Pollsters Look to Fix Flaws After 2020 Surprises

WASHINGTON — The polling industry required an audit after the 2020 election. Poll results overstated Joe Biden’s lead over Donald Trump in the national vote for president by so much that analysts suggested: “the polling industry is a wreck, and should be blown up.” Yet in the same election, polling about specific issues of national concern wasn’t off by the same order of magnitude. 

Public opinion research has been volatile lately, but at a recent discussion on the future of the industry held by the think tank ThirdWay, pollsters argued that the system isn’t broken, it just needs a few thoughtful methodological tweaks. They also said the biggest lesson learned from 2020 is that polls aren’t predictions. 

Polling organizations attempt to collect responses on a number of closed- and open-ended opinion questions from the best representative sample of the American voting public. But while math and computer analytics are used, polling remains as much an art as a science. Subtle trends in voter — and non-voter — opinion may be equally as important as a respondent’s outright claim to support a particular candidate.

And sometimes even the pollsters are surprised. 

“In 2020, we had the highest [voter] turnout since 1900. I was not expecting that, “ said Kate Catherall, executive vice president of Political Strategy at Avalanche Insights. “I was more concerned about voter motivation… which is harder to track.”

Josh Ulibarri, partner at Lake Research Partners, did anticipate an increased participation rate but was stunned that higher turnout didn’t favor Democrats as much as they expected. 

“High turnout did not help us in all of these states, sending an absentee ballot to every voter didn’t help us in all of these states. We’ve been under the observation as progressives — and Democrats many of us — that higher turnout would always benefit us, and that’s not true in some of these states,” he said.

But while some pollsters admitted to a disconnect between polls and final votes, others suggested that it was not polling, but poll interpretation, that was off. 

“I was surprised by very little,” said Angela Kuetler, senior vice president at Global Strategy Group. “We certainly had our misses, like so many [polling firms] did… but we need to do a better job of talking about our data and educating people about how to read our data. Multiple factors… create a disconnect between what the public perceives polling to be and what polling actually is.”

Evaluating 2020 election polling, Third Way’s panel of pollsters agreed that spiking partisanship coupled with declining trust of both societal organizations and polls themselves will be a major concern for polling going forward. 

“Social trust is the big sticky problem we’re trying to deal with,” said Catherall. “I don’t think Trumpism is going away. Social distrust isn’t going away. And I think those things probably are getting worse [based on] the indicators we have.”

Though one fix to address the bias of mistrust may be to focus on the nuances of issue-based polling and ask more open-ended, and less yes-or-no questions. 

Kuetler says the public is not as divided on issues as they are on candidates, so approaching opinion polling from a different perspective and experimenting with more attitudinal questions could be the key to addressing social mistrust.

The industry also needs to take a hard look at how they can best reach out to voters and non-voters to achieve a truly representative sample. 

“Having low social trust doesn’t mean you’re more likely to answer an email, or click a link, than answer the phone, so it’s not something that can be solved with mode,” said Catherall. “There is literally no way to do this that is perfect. There will always be some bias.” Still, she does believe it is critical for the future of polling to reach people where they are.

“Fundamentally, there are people we are just missing in our polls,” said Kuetler. “Multi-modal is the way to go.” She advocated for text-to-web surveys, online panels, phone calls, and even Facebook to reach a wider sweep of opinions. 

“The future is mixed mode… [if we] take all of them, we’re hopefully getting a larger swath of the electorate; hopefully getting the right mix.”