You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Activists promoting Democratic unity at January’s Women’s March in Orange County, California.

Hate Is on the Ballot

The hidden dynamic that’s transformed our politics—and will loom large in the 2020 election

Activists promoting Democratic unity at January’s Women’s March in Orange County, California.

On election night 2018, newly re-gaveled House Speaker Nancy Pelosi presided over a celebratory press conference after the Democratic Party’s recapture of a majority in the U.S. House of Representatives. At the center of the party’s success, Pelosi explained, was its “For the People” agenda—and particularly the party’s laser-focus on health care access and the protection of preexisting conditions in Obamacare. Adding a new spin on Tip O’Neill’s timeworn political adage “all politics is local,” Pelosi triumphantly declared “all politics is personal.” This pragmatic emphasis on basic economic safeguards, Pelosi argued, had powered a historic blue wave. Ultimately, Democrats managed to flip 40 Republican-held House seats, ousting 31 Republican incumbents in the process.

Health care was a salient policy issue in 2018, and likely the most salient issue. The Republican Party’s attempt to repeal Obamacare had produced raucous town halls across the nation and sit-in protests on Capitol Hill throughout 2017. In a suitably dramatic endgame for a law that dominated American politics for a decade, the repeal measure was voted down around two o’clock in the morning, with Arizona GOP Senator John McCain’s downturned thumb drawing shocked gasps from his Senate colleagues. On the stump, Pelosi’s red-to-blue Democratic hopefuls stressed the GOP’s efforts to take down the Affordable Care Act to the exclusion of almost everything else, especially Donald Trump. Aside from a few renegades, most of the 2018 Democratic hopefuls avoided talking about Trump. Democratic consultants warned candidates that focus group data suggested voters wanted the election to be about “more than just Trump,” and the Democratic candidates weren’t keen on riling up Trump’s faithful base.

Once the final seat tally was known, the takeaway from the Democrats’ success in the House was that the strategy had worked. By focusing on a key “bread-and-butter” issue, Democrats had flipped districts Trump won in 2016 by getting moderates and disaffected Republicans—particularly college-educated, suburban women—to break ranks with the GOP in favor of Democratic pragmatism. It was a nice, tidy explanation for what happened. But it is also largely wrong.

To be sure, Republican and independent voters who supported expanding or preserving Obamacare were more likely to cast ballots for Democratic House candidates. Data from Democracy Fund reveals that 23 percent of Republicans who wanted to see the law expanded reported they planned to vote for a Democratic House candidate. But here’s a hitch: There were also Democrats against the ACA, and 31 percent of these Democrats indicated they intended to vote for a Republican House candidate—more than enough to offset any Democratic gains on the health care issue alone. These numbers also don’t square with exit polling, which finds that 95 percent of Democrats voted for a Democratic candidate for the House, while 94 percent of Republicans voted for a Republican House candidate.

But there was something very different about the 2018 cycle—and it’s something that the health care thesis fails to address. That something was voter turnout. After waning to historic lows in 2014, voter turnout in 2018 reached historic highs, smashing even the very high expectations set for the cycle in my own forecast. The 53.4 percent turnout in 2018 was closer to what we typically see in a presidential election than a midterm cycle. It dwarfed the 2014 cycle’s midterm turnout of 36.7 percent by nearly 17 points. And for all the talk of how central health care was that year, it was certainly not any more salient than it had been in the other elections since Obamacare was first enacted in 2010—all cycles marking solid-to-historic Republican gains in Congress. In the fateful 2010 balloting—the Tea Party movement’s great coming-out moment—the GOP netted 63 House seats, even while the nation’s economy continued to hemorrhage from the Great Recession that began under President Bush.

Indeed, if the health care policy debate should have been driving turnout in any major recent cycle, it was in the 2016 election, not 2018. The 2018 congressional battle over health care only came to pass, after all, because Trump ran hard on the pledge to “repeal and replace Obamacare,” and won.

Despite the genuine possibility that Obamacare could be replaced, overall turnout remained close to that of the 2012 and 2008 elections, and actually declined among key components of the so-called Obama coalition. Obamacare’s very survival was on the 2016 ballot, and the electorate broke out into a sustained collective yawn.

So what, then, really drove the dramatic surge in voter turnout in 2018? It happened for one simple reason—or, rather, because of one simple man: Donald J. Trump. Trump’s surprise victory on election night 2016 set into motion conditions that all but guaranteed Democrats would take back control of the House of Representatives two years later, even as the GOP managed to hang on to a narrow majority in the Senate.

