Something’s Going Wrong With The Polls When Trump is On The Ballot

This year, none of the big prognosticators nailed that brag-worthy, March Madness-bracket-esque, rub-it-in-your-face perfect electoral map forecast.

This is a companion discussion topic for the original entry at
1 Like

I found the polls really annoying
This is what I mean an when I say “annoying”…
Fivethirtyeight, Nate Silver’s polling site with uses an aggregate of 40,000 scenarios to come up with an estimate, statistical estimate of the Electoral College votes got it wrong. From what he posted in 11/1/2020 it looked to me to be a Biden blow out which is what I very much was looking forward to. Here is his graph…

Note Biden’s projected estimate of EV. It shows clearly 348 votes. That is not a close election by any measurement I am aware of. And if you look carefully there is an 80% estimate range. By 11/1 Biden was almost clearly separated from trump. Given this I assumed (incorrectly as it turned out) that trump was gonna get his fat ass spanked bigly. But Nate got it wrong. The same thing with polls happened in 2016. How can polls be trusted after this? Biden’s actually EV is supposed to be 306 votes.

I hate it when I make spelling mistakes


Maybe the problem isn’t with polls, but with our expectations of how accurate polls should be. Predictions about the future are tough, after all.


Dave Wasserman consistently rang a bell about the Trump team’s efforts to register and turn out the millions of white non-college voters who sat out 2016. For whatever reason, most pundits and pols paid no attention:


If I had anything to say about it, there should be a harsh reckoning about why the polls underestimated candidates in the same direction and similar magnitude in the same states in the last three election cycles,” Morris said.

Hmmmm, Highly improbable…cheating GOP


ikr. Trump supporters are liars, too? Who woulda thunk it?



I think the thing that most folks missed was the huge numbers 45 got
probably skewed Likely voter polls.

On a side note it is pointless for the base and the left wing of the party to be fighting over 2020 down ticket results. Trump getting 10 million more voters than in 2016 was not on anybody’s bingo card. I thought we hit peak stupid in 2016.

I was naïve
I was wrong


Breaking not News!!!

WASHINGTON —Donald J. Trump was reportedly furious after the White House’s Human Resources Department notified him that it had scheduled his exit interview, the head of H.R. has revealed.

Carol Foyler, the H.R. chief who was the target of Trump’s wrath, said that the exit interview is a “valuable tool” to help make the White House a better working environment.

“Everyone who leaves the White House has to participate in an exit interview,” she said. “That was true of the nine thousand people who left during the past four years.”

The H.R. executive said that she hoped that the interview could help Trump explore any workplace issues that led to his departure.

“The previous three people who had his job held it for eight years and he only lasted four,” she said. “It would be helpful to get his perspective on why things didn’t work out for him here.”


I find the various theories all somewhat persuasive, but I hope to heck that either Trump is a sui generis problem, or pollsters figure out some what to accommodate for non-response bias or whatever.

Political twitter has been aflame with “polls are dead, quit even trying!!!”, which is obviously completely absurd. The best way (albeit a clearly flawed way) to predict who people will vote for is to ask them who they plan on voting for. The alternatives are stupid bullshit like counting yard signs or crowd size at rallies, or just not even attempting to predict the winner at all. These are obviously much worse.


Personally the’ low social trust’ seems the most reasonable explanation, hatred of institutions, the press and others in general is most definitely on the rise for republican voters. But I’d bet this is also the last one pollsters are willing to entertain since it break their fundamental principle that the types of people who will answer a poll are not politically different from the types of people who won’t as a whole. That idea has never sat right with me, but it has seemed to work out until recently.

Also I have a hunch that they way right-wingers use social media to campaign may play into this too. Too long to get into here but basically the ‘Cambridge Analytica’ model of using the massive amounts of social media data on a person and very specifically tailoring ads for just them based on psyc models is giving them and advantage with voters that traditionally doesn’t poll. (this I think is the big yet seemingly invisible elephant in the room Dems need to address with how they campaign).


“These voters don’t trust institutions or each other, and feel pressured by society to hold certain beliefs so they don’t speak openly about theirs,”

May be we should take this as social commentary rather than humor:


Once again, 538 published a probability distribution, not an estimate of 348. 348 is the median value in the probability distribution graph you posted. You looked at a graph that said 50% chance of more than 348 for Biden, and 50% of less than 348 for Biden, and yelled “I SEE A PONY! I’M GONNA GET A PONY!”


IMO the issue is Trump’s racism appeals to FAR to many, and they are embarrassed about it.


538 is not a polling site


And while these kinds of people may have always distrusted institutions and declined to talk to pollsters, it’s become more of a problem now when they’re increasingly siloed in one party.

It’s a problem for the GOP too in the long run.


Andy Borowitz is a national treasure.


You are, of course, completely correct. As a whole, American understanding of probability and statistics can never over-disappoint me. It’s just pitiful and pitiable.


538 had a close Biden win as a fairly probable option.

It got the state level stuff way wrong. Cough Collins cough.


Article should have delved into the enormous misses on senate polls. Collins overperformed by 18%. IA, AK, missed by double digits. MT, NC, SC plausibly within the margin of error, but all missed on the same side.


One thing that I never want to hear about again is the angst of the
“ undecided” voter. When Attila the Hun is running against Tom Hanks and people say they are undecided, they are lying.

Also, the reliance on phone calling is absurd in the current plague of nuisance phone calls. We don’t answer the landline that we keep as a junk number. Like so many others, we just check for messages. Pollsters need to figure out a new paradigm. Clearly, the old one ain’t working.