Ariel Edwards-Levy is a senior reporter and HuffPost's polling editor. As you can imagine, she's been busy trying to tell us what polls tell us about, well, us. And she does it with unique humor and accessibility. So far, she's unpacked why polls underestimated Donald Trump yet again. She's also looked at why voters support both liberal ballot measures and conservative politicians, and how people are feeling about the government response to an unprecedented third wave of COVID-19. Her Twitter feed is also brimming with other insights and puns.
Must Reads talked to Ariel about polling and 2020.
So, were the polls wrong again this year?
Yes, I think that's pretty unambiguous. It's still too soon to know the exact magnitude of the error, which is going to be important. At the national level, at least, the presidential results may end up only several points off, which would be well within historical precedent.
But especially at the state level, we're seeing significant, systemic misses in terms of underestimating support for Donald Trump and other GOP candidates. Given that it's the second presidential election in a row where that's been an issue, that is obviously going to get a lot of well-deserved scrutiny.
What's going on there?
It's way too early to offer any definitive answers at this point, but broadly speaking, there are at least two buckets of possibilities to look into. One is that pollsters got the turnout modeling in this election wrong — for instance, underestimating how many GOP voters would end up casting ballots. Given the way that the pandemic reshuffled voting this year, that seems distinctly plausible. The other has to do with pollsters' ability to draw a representative sample of the public.
One concerning and also quite plausible theory is that a certain type of Republican-leaning voter has stopped responding to polls at all, and that this bloc isn't limned along any neat demographic lines that could be simply ironed out with weighting. This theory is distinct from the popular idea that these voters are responding to polls but actively lying to pollsters, which is not entirely outside the realm of possibility, but it's something most pollsters I've talked to view as a less-likely explanation.
Ultimately, there may not be any singular root cause to point to. Instead, there might be a cavalcade of smaller issues, in some cases differing between pollsters, modes and states, and some will probably be more easily fixable than others.
What drove you to polling? Like, why would you do that to yourself?
It was sort of an accident of fate! When I was finishing up an internship and looking for a job, the polling position happened to be what was open. I knew nothing about polling, but figured I'd give it a go. Why I ended up staying with it and what I find valuable about it is pretty similar to what attracted me to journalism in the first place: I love getting to talk to people around the country, to hear about what's happening in their lives, what they care about, and how they're following and reacting to current events. Public opinion research, ideally, is a way to do that at scale, and add a sort of macro-context to the individual stories we report on.
Do you have a preferred Nate? How do polling editors feel about other polling editors?
I have to put in a plug for my similarly named former HuffPost colleague Natalie Jackson, who's now doing thoughtful work at the nonprofit Public Religion Research Institute. I think everyone working at the intersection of journalism and survey research is having to grapple with some really complex questions.
If you've done this for a while, one thing you become painfully aware of is the inherent uncertainty around polling, and the real limitations on what it can usefully tell you. Trying to communicate that both internally and to the public is a central challenge with no easy answers. So, generally speaking, I respect people who are trying to do that work.
What are you watching most closely going forward?
From the mechanics-of-polling standpoint, I'm obviously concerned with getting a better sense of what drove the 2020 error. On the public opinion side, the start of the new administration is going to be a fascinating time. We've seen a lot of public liberalization on issues like immigration, free trade, the role of the government during the Trump era, and one theory has been that that shift came partly out of opposition to Trump. I'm curious to see the extent to which those shifts hold under a Democratic administration.
|