to vote. After all, “45% of Republicans with degrees, compared to 23% of Democrats with degrees, said they feared that their careers could be at risk if their views became known.”

So the polling got the less inhibited white non-college-educated Trump voters right, but the graduates very wrong: “The exit polls show that Trump ran even among white college graduates 49-49, and even had an edge among white female graduates of 50-49! This puts pre-election surveys out by a whopping 26-31 points among white graduates.” The threat of wokeness both alienated educated white voters — and caused more of them to vote Trump than anyone expected. The problem with woke media is that they mislead Democrats who then misread the country.

Another part of the Sullivan post that I liked:

And this is where I think I have been wrong about Trump’s appeal, and where I think I’ve misunderstood why otherwise decent people could support such a foul disrupter of democratic norms. Many of them simply didn’t take Trump’s threat to our system seriously. They took all his assaults on democracy as so much bluster from the kind of car salesman he is. They deal with this kind of bullshit all the time, took liberal democracy for granted and saw little reason to fret about its future. The writer Jamie Kirchick says that everything Trump says makes sense if it is preceded by the following words: “And now, Donnie from Queens, you’re on the air.” Many people heard Trump exactly that way, and couldn’t see what all the fuss was about.

The one thing I think Sullivan, along with a good many people I'm seeing on FB, gets wrong, is concluding that the fact Trump has apparently lost says something important about the American people. Trump losing has important implications for the next four years. But in an election this close, where the result would have been reversed if one percent of the votes switched from Biden to Trump,  which side of the win line the outcome came out on says very little about the electorate.

Sullivan provides parts of his output free by email, which is how I am getting it, a larger amount, along with reader comments, for subscribers on Substack. I considered subscribing, as a substitute for Slate Star Codex until it reappears, possibly also on Substack. But unlike SSC, Sullivan's The Weekly Dish posts only a selection of reader's comments, and I don't feel entirely comfortable participating in a conversation where whether my comments appear is up to someone else. 

But I may change my mind. Certainly this post was worth reading. 


RKN said...

The one thing I think Sullivan, along with a good many people I'm seeing on FB, gets wrong, is concluding that the fact Trump has apparently lost says something important about the American people. Trump losing has important implications for the next four years. But in an election this close, where the result would have been reversed if one percent of the votes switched from Biden to Trump, which side of the win line the outcome came out on says very little about the electorate.

What's deeply disturbing to me is that tRump is even close in this election given the clear and unambiguous evidence from the past four years that he is a truly horrible human being. If a Biden win says anything about the electorate at large it's that enough of them (70+ million) agreed with this assessment and felt strongly enough about it to vote this oaf out of office.

If you find Sullivan's views on tRump worthwhile I can recommend a recent discussion between him and Sam Harris about tRump on Harris' podcast.

Ayreon said...

I don't know that you can make those conclusions from the exit polls. I would expect them to be even less accurate than the regular polls. If you're afraid to tell someone you intend to vote for Trump on the phone, why would you admit to just voting for him face-to-face?

If we can't trust regular polls too much, we can trust the exit ones event less.

I haven't investigated it too much yet, but from looking at the maps it seems like the kind of places where highly educated whites live went further against Trump.

Tim Lambert said...

Andrew Sullivan is not left of centre. He is an anti-Trump conservative, a big fan of Reagan and Thatcher. That the election result proved to him that wokeness was bad is not perceptive. He strongly believes that wokeness is bad and would have sincerely argued that the election result proved wokeness was bad no matter what the result.

I don't buy the shy Trumper argument. I believe pollsters have checked this by asking if they voted for Trump and the percentage of the people who say they voted matches what he got. More people say they voted than actually did vote so you can detect that people lie to pollsters this way. Just about whether they voted, not about which candidate they preferrred.

Benjamin Cole said...

I think you are onto something here.

No one fears for freedom of expression from the modern right wing. But evidently, on many college campuses and media outlets, one can fear the left wing.

Then we had the situation in Seattle of the Chaz zone, and many urban districts with boarded-up windows. The left-wing was put into the corner of defending these results.

The modern day hero of the left-wing is not Rosa Parks, an entirely admirable human being, but rather George Floyd.

If the GOP can find a charismatic leader who sheds most of Trump's negative peersonal traits, they should be able to annihilate the Democratic Party in 2024.

Michael Wolf said...

The refrain is that Trump's supporters take him seriously but not literally and his detractors take him literally but not seriously. While this doesn't explain everything, it can explain a lot.

For example: "We're going to build a wall, and Mexico is going to pay for it!"

