The myth that could cost Democrats the next election

Whenever Democrats lose an election, a debate inevitably ensues over whether they were done in by an unenthused base or an alienated swing electorate.

No matter how many times this drama gets restaged, the parts and scripts remain largely the same. Center-left Democrats insist that their party must win over swing voters with moderation, while some progressives maintain that it actually needs to mobilize its disaffected core constituencies with a more radical agenda.

Pramila Jayapal, chair of the Congressional Progressive Caucus, voiced the latter perspective in a pair of X posts this week, saying, “the true swing voters don’t swing between Republicans and Democrats. They swing between the voting booth and back to the sidelines if they’re being ignored or taken for granted.” She further argued that the “true swing voter is our multiracial, multigenerational base that needs to see we’re fighting for them.”

For some progressives, Jayapal’s argument is proven by one critical fact about the 2024 election: Kamala Harris received far fewer votes than Joe Biden — and this decline in support for the Democratic nominee was larger than the increase in support for Donald Trump. As of this writing, Trump’s vote total in 2024 is 2.6 million higher than in 2020, while Harris has fallen more than 7 million ballots short of Biden’s vote total four years ago. (The latter margin is likely to narrow a bit, since slow-counting areas are disproportionately Democratic.)

What’s more, in the 47 states where the vote count is largely complete, the most Democratic counties in 2020 delivered 1.9 million fewer votes for Harris than they had for Biden, even as predominantly Republican counties generated 1.2 million more votes for Trump this year than in 2020.

Nevertheless, Jayapal is wrong to suggest that swing voters — which is to say, voters open to changing their partisan allegiance from election to election — are an extinct species. And the broader idea that Democrats only need to worry about mobilizing their base, and can best do this by putting forward a bold, progressive agenda, rests on multiple false assumptions.

If Democrats fall prey to such fallacies, they will have a harder time rebuilding a robust, majority coalition in the years to come.

There are five basic problems with Jayapal’s outlook:

1) Democratic turnout did not collapse where it mattered

At this point, it is not actually possible to know with certainty whether turnout or persuasion drove this year’s changes in each party’s vote total. That’s because we don’t yet know 1) how Americans who turned out in 2024 voted last time nor 2) whom 2024 nonvoters would have supported, had they decided to cast a ballot.

For these reasons, even in counties where the Democratic nominee’s vote tally fell by much more than Trump’s rose, it is possible that a broad increase in support for the Republican nominee drove both changes: If voters and nonvoters alike became more pro-Trump in these areas, then falling turnout might not have mattered.

That said, in such counties, it seems likely that a decline in Democratic turnout contributed to Harris’s disappointing showing. But the places where the Democratic nominee’s vote tally collapsed tend to have one thing in common: They’re mostly located in safe blue or red states.

In the places that actually mattered, Harris did not earn dramatically fewer raw votes than Biden. To the contrary, in four swing states — Georgia, Nevada, North Carolina, and Wisconsin — she actually won more votes than Biden did in 2020.

Nevertheless, despite apparently mobilizing more Democratic voters in these states, Harris lost them all. Even if population growth partly explains Harris’s gains, especially in the Sun Belt, it is extremely unlikely that we’d see this pattern of results if swing voters did not exist.

2) In the last four federal elections, millions of voters switched their partisan allegiances

Although we don’t yet know how much party-switching occurred in 2024, we have a clearer picture of previous elections. And in 2016, 2018, and 2020, millions of voters changed sides.

According to an analysis of high-quality survey data from the University of Virginia’s Center for Politics, between 6.7 and 9.2 million Americans voted for Barack Obama in 2012 and then Trump in 2016.

Two years later, Democrats dominated the 2018 midterms, winning the House popular vote by 8.6 points (in 2016, Republicans actually won more House votes than Democrats did). Although many assumed that this was the result of a Resistance-fueled surge in Democratic turnout, 89 percent of the party’s improvement derived from voters switching their partisan allegiances, according to the Democratic data firm Catalist.

In 2020, 2.43 percent of voters reported voting for the major party they had opposed in 2016, according to a 2023 study. This was an unusually low level of vote switching but still suggests that 3.8 million voters backed the Democratic nominee after supporting the Republican one four years earlier, or vice versa.

Finally, in the 2022 midterms, GOP gained ground with both rural and white working-class voters, due in part to vote switching among those who had backed Democratic candidates in 2018, according to Pew Research.