My forecasting model for the 2018 midterms predicted an enormous Democratic wave in the House, mostly by focusing on a dynamic known in political-science circles as negative partisanship. The idea behind negative partisanship is simple, harking back to Henry Adams’s definition of politics as the “organization of hatreds.” The determination to vote out the opposition—and the broader trend of acute polarization within the American political system—has altered virtually every facet of our political life. Negative partisanship is affecting the behavior of voters and reshaping the voting coalitions aligned behind each major party.

Negative partisanship is also the reason why the pending 2020 presidential and congressional cycle doesn’t call to mind charged modern ideological battles such as 1964 and 1972 so much as the fateful election of 1860, which ended up kicking off the Civil War. Then, as now, the nation and its elected leaders were divided into two sharply opposed factions, harboring deep-seated cultural and philosophical resentments toward each other. Then, as now, each side gravitated toward intractable positions while showing little appetite for continued compromise. And then, as now, demagoguery and nativism had reached a fever pitch, political civility a new low, and conspiratorial thinking had defiantly overtaken the political mainstream. And then, as now, all these forces converged to trigger widespread institutional breakdown under the strain of political conflict, with profound results for the country’s future moral direction. Slavery and sectionalism were, of course, the forces that scuttled the antebellum status quo in 1860. Today’s pitched mood of political crisis stems from more diffuse sources of cultural change.

It’s become commonplace to date the origins of today’s confrontational political mood to the rise to power of Newt Gingrich’s Republican Revolution in another landmark midterm election in 1994. But here, too, the negative partisan turn in our politics has deeper roots—it stretches back to the 1950s and 1960s, a time of great consequence in American political development. Back then, the long-running, and dominant, Democratic coalition was beginning to fall apart. The tensions long assailing the national Democratic Party finally burst open with the passage of the Civil Rights Act of 1964 and the Voting Rights Act of 1965, ending more than 30 years of an ethically unsustainable alliance between Northern liberals and right-leaning Southern segregationists. The Dixiecrat revolt of the late 1940s plunged the party into two decades of electoral turmoil that would ultimately lead to the Reagan revolution; the Gingrich-led capture of Congress—also significantly conjured by the fractious politics of health care—served as an important capstone in this transformation, ending the Democrats’ decades-long run of dominance on Capitol Hill.

Cultural changes produced additional varieties of political polarization. Second-wave feminism liberated women from reproductive captivity, while upending traditional gender roles and patriarchal assumptions about social power. Legislation such as the Higher Education Act of 1965, and particularly the successful 1972 Title IX amendment to the legislation that added protections for sex discrimination, made it clear that not only could the problems women were facing “be named,” they could be remedied via activism, lawsuits, and federal legislation. The Warren and Burger courts sharply redefined the state’s relationship with citizens, and the state’s relationship with religion, reinforcing the separation of church and state by outlawing prayer in schools and other publicly sanctioned displays of faith. And as if all of that wasn’t enough, the Immigration and Nationality Act of 1965 abolished the “national origins formula,” opening mass immigration up for the first time to nonwhite Europeans.

This wide-ranging battery of postwar changes in our cultural and political scene also got outsize attention in a new media landscape, from the rise of cable television and talk radio. And the internet’s emergence fundamentally changed the media environment, leading to the proliferation of “outrage media”—outlets that prospered by driving traffic to their platform via scandal, controversy, and conspiracies and teaching Americans to hate their neighbors.

Katie Porter (right), one of the Democratic House members swept into office on the blue wave of the 2018 election, joins the Orange County Women’s March.

Against this backdrop of roiling cultural upheaval, our politics has remained rigidly binary—and indeed has grown far more brittle over time. Partisans in this culture struggle have sorted themselves ideologically into two parties that are now devolving into bitterly opposed enemy factions in the 1860 model. In 2014, pollsters with the Pew Research Center found that 27 percent of Democrats and 36 percent of Republicans believed their partisan counterparts were a “threat to the well-being of the country.”

Now, just three years into the Trump era, and a whole new brand of negative partisan division, my survey finds that these indexes have skyrocketed. Seventy-one percent of Republicans and 55 percent of Democrats now regard the opposition party as a force that stokes baleful national decline. In a study of what they call “lethal mass partisanship,” political scientists Nathan P. Kalmoe and Lilliana Mason found that 15 percent of Republicans and 20 percent of Democrats agree that the country would be better off if a large number of opposing partisans in the public today “just died.”

