Even in a high-turnout election like 2024, an estimated 38% of eligible Americans did not vote. Over the years, Democrats have become accustomed to explaining away their electoral losses by lamenting that they would have performed better if only more people had turned out to vote. To be sure, nonvoters interviewed by pollsters have consistently reported that they prefer Democratic candidates over Republicans. Or at least this was true up until 2024.
Our analysis of Cooperative Election Study (CES) data reveals a big shift. In his third run for the presidency, Donald Trump not only won the popular vote for the first time but also captured the support of Americans who chose not to vote at all. This finding builds on recent research, including Catalist analysis, which documented Trump’s gains among what they call “irregular voters” – people who vote inconsistently across election cycles. Our CES data confirm and extend these findings, showing that Trump held a 4-point advantage over Kamala Harris even among Americans who were eligible to vote in 2024 but stayed away from the polls.
Perhaps even more notably, these same nonvoters who broke for Trump simultaneously supported Democratic House and Senate candidates at higher rates than people who actually cast ballots.
What does this mean, exactly? These findings suggest that if every eligible American had voted in 2024, Trump would have won the presidency by an even larger margin. In congressional races across the country, however, Democratic candidates may have performed better.
The trends in nonvoter preferences
Since 2008, nonvoters have reported that they preferred the Democratic candidate for president, sometimes by a wide margin. Barack Obama held a commanding 18-point lead over John McCain among nonvoters in 2008. Four years later, Obama had a 20-point lead among nonvoters when he faced Mitt Romney in the 2012 presidential election, as the figure below shows. But the partisan gap among nonvoters has gradually narrowed since the Trump era began, shrinking to 7 points in 2016 and just 5 points in 2020.
Figure 1
The 2024 election saw the culmination of this trend with Trump leading Harris 44% to 40% among Americans who did not vote, according to the survey data. The actual national overall vote share in 2024 was 49.8% for Trump and 48.3% for Harris, with Trump claiming a 1.5-point margin. But had everyone actually voted, the CES data suggest Trump’s margin would have increased to 2.6 points (making the final tally 50% to 47.4%). Rather than helping Harris close the gap, universal turnout would have given Trump a slightly larger victory.
Nonvoters still lean towards Democrats in down-ballot races
While nonvoters preferred Trump to Harris in the presidential contest, they otherwise leaned slightly Democratic overall. We can see this when we look at party identification. In 2024, 39% of nonvoters identified as Democrats, while 37% identified as Republicans. However, this party identification advantage was much narrower than it had been a decade earlier. In 2012, about half of nonvoters said they were Democrats and only about 30% identified as Republicans.
In 2024, 90% of Democratic nonvoters still supported Harris, but Republican nonvoters were even more unified – 95% backed Trump. More significantly, independent nonvoters broke 54% for Trump. So not only were there fewer Democratic nonvoters than in prior elections, but Trump also enjoyed stronger support from Republican nonvoters than Harris did from Democratic nonvoters. And Trump claimed more support from independent nonvoters as well.
Despite nonvoters’ preference for Trump, the small Democratic advantage in party identification translated into continued support for Democratic candidates in down-ballot races. Nonvoters who expressed a preference supported Democratic House and Senate candidates by 2 points (50% to 48% – see figure 2). But in the actual voting tally, Democrats lost the popular vote in House races by about 3 points and won the popular vote in Senate contests by about 1 point.
Figure 2
However, the Democratic advantage among nonvoters in down-ballot races also seems to be eroding. For example, in House races, the survey data show the gap has been steadily closing – and shrank significantly after 2020. This suggests that Republican gains among nonvoters may be about more than just Trump’s personal brand.
These patterns reveal how nonvoters are shifting to the right across the board. As the party identification gap among this group of Americans has narrowed, they now prefer the Republican for president, and their House preferences are becoming more Republican over time. Yet, these voters did lean Democratic for congressional races in 2024, and were more supportive of Democratic House and Senate candidates than actual voters were. The question is whether this represents a temporary blip or the early stages of a broader Republican realignment among this disengaged group.
The universal turnout scenario
Importantly, Trump’s advantage among nonvoters wouldn’t have changed the outcome on election night. This is generally the case with nonvoters, as most elections are not close enough for these potential votes to tip the balance. In 2024, Trump won all six of the most decisive swing states. Nonvoters in three of these states did prefer Harris more than voters did, but it is unlikely that universal turnout would have flipped any of these states.
Figure 3
Figure 3 shows some estimates from the CES data. In four key swing states – Arizona, North Carolina, Pennsylvania, and Michigan – nonvoters preferred Trump at even higher rates than actual voters did. Universal turnout in Pennsylvania, for instance, could have increased Trump’s margin up to 4 percentage points.
In Wisconsin and Georgia, universal turnout would have moved the results closer to a Harris win. Would that have changed the election outcome? There is no evidence that the final outcome would have changed even in these two states where nonvoters preferred Harris. In any event, Trump still would have won the presidency.
The overall pattern is clear: While universal turnout could have produced some minor differences, it wouldn’t have changed the fundamental electoral map.
Additionally, we examined competitive Senate races to determine if the senatorial Democratic advantage among nonvoters would have been enough to change any outcomes. The answer is no, but several races would likely have been closer.
Figure 4
In states that elected Democratic senators, nonvoters were consistently more Democratic than actual voters, which would have expanded Democratic victory margins under universal turnout. Virginia showed the most dramatic difference. Universal turnout would have increased the Democratic margin by nearly 5 percentage points, as shown in figure 4. Similar patterns emerged in Wisconsin, Arizona, Nevada, and Michigan, where Democratic victories would have been more decisive with full participation.
In states where the Republican candidate won, the pattern was mixed. Pennsylvania stood out as the only state where nonvoters were actually less Democratic than voters, meaning universal turnout would have made the Republican victory even more decisive. But in Texas and Ohio, nonvoters could have helped Democrats, making both races slightly closer.
Despite these patterns, it’s unlikely that universal turnout would have flipped a single Senate seat. The main effects would have been making some Republican victories smaller and some Democratic victories larger, with no net change in the actual balance of power on Capitol Hill.
Why are nonvoters shifting rightward?
Over the past two decades, the CES data reveal a massive shift in the vote preferences of nonvoters, culminating in Trump emerging as the preferred 2024 candidate among this group. There is some evidence suggesting that this coalition of nonvoter Trump supporters may prove fragile. Recent polling from Pew Research shows that Trump’s approval among 2024 nonvoters has already dropped 13 percentage points since taking office.
Then again, one key reason for this shift is the growing diploma divide in U.S. elections over the same period. Less-educated Americans have increasingly voted Republican. About 80% of nonvoters do not have a college degree. In 2012, nonvoters without a college degree preferred Obama by a 20-point margin. But by 2024, these non-college nonvoters supported Trump over Harris by 7 points.
The conventional wisdom that higher turnout benefits Democrats rested on the belief that nonvoters or low-propensity voters were a reliably Democratic-leaning group. The 2024 CES data suggest that this assumption may no longer hold. If Republicans continue to win support from both voters and nonvoters, Democrats face an existential challenge that may require rethinking their entire electoral strategy. But if the observed shift is Trump-specific or a temporary blip, Democrats may find relief in 2028 and beyond.
Caroline Soler is a recent graduate of Tufts University, majoring in political science and mathematics, and is currently a research associate for the Cooperative Election Study.
Brian Schaffner is the Newhouse Professor of Civic Studies in the Department of Political Science and Tisch College at Tufts University. He also serves as a co-director for the Cooperative Election Study.