All this indicates that swing voters, as conventionally defined, very much exist. And while small in number, in a closely divided country, their shifting whims can be decisive (especially since winning over a swing voter is twice as valuable as turning out a base voter, since the former not only adds to your tally but subtracts from your opponent’s).

3) Pre-election polling indicated that a significant number of Biden 2020 voters were shifting towards Trump

In the New York Times/Siena College’s final polls of the battleground states, 6 percent of Trump supporters said they’d backed Biden in 2020, while 4 percent of Harris voters said they had backed Trump. The pollster’s final national poll showed more modest — but still significant — vote switching, with 4 percent of Biden 2020 voters favoring Trump, and 3 percent of Trump 2020 voters backing Harris.

This indicates that the same basic pattern witnessed in 2016, 2018, 2020, and 2022 — of a small but significant share of the electorate switching their partisan preference — continued in 2024.

4) The realignment of working-class and college-educated voters since 2012 cannot be explained by changes in turnout

In 2012, Democrats won 45 percent of white, college educated voters, according to Catalist. That figure rose to 46 percent to 2016 and 54 percent in 2020. Democrats also did far better with white college graduates than Obama had in the 2018 and 2022 midterms, winning 54 percent of the bloc’s vote in the first midterm, and 50 percent of it in the second.

To believe that this was entirely attributable to turnout — rather than voters switching sides — one would need to posit that white college-educated Republicans abruptly and durably became less likely to turnout for elections in 2016, while white college-educated Democrats happened to increase their turnout rate in a sudden and lasting fashion at the very same time.

This does not seem plausible. And much the same can be said of the shift in white non-college-educated Americans voting behavior since 2016. These shifts are simply too large and persistent to purely reflect asymmetric changes in turnout rates within these populations.

5) Democratic-leaning nonvoters are not especially progressive

Turnout is still a key determinant of election outcomes. Democrats would almost certainly have done better in 2024 if every Biden 2020 voter had shown up at the polls.

But there is little basis for the idea that Democratic-leaning nonvoters would turn out en masse, if only their party became more progressive.

In reality, swing voters and low-propensity partisans tend to move in the same direction. Which is to say: When a party does better with swing voters, it typically also mobilizes more of its base than the other party does. This is partly because the forces that lead a party’s voter to switch sides — and the forces that lead them to drop out of the electorate — are often largely the same.

According to a study by Ohio State University political scientist Jon Green, Obama voters who agreed with Trump on some major issues — such as immigration, climate change, or gun control — were more likely than other Obama voters to defect to the GOP in 2016, even after controlling for other variables. And the same was true of Obama voters who expressed sexist views in opinion surveys.

Neither of those findings are too surprising. More notable though, is that Obama voters who were either sexist or conservative on some major issues were also more likely than other Obama voters to sit out the 2016 election. Thus, Trump’s positioning on issues like immigration, guns, and climate not only persuaded a key chunk of Obama voters, but also seemingly demobilized other Obama voters, who no longer felt a strong preference for the Democratic nominee. (Notably, 2012 Mitt Romney voters who agreed with Hillary Clinton on some major issues were also more likely than other Romney voters to either support Clinton or sit the election out. Unfortunately for Democrats, these voters were less numerous in key battleground states than Obama-to-Trump voters).

Green’s study is consistent with other survey data showing that low-propensity Democratic voters are significantly more moderate than Democrats who reliably show up for elections. In truth, Americans who want the Democratic Party to be more uniformly progressive are, by and large, the most reliably Democratic voters in the country. It is Democrats with more heterodox views — those who are progressive on some issues and moderate or conservative on others — that the party is most at risk of losing to either Republicans or the living room couch.

This said, ideologically extreme candidates do appear to have an influence on turnout, but only because they tend to mobilize the other party’s voters in opposition, according to a 2018 study from political scientists at Stanford and UCLA.

None of this means that Democrats must embrace across-the-board centrism in order to compete for the presidency in 2028. A large segment of both swing voters and Democratic nonvoters share a deep skepticism about the political system and broadly progressive views on various economic issues. A certain version of economic populism might play well with both constituencies.

Further, Trump is likely to implement many controversial policies over the next four years and could very well undermine the economy’s performance with his trade and immigration agenda. It is therefore possible that he will solve the Democrats short-term problems for them.

But if you believe that Democrats should aim to build a large national majority, then there is no alternative to heeding swing voters’ concerns — not least because your party’s turnout targets are likely to share their complaints.