Republican presidential nominee Donald Trump arrives at a campaign rally on November 8, 2016, in Grand Rapids, Michigan.
CNN  — 

President Donald Trump’s political base may not be as impregnable as commonly assumed. And that could have big implications not only for the Democrats’ strategy against Trump in 2020 but also for their choice of a presidential nominee.

Detailed new research by the Democratic voter-targeting firm Catalist found that the party’s big gains in the 2018 congressional election were fueled not only by unusually high turnout among voters sympathetic to the party, but also by larger-than-expected defections from the GOP among voters who had backed Trump two years earlier.

Those findings offer potentially critical evidence as Democrats are debating the best approach to beating Trump in 2020. On one side are progressive activists who say the party should prioritize mobilizing nonvoters, particularly young people and minorities, with an unabashedly liberal agenda. On the other are centrists who say Democrats can’t tilt so far left on issues such as single-payer health care and the Green New Deal that they alienate swing voters who backed Trump in 2016 but may be open to reconsidering now.

Sens. Bernie Sanders of Vermont and Elizabeth Warren of Massachusetts, most recently at last weekend’s California Democratic convention, have made the former case, while former Vice President Joe Biden, along with several second-tier hopefuls such as former Colorado Gov. John Hickenlooper and Sen. Michael Bennet of Colorado, have most explicitly identified with the latter argument.

Rather than picking one path, the new Catalist data on 2018 signals that Democrats need to do some of both in 2020. But, on balance, its analysis found that a clear majority of Democrats’ gains from 2016 to 2018 came from voters switching their preference, rather than from changes in the electorate’s composition.

“The number one thing I would say is winning elections isn’t just about mobilization,” said Yair Ghitza, Catalist’s chief scientist, in an interview. “I do think that’s something some people argue, and it’s gained a bit of traction. What I try to point out here is that mobilization is incredibly important. But the idea that there are literally no swing voters left, is, I think, a misreading of a lot of the data that’s out there.”

Catalist analyzes electoral results by studying state voter files on who actually voted in each election. The state files provide precise information on some demographic characteristics of the voting population – particularly gender and age – and Catalist uses advanced statistical techniques to fill in its portrait of data that is available only in some places, such as race and education. Then it combines its turnout data with polling analysis and precinct-level results to produce its estimates of how each group in the electorate voted.

Its approach represents an alternative measure of voting behavior to the more familiar Edison Research exit polls (CNN is a member of the consortium that produces those polls), but for 2018 the two methods produced similar results on voting behavior for the key groups.

In the new analysis, Ghitza sought to quantify how much of the improvement in Democratic performance in 2018 compared with 2016 had resulted from changes in who voted versus shifts in preferences among the voters.

In 2018, Catalist calculated, Democrats won the total popular vote in House elections by 7 seven percentage points (after making projections for uncontested races). That was a gain of about 5 percentage points from Hillary Clinton’s popular-vote margin over Trump in 2016.

That change derived from three big sources: who left the electorate between 2016 and 2018, who entered it and the changing preferences of voters who participated both times.

Gains from switches by voters

The falloff from voters who participate in the presidential election but then sit out the next midterm has become a huge problem for Democrats as their coalition has grown more dependent on young people and minorities; both of those groups turn out much more reliably in presidential than in midterm elections.

That’s meant that considerably more Democratic than Republican voters typically stay home in the off-year election. That falloff was so severe from then-President Barack Obama’s re-election victory in 2012 to the GOP sweep in 2014, for instance, that Ghitza calculates it cost Democrats fully 6 points in their share of the total vote.

Even comparing 2018 with 2016, more Democrats than Republicans stayed home, Catalist found. But because of the turnout gains among key party constituencies, that drop-off was much less of a problem than it typically is for Democrats. About 40% of all voters who participated in both the 2008 and 2012 presidential elections stayed home in the next midterms; but in 2018, only about 27% of 2016 voters sat out, Catalist found.

Moreover, the mix of voters who fell off last year was less lopsidedly Democratic than in the past: Republicans also suffered a drop-off, particularly among non-college whites and rural whites, two of Trump’s key groups. Minorities, who usually slip as a share of the midterm vote, represented almost exactly as much of the vote in 2018 as they had in 2016. And while young people still declined as a share of the electorate in 2018, they did not do so nearly as severely as in the previous two midterm elections.