In view of the GOP’s intense loyalty to Trump now, it’s hard to recall that the nomination of Trump as the standard-bearer of the Republican Party was the culmination of a hostile takeover of the party. It also defied most every form of conventional wisdom about who can win a presidential primary. This was especially true for the “party decides” hypothesis popularized in 2008 by political scientists Marty Cohen, David Karol, Hans Noel, and John Zaller. Not only had Republican elites failed to exert influence on the selection of the party’s nominee, they were completely railroaded by the party’s base, turning the theory of “elite dominance” in nominee selection on its head.

Few elected Republicans attended the 2016 Republican nominating convention, let alone agreed to speak at it. Trump was a pariah among Republican elites even though Republican voters loved him. In the general election, Trump defied every principle of electability. He was crass, rude, unapologetically unqualified. He claimed a judge had been unfair to him in a civil suit because of his Mexican ancestry, told his supporters to rough up protesters at his rallies, and viciously attacked a Gold Star family. But none of these things fazed Republican voters, because the most important thing Trump had going for him was the “R” next to his name, and an uncanny ability to remind people that as bad as he was, the “enemy” was worse. In the darkest days of his campaign, in the week after the release of the Access Hollywood video in which Trump gloats about grabbing women “by the pussy,” Trump would whine to his rally audiences about how much they needed him, that he was their last line of defense against “Crooked Hillary” and a Democratic Party who wanted to end their way of life. He was right, of course, to deploy this strategy. His highlighting of the alternative, Hillary Clinton, tapped right into polarization’s key ingredient: negative partisanship.

Despite these all too plain and all too radical trends of negative partisanship, election analysts have been slow to adjust to the new polarized era. Nor do they seem to appreciate the longer-term historical context in which individual election cycles play out. The parties have transformed over the past 100 years—once urban, the Republican Party has become a rural party. Once rural, the Democrats have become an urban party. And America’s suburbs are now our premier political battleground.

One vital consequence of this set of regional convulsions is sure to play a pivotal role in the 2020 cycle: America’s suburbs were once a reliable bulwark for the GOP in much of the country, but are now realigning toward the Democrats, as college-educated whites grow disenchanted with the Republican Party.

Long before Trump announced his long-shot presidential run in 2015, the GOP was hemorrhaging younger voters, especially within the past two generations of college graduates: millennials and Generation Z. Trump’s surprise victory in the 2016 election was a boon for the party’s short-term policy interests and efforts to stack the federal judiciary with conservative judges—but it was also the death blow to the GOP’s efforts to appeal to the most racially diverse and best-educated generations in the country’s history.

The millennial generation became fully enfranchised for the first time with the 2016 presidential cycle. And the front end of Gen Z reached voting age at the same time. Together, millennials and Gen Z voters are poised to make up as much as 37 percent of the electorate in 2020.

And both generational cohorts are still recovering from the economic body blows of the 2008 meltdown. Front-end millennials are nearing their forties—they have mortgages, kids, and crushing student loan debt. As people mature and their responsibilities increase, so does their propensity to vote—as we can plainly see in the cases of the boomers and Gen Xers.

Volunteers gather signatures at the Orange County Women’s March.

While millennials may be finally coming into their own, politically speaking, the members of Gen Z have a long journey ahead of them. The youngest cohort of voters (under 24) have poor turnout rates. Voters in this age bracket averaged just a 43 percent turnout rate between 1976 and 2016, compared to the 65 percent turnout rate of voters over age 25 and the nearly 80 percent turnout rate of the oldest cohort.

Indeed, turnout remains relatively low for all voters under 40. The population pyramid and the voting population pyramid have in many ways become inverted. This is what’s allowed silent voters and boomer voters to make up the majority of the electorate, even though they represent a smaller portion of the country’s overall population.

All told, there are 160 million prospective millennial and Gen Z voters out there; they represent a reservoir of potential Democratic voters who, as a simple question of math, must inevitably overtake the aging, white, and conservative one.

But this wishful liberal talking point has consistently failed to account for the strong patterns of political behavior within each voting demographic. In their rosy 2002 prediction of inevitable Democratic dominance, for example, Ruy Teixeira and John B. Judis overlooked pervasively low voter turnout rates within the Democrats’ modern coalition of lower-income, minority, and younger voters. While Republican turnout proves to be more steady across election cycles, Democratic turnout fluctuates wildly between presidential and midterm elections, as well as between state-level and national-level elections.

