Adam Bonica, Rachel Funk Fordham, Jake Grumbach, and Ernesto Tiburcio pose the above question and answer No. Their key evidence is in these two graphs they made from the 2024 Cooperative Election Study, showing reported party identification and reported vote preferences for people who said they voted, people who said they were registered to vote but did not vote, and citizens who said they were not registered to vote:

As Bonica et al. point out, you can’t take those vote preferences of the registered nonvoters too seriously–the very fact that they didn’t vote offers some evidence that their preferences weren’t so strong–but, still, to the extent these numbers are to believed, this suggests that, when it comes to election time, the best strategy for the two parties is not persuasion but mobilization (or counteracting the mobilization efforts of the other side):

– Democratic campaigns should put more effort into persuading their supporters to turn out and vote.

– Republicans campaigns should put more effort into dissuading Democrats from voting.

Retrospectively, it seems that Republicans have been more effective than Democrats at persuading their supporters to come to the polls: the Republicans have left far fewer registered voters “on the table.”

One open question is, what are these respondents’ voting histories? Registered nonvoters include people who are newly registered but then didn’t vote, along with people who’ve been registered and haven’t been voting for a long time, and everybody in between. Both parties are always trying to register new voters, but that won’t get you much if those new registrants are voting at a low rate.

The last time we looked at this question was in a post from 2007, What difference would it make if everybody voted?. We shared three findings from surveys:

1. Mark Baldassare compared likely voters in California to those who were not registered to vote, and found that the nonvoters had much more negative views about the governor. That makes sense: I’d expect nonvoters to generally be more pessimistic about politics. Yes, you will have some angry people who turn out to throw the bums out, but in general we think of voting as something that’s done by people who feel more engaged with the political system.

2. Looking at data from the 1990s, Benjamen Highton and Raymond Wolfinger estimated that, if everyone who was eligible to vote had voted, Bill Clinton’s vote margin would have increased by about 4 percentage points in 1992 and about 5 percentage points in 1996. They report this as, “voters differ minimally from all citizens; outcomes would not change if everyone voted,” which seems strange: a shift of 4 percentage points in vote margin is a lot, no? In retrospect, I think Highton and Wolfinger were overly influenced by the era in which they were writing: From 1980 through 1996, there were five presidential elections, none of which were close. Since then, though, nearly all the presidential races were decided by less than 4 percent of the vote. Also there’s Congress, where the Republicans maintained control of the House of Representatives with very slim margins in 1996 and 1998.

3. Jan Leighley and Jonathan Nagler report that, nonvoters have much different attitudes on several key issues, compared to voters. They argue that even if increased voter turnout would not result in much change in vote margins, it would shift politics by changing the salience of issues and motivating candidates to take more populist stances.

The work of Bonica et al. is different from what I’ve seen before in that they focus specifically on people who are registered to vote but don’t turn out: this is a relatively small subset of the nonvoters but is important because these are people who could be affected by mobilization in the final stages of the campaign.

To get a handle on what’s known about this group, I googled “registered nonvoters,” and the following items popped up:

– This 2017 report from Pew Research, “How ‘Drop-Off’ Voters Differ From Consistent Voters and Nonvoters,” which defines “nonvoters” as “those who were registered to vote, but did not cast ballots in any of the most recent national elections.” They report party ID as 51-47 Republican among consistent voters, 40-58 among drop-off voters, and 37-58 among nonvoters. This, from 2012-2016, is consistent with Bonica et al.’s results.

– A New York Times article from 2019 reporting “an analysis of 623 registered voters who live in battleground states and stayed home in both 2016 and 2018.” They look at hypothetical 2020 vote preferences and find no differences compared to people who vote. This is a small sample, it’s a year before the election, and it’s only in a few states; still, it provides some evidence on 2020 being different.

– A 2025 article from the American Enterprise Institute reports from a 2024 Pew survey that “The top reason for not voting, at 35%, was the belief that their vote wouldn’t count. Almost as many, 31%, said they didn’t like the candidates. . . In the Pew survey, 94% of self-identified voters said voting for them was very (79%) or somewhat (15%) easy in 2024. Two percent said it was very difficult and 5% somewhat difficult.”

