In 2016, former President Donald Trump was a political outsider looking to win the GOP nomination for president. In part, his campaign sought to appeal to voters who were disenchanted with politics.
Despite that, though, Trump wasn't significantly more popular with infrequent voters than with consistent voters. In a poll I conducted in January and February 2016 via GfK's KnowledgePanel, respondents who hadn't voted in any of the three prior general elections (2010, 2012 and 2014) supported former Secretary of State Hillary Clinton by 3 percentage points. For respondents who had voted in all of those elections, Clinton's lead was only slightly larger: 6 points.
Today, though, this mild "participation gap" has become massive. Between Feb. 20 and March 18, 2024,* Gall Sigler and I oversaw a survey, fielded by NORC, of 2,462 English- and Spanish-speaking adults living in the U.S. According to public records, 63 percent of our respondents who reported being U.S. citizens turned out to vote in 2020.**
And when we broke out respondents by their voting history, we found dramatic differences in whom they support for president in 2024. President Joe Biden performed much better among frequent voters, while Trump had a large lead among people who haven't voted recently. Specifically, among respondents who voted in the 2018, 2020 and 2022 general elections, Biden outpaced Trump 50 percent to 39 percent. But among respondents who were old enough to vote but voted in none of those three elections, Trump crushed Biden 44 percent to 26 percent.
This pattern helps us understand observations about contemporary politics that seem perplexing at first blush. For instance, the party holding the presidency usually loses ground in midterm elections. But in 2022, Democrats managed to hold or even win key Senate and governor seats, and they only narrowly lost the House. Since turnout in midterms is substantially lower than turnout in presidential elections, Democratic candidates may have benefited from the party's newfound strength with more consistent voters. Relatedly, we might expect Biden's standing in the polls to tick up slightly in the coming months as more pollsters start to screen for likely voters, not just registered voters.
Still, these results are a cautionary tale for those who would extrapolate Democrats' strong performance in 2022 or recent special elections ahead to this November. The 2024 election will almost certainly have turnout far higher than those races. It will also include millions of voters who sat out 2020 — in the GfK study I mentioned above, 10 percent of validated 2016 voters sat out the 2012 presidential election — not to mention millions of people who tell pollsters that they aren't likely to vote and then do.
That said, millions of people who voted in the turbocharged turnout environment of 2020 will also likely take a pass on 2024 — and this change to the electorate could benefit Democrats. In our recent NORC poll, respondents who voted only in 2020 but skipped 2018 and 2022 leaned Trump by a large 23-point margin.
The fact that Trump did better among less frequent voters sheds light on another widely discussed trend: the declining levels of support for Democrats among Black and Hispanic Americans. Our NORC poll suggests that erosion is especially pronounced among less frequent voters. For instance, Biden led among Black respondents who voted in 2020 by 64 points, but he led among those who were registered but didn't vote by just 11 points. Biden's margin over Trump among Black citizens who don't appear to be registered to vote*** was also just 16 points. So it's not that Biden is uniformly underperforming with Black Americans — it's that he is underperforming specifically with Black citizens who don't consistently vote.
We saw similar patterns with Hispanic voters, though they were somewhat less pronounced. Among Hispanic 2020 voters, Biden was up by 8 points, while among Latinos who were old enough to vote in 2020 but didn't, he was up by only 3 points. So while there has been a lot of ink spilled about a potential "realignment" of these voters, this poll suggests that the extent to which those prognostications come true will hinge a lot on their turnout.
Of course, it's unlikely that the experience of voting itself is what's driving the participation gap. Instead, other personality traits that lead people to vote either frequently or infrequently are probably also pushing them toward different candidates.
One such trait is political engagement. Among those who follow politics closely, Trump's violations of democratic norms and false claims about the legitimacy of the 2020 election may have cost him some support, even as less engaged Americans react differently to Trump or to Biden. In our poll, even after accounting for respondents' race, partisanship and age,**** those who voted in 2022 were more likely to say that Biden "probably" or "definitely" won in 2020 than those who did not.
We might also ask a more basic question: Is the poll right? In an era of low response rates, when survey samples are often disproportionately highly engaged, it's a fair question. That said, these results came from a high-quality poll that took significant efforts to reach less engaged voters, including supplementing online surveys with phone surveys, interviewing in Spanish as well as English and oversampling Black and Hispanic Americans. Our results were also consistent with other surveys that match respondents to vote history records. And we saw similar patterns across Americans who identified as Black, Hispanic and white, a consistency that makes them harder to explain away.
Certainly, both parties will still have a strong incentive to register and turn out their core supporters this fall regardless. But while some analysts have argued in the past that nonvoters lean disproportionately Democratic, our poll suggests that, right now, that's not true. If the poll is right, it's Republicans who should be rooting for high turnout in November.
*While such a lengthy field period is unusual for campaign polls, it gives people — especially those who aren't eager to answer questions about politics — more time to respond.
**We counted respondents who didn't match to a voter file record as nonvoters, but it's possible that some of them may have voted but not matched to our survey. One thing that could cause a legitimate voter not to match to the voter file is if they have moved since the prior election; another is if their name was misspelled.
***That is, they didn't match to a voter file record but were old enough to have registered.
****Based on a linear regression.