That’s one key reason that the Obama coalition vanished from the electoral scene almost as rapidly as it emerged as a potential major political force. The consequences of this crash were readily apparent just two years into Obama’s presidency, with the Democrats’ devastating losses in the 2010 midterms. In that disaster, Democrats sacrificed 63 House seats; in 2014, they went on to lose nine Senate seats (and their majority in that chamber). Democrats carried 57 percent of independents in the 2006 midterms, but just 37 percent in the 2010 midterms—pretty much changing places with their Republican counterparts in the independent camp over that four-year span.

The GOP dominated independents in 2014 as well. But here’s something you almost never hear an analyst say about independents: Although they are important to election outcomes, they are a necessary, but by no means sufficient, element in a successful electoral strategy.

Even though we commonly assume that independents make up roughly a third of the electorate, the pool of persuadable independents is actually quite small, just 7 percent of the total electorate, according to the Pew Research Center’s most recent analysis. This is because most independents admit to being “leaners”—bringing them into fairly reliable affiliation with the Republicans or the Democrats. Research shows these leaners think like, feel like—and, most important, vote like—“soft partisans.” In fact, many leaners are what political scientists Samara Klar and Yanna Krupnikov call “embarrassed” partisans—people too ashamed to admit their partisan dispositions even to themselves.

This dynamic leaves a quite small pool of “pure” independents truly open to persuasion, especially late in a campaign season. That’s especially the case after voters have been reminded why they vote for one party, and not the other, by a barrage of campaign appeals that key into their latent partisan dispositions.

It’s also important to understand that polarization and the electorate’s increasingly hyperpartisan outlook cause high levels of inelasticity in voting behavior. We see this play out daily in the static polling on Trump’s approval numbers. These figures are virtually unresponsive to events, even to dramatic shifts in the political landscape such as Trump’s impeachment. We can also see this bedrock level of partisan attachment in election outcomes such as the special Senate election in Alabama in late 2017. In that contest, Roy Moore, the Republican nominee, faced credible accusations of child sexual abuse but still managed to accrue 48.3 percent of the state’s vote share. That outcome furnished a prime case study in polarization, with fully 91 percent of the state’s Republican electorate voting for Moore.

It’s true, of course, that partisanship has always shaped public opinion, buffering politicians through scandals and providing a floor of support for candidates in general elections. Even so, however, the power of partisanship has steadily increased over the past decade. On average, 90 percent of Republicans vote for their party’s nominee in any given election, while 90 percent of Democrats do the same. There are exceptions, of course. But partisanship overwhelmingly dictates the voting choice for most people in most elections—and so, ultimately, the main factor driving election outcomes is the relative proportion of Republicans and Democrats (and their respective leaners) in the electorate. That means the party that best turns out its own voting coalition will have an edge.

Keep in mind as well that the American system is marked by strikingly low rates of voter participation, while presenting great variations in turnout across cycles, contest-types, and states. Much of the movement in vote share between parties from one cycle to the next is thus driven not by the shifting preferences of a relatively fixed pool of voters, but by the movement of voters into and out of the electorate. Of course, this steady churn through cycles of participation also includes the entrances and exits of independents.

And what drives people to vote? For many voters, especially intermittent or newly engaged ones (i.e., people activated by the election of a new president), negative partisanship is a critical factor. And due to the polarizing logic of the present electoral landscape, it’s easy to behold the power of negative partisanship nearly everywhere—in news cycles, social media feeds, campaign debates, and impeachment proceedings. But for ordinary voters, negative partisanship chiefly follows the threat response they register in reaction to the opposition party’s control of government.

Under the spell of negative partisanship, the campaign arms of both parties, their candidates, and their supporting groups all work in concert to reinforce and validate this set of emotional reactions. (Some more effectively than others, it should be noted. Republicans are particularly adept at emotive campaigning. This is because Republicans understand “soft” voters are most likely to be activated by “stakes” messaging—emotive appeals that tap directly into the threat response that characterizes negative partisanship. Democrats appeal to people’s heads, Republicans to their guts, and most voters are not politically cerebral.)

Based on my own research, I’ve concluded that negative partisanship is more than simply a generic background feature in most two-party contests. Indeed, it looks increasingly to be the main driver of the long-established pattern in our modern politics known as the “midterm effect”—in other words, the process by which the president’s party typically loses congressional seats in midterm elections.