– Here’s the link to the aforementioned Pew survey, which also reports that “Among those eligible to vote who say they did not cast a ballot, 42% say they wish they had voted while 57% say they do not. These shares are similar to other recent presidential elections: 45% of nonvoters said they wished they had voted following the 2020 election, and 44% said this postelection in 2016.” I don’t see anything there on the party identification or candidate preferences of the nonvoters.

Juxtaposing this last finding with the results from the CES, the question arises: how were the Republicans so much more effective than Democrats at getting their occasional supporters to show up and vote? This seems like a consistent pattern:

– As noted above, a Pew survey estimated that, of the registered nonvoters in 2016, 37% were Republicans and 58% were Democrats.

– The CES breakdown of registered nonvoters in 2024 was 29% Republicans, 53% Democrats.

– According to this Pew report, nonvoters in 2020 preferred Biden to Trump, 50% to 35%, but they were split in partisanship (“46% were Democrats or leaned Democratic, while 41% were Republicans or leaned Republican”). This is all citizen nonvoters, not just registered nonvoters, but still it seems that 2020 was different than 2016 and 2024.

– This 2025 PRRI report shares this estimate from the 2024 race:

They don’t distinguish between registered and nonregistered nonvoters.

– This pre-election Pew poll from 2016 reports, “Nonvoters favor Barack Obama over Mitt Romney by a wide margin (59% to 24%) . . . . Nonvoters express more liberal opinions than voters on a wide range of domestic and foreign policy issues. But the differences are far less pronounced on social issues, such as gay marriage and abortion.” Again this is not restricted to registered nonvoters.

Why is this a controversy?

For the longest time, it’s been recognized that frequent voters are more Republican and have more conservative positions on economic issues compared to occasional voters and nonvoters. This is what Bonica et al. found in their analysis of the 2024 CES, and it’s consistent with most of the other survey data referenced above, with the possible exception of 2020.

In their post, Bonica et al. contrast their findings with a recent New York Times article that featured an analysis by David Shor, who pointed to a poll from May 2024 that estimated Trump and Biden being close to tied among people who had voted in 2020, but Trump having a 15-percentage-point lead among respondents who had not voted in 2020. Bonica et al. argue that Shor got a different answer because he was analyzing just one poll which was taken while Biden was still running, also they say it’s “unclear if non-voters in that May poll truly represent the millions who ultimately stayed home in November.” Also there’s a difference between nonvoters in 2020 and nonvoters in 2024: that’s another moving target.

Bonica et al., and all the other sources other than Shor, estimate that nonvoters continue to be more Democratic than voters, both in party registration and candidate preferences. This pattern can be explained in two ways which sound completely opposite!

Explanation 1: Regular voters are more likely to support Republicans; irregular voters and nonvoters are more likely to support the Democrats. This is consistent with the historical pattern of Republicans being wealthier and more politically engaged.

Explanation 2: Republicans have been more effective at mobilizing occasional voters. This is consistent with the idea that Republican candidates in recent years have a special appeal to less educated and disaffected voters, an appeal we associate with Donald Trump but which seems to have been there in 2012 as well.

The funny thing is how different these two stories sound. The first explanation has engaged Republicans outmobilizing disaffected Democrats. The second explanation identifies the Republicans as appealing to the disaffected. Same data, same statistical inferences, opposite stories.

Bonica et al. write that in 2024, “higher turnout would have helped Democrats: Democrats turned out at a lower rate than registered Republicans (83% to 90%), and non-voters leaned Harris.” Again, this is consistent with the traditional story that Republicans are more politically engaged than Democrats, and it’s also consistent with the new story that Trump has been effective at getting less-engaged people to come to the polls.

I ran this by Don Green, who expressed concerns about relying on CES. After survey adjustments, you’ll get the right vote proportions by state, but subgroup analyses can be questionable because of differential nonresponse. But the general pattern of nonvoters being more Democratic than voters seems to be borne out by most of the other surveys.

Don also reminded me of the difference between aggregate and marginal. Whatever the Republicans or Democrats do to mobilize a potential voter (or to dissuade a potential voter from the other side to turn out) will be affecting the marginal voters: those people who are on the borderline of whether or not to vote. These marginal voters will not in general look like the aggregate population of nonvoters, or even the aggregate population of registered nonvoters.