The overall result was voters who sat out the 2018 midterms after voting in 2016 cost the Democrats a manageable 2 percentage points in the total vote last year, only about one-third of their crushing decline in 2014.

And last year, Democrats offset that loss through the other major factor that shifted the electorate’s composition: new voters. Catalist found that about 13% of the 2018 voters, some 14 million people, had not voted in 2016. That was a significantly bigger surge of new voters than in 2010 and 2014, when about 9% of the electorate had not participated in the previous presidential election.

And while the new voters had favored Republicans by 2 percentage points in 2010 and by a solid 7 percentage points in 2014, they provided Democrats a resounding advantage of 21 percentage points last year.

In all, Catalist calculated, new voters swelled the Democrats’ total share of the 2018 vote by about 2.6 percentage points. When combined with their loss of around 2 percentage points from 2016 voters who sat out 2018, that meant changes in the electorate’s composition contributed about half a percentage point to their overall vote gain from the presidential election to the midterm elections. That was a vast improvement from the midterm elections under Obama, when Democrats were hurt by the composition of both the drop-off and new voters.

“There was a massive turnout boost that favored Democrats, at least compared to past midterms,” Ghitza wrote in a recent Medium piece explaining his research.

But by itself, that roughly half-point improvement in the Democratic vote from changes in the electorate’s composition would not have been nearly sufficient to drive the party’s sweeping gains last year.

“If turnout was the only factor, then Democrats would not have seen nearly the gains that they ended up seeing,” Ghitza wrote.

In fact, Catalist calculated that nearly 90% of the Democrats’ increase in their total vote from 2016 to 2018 came from switches among the roughly 99 million people who participated in both elections. Catalist projects that Democrats virtually broke even among those two-time voters in 2016 but won them by about 5 percentage points in 2018.

Vote switching, as opposed to shifts in the electorate’s composition, accounted for about three-fourths or more of the Democrats’ improvement compared with the 2016 presidential results in a wide variety of states, Catalist found. Those included their Senate victories in Pennsylvania, Michigan, Wisconsin, Ohio, Nevada and Arizona, as well as governor’s victories in Nevada, Michigan and Maine.

Turnout and persuasion both matter

Catalist’s research methods didn’t allow it to estimate how many of the voters who moved to the Democrats in 2018 had switched from voting for Trump in 2016 versus how many switched from supporting one of the third-party candidates, libertarian Gary Johnson and Green Party nominee Jill Stein. But Ghitza said that estimates he’s seen in other private Democratic analyses of the election are that Trump voters accounted for about three-fourths of the switchers and former third-party voters about one-fourth.

The overall level of movement from Trump doesn’t signal that his base is cracking. But it does suggest that it’s a mistake to view all of his voters as immovably locked in behind him.

“He got votes from people who aren’t considered his base in 2016,” Ghitza said. “And so this could certainly be consistent … with the idea that he has a floor and those people aren’t leaving him but there are still swing voters out there that could be had.”

These new findings aren’t likely to end the roiling debate among Democrats about whether to emphasize mobilization of base voters or persuasion of swing voters in 2020. Comparing the 2018 vote with the results in 2014, as opposed to 2016, tilts the outcome so that mobilization looks relatively more important, Ghitza notes.

And the likelihood that total turnout could increase significantly in 2020 – Catalist projects that over 15 million more people may vote next year than in 2016 – will also encourage Democrats who want to prioritize mobilization, since many newly eligible voters are young and nonwhite.

The big question for that strategy is whether the Democratic message – and nominee – can inspire more turnout from those groups without becoming so polarizing that they help Trump mobilize his own core supporters, or alienate the swing voters who abandoned him in 2018.

“There’s something to that argument” for mobilization, Ghitza says. “But it’s clear that it is best to both mobilize and persuade, and to find a message that can do both. If the Democratic candidate has a message that only appeals to certain pieces of the country, then those mobilization advantages could end up being counteracted by increasing support for Trump on the other side.”