*Literally but not seriously: Trump is promising to build a literal border wall across the entire Mexican/US border, and will somehow try to extort Mexico to pay for it. This is a preposterous idea, utterly ludicrous, Trump is an idiot for saying it and his supporters are stupid to believe him.

*Seriously but not literally: Trump is going to be tougher on enforcing our immigration laws and on border security in general, some of which will involve shoring up or building new border walls in certain places. Via trade deals/tariffs/other policy, it'll end up being that Mexico will essentially cover some of that cost, but it won't be that their government literally makes a cash payment to ours for the wall.

Now the latter comment may be economically incorrect for technical reasons, but you see the difference.

In 2015 I was still a little bit too inside a bubble of sorts to be able to parse this out. But if you do this with almost everything he says, the stuff he is relentless mocked and scorned for, it's usually the ones doing the mocking who don't get it. Now it's true that he's almost speaking a different language, one they don't know. But instead of having the humility to recognize that and learn how to speak it, as I did (it's really not that hard, at all), they quadruple down on their Orange Man Bad rhetoric and opinions, without understanding him or why he gets the support he does.

I don't have any delusions that Trump is playing 17D Chess or is the savior of our country as many of his supporters do. When graded without a curve, I'd give him an F, just a slightly less bad F than Obama and Bush. On a curve, I think he's better than either and would be better than Biden - but still by no means good in any truly meaningful sense. But he and many of his supporters are far less stupid, and certainly far less evil, than they are purposely made out to be for propaganda purposes.

William H. Stoddard said...

What Sullivan seems to miss is the argument that Biden and Harris, or the Democratic Party more generally, are threats to liberal values of freedom of expression.

In the first place, they have gone along with, and indeed profited from, a political climate where accusations of racism are freely made against people who dissent from progressive orthodoxy, where they can lead to storms of online condemnation, and where such condemnation can take away jobs or end careers.

In the second place, they have been all but silent about the takeover of multiple America cities by violent mobs, nominally protesting in favor of black rights, but in fact creating a situation where no rights are protected, and exposing nonwhite neighborhoods to economic collapse and increased violent crime. There's considerable evidence that these mobs consider any advocacy of non-Left views, or even simple refusal to speak, to be cause for retribution ("silence is violence"), which makes speaking out against the left actually dangerous in some situations.

There is also the complicity of social media in letting leftist statements by made without hindrance, but applying strict standards to nonleftist statements. They seem rather like the bat in the fable, who told the mouse-eating predator that he was a bird and the bird-eating predator that he was a mouse: They claim immunity from prosecution on the ground that they are neutral platforms but in controlling what content gets through, they are acting as publishers, but without being vulnerable to lawsuits for defamation as publishers are. Under Trump's presidenccy, the Senate was moving toward challenging this legal privilege; it can't be expected that such a challenge will get anywhere under a Biden administration.

The United States has long been an outlier in the West in having constitutional rules that actually protect freedom of expression (as opposed to countries like Canada and Germany where restrictions on that freedom, at the discretion of the legislature, are built into the national constitutions). That exceptional status is under threat from the Democrats. It doesn't seem ever to have been under threat from Trump, which is reason to support him.

DinoNerd said...

I'm receiving the same free material. I hope Sullivan's overall thesis is right - that the US people can collectively demonstrate some amount of what, to me, is common sense, in spite of all the obstacles to that.

OTOH, I agree with Tim Lambert, who described Andrew Sullivan as an anti-Trump conservative. Sullivan's not left wing, except maybe by the standards or your or my childhood. (He takes a certain amount of civility towards e.g. gay people for granted, like just about everyone these days.) He might be somewhere in the muddled center, or he might be outright right wing - I don't really know. But I can't see him as anywhere left wing.

I also chose not to subscribe to him, not because of selective publication of comments, but because he's not cost effective, compared to other sources of political essays - mostly news media - where I cannot converse with the essayists. I feel confirmed in this decision by the observation that he's sending me material more frequently now that he's paywalled his main material; clearly, he's trying a bit too hard to get my money.

Anonymous said...

There was a systematic polling error. Preference falsification is one possible explanation of a polling error. Identifying a more specific polling error does add credence to preference falsification, but it seems like pretty weak evidence to me.

And let me repeat the two things Ayreon said: (1) do we actually know a more specific polling error? (2) if this proposed error is what happened, we could get more evidence by looking at districts where polling errors were worse. [We don't actually have polling by district, so it is a more complicated inference.]

Anonymous said...

You should link to the post, even if it is paywalled.

David Friedman said...

I was interpreting Sullivan as left of center on the basis of other things in the post, but I could easily be wrong.

Eugine Nier said...

The most important lesson of this election is just how brazen the Democrats are willing to be in their committing of fraud.