The midterm effects of the past two decades have been particularly powerful due to the interlocking forces of polarization and negative partisanship. As turnout surges and declines, the demographic composition of the electorate changes in ways that favor one party or the other. During the Obama era, for example, Republican voters were fired up and Democrats complacent, feeling comfortable with their guy in the White House. But once Trump was elected, the pendulum swung the other way, firing up Democrats and resetting the electoral map. This is the deeper story of the 2018 wave election that went well beyond a standard-scale correcting “effect.”

You can get a clearer picture of how negative partisanship can amplify turnout and alter electoral outcomes by looking at the recent electoral history of Virginia. Back in the state’s 2013 gubernatorial election, when complacency within the Democrat’s coalition reigned supreme, Democrat Terry McAuliffe barely eked out a narrow victory over the controversial Republican nominee Ken Cuccinelli, thanks in no small part to a “spoiler” Libertarian candidate, Robert Sarvis. His 6.5 percent share of the vote helped obscure an important Democratic breakthrough: For the first time in five decades, the state’s governorship had not gone to the party opposite the president’s party.

Even though the educated suburban population swelled in northern Virginia, and the state capital of Richmond got swept up in a tech boom and a wave of gentrification, Democrats still struggled in their 2014 effort to reelect incumbent Senator Mark Warner. But here, too, there was easy-to-miss but significant trend: Warner’s narrower-than-expected victory came in a cycle that virtually wiped all the nation’s remaining Southern “Blue Dog” Democrats out of office. The party also underperformed expectations in Virginia’s 2015 state legislative cycle, gaining just one seat in the House of Delegates and none in the state Senate.

Regarding the 2017 gubernatorial race, most analysts agreed that, given Trump’s election, Virginia was likely to reinstate its pattern of selecting a governor from the party opposite the president. But it turned out that the political observers in the state were dramatically underestimating the potential size of the Democrats’ advantage in Virginia. This was true both for the marquee gubernatorial race and in that year’s many down-ballot contests for the House of Delegates, where the Republican Party held what looked like an insurmountable majority.

The demographic muscle that the party never managed to effectively harness during the Obama years was now flexing itself for the first time. It’s hard to overstate the impact of this shift. In three short years, Virginia’s entire political order was transformed. Democrats flipped three seats in the House of Representatives despite a partisan gerrymander in 2018. And over the course of two state legislative cycles, the party netted a total of 21 seats in the House of Delegates and two state Senate seats. Democrats now have a liberal governing trifecta in the former capital of the Confederacy.

Those three pivotal Virginia House seats combined with 37 other Democratic pickups in the House of Representatives in 2018. But the story of exactly how these districts flipped has been largely untold. Heading into the 2018 cycle, most election analysts recognized the strong potential for a blue wave, at least in the House. The polling on the “generic ballot” question, gauging voter support for a Republican or Democratic House candidate based solely on party affiliation, indicated a large, persistent advantage for Democrats throughout 2017 and 2018.

But it was only after the great Virginia realignment that most analysts understood a critical truth of the negative-partisan political world: America’s suburbs would be at the heart of any wave election.

Even with such forecasts trending strongly in the Democrats’ favor, it was still difficult to grasp the larger dynamics and the potential impact change coming to the House. When I released my House forecast in July 2018, the generic ballot had been in the midst of a narrowing period between April and June. The consensus take was that Democrats probably had a shot at netting the 23 seats they would need to take control of the House of Representatives—but not much more, in all likelihood. This prediction was in turn grounded in the recognition that 23 Republican-held House seats had broken in favor of Hillary Clinton in the 2016 election and thus were prime pickup opportunities. But the analysts couldn’t make out what was happening in a second set of districts, ones that broke in Trump’s favor in 2016, but still bore some resemblance to those Clinton districts.

These Trump districts had competitive partisan voter indexes (PVIs), which are measures of the level of party competition in each district, higher rates of racial diversity, and, most important, highly educated populations. Although Trump carried them all, he fell shy of Mitt Romney’s 2012 vote share in them. (In more rural districts, Trump overperformed Romney’s vote share, evidence of a long-term realignment in the rural interior now broadly known as Trump Country.) My research showed that these districts met the baseline demographic requirements to produce the type of turnout surges needed to break in favor of Democrats. And due to the galvanizing power of negative partisanship—Trump’s gift to the Democrats in 2018—a turnout surge seemed likely in these districts. My model thus successfully identified these districts as “temporary” Trump districts. They were, to put things a bit more technically, districts that broke Republican in 2016 because Democrats had underperformed in turnout due to in-power complacency, particularly among younger voters and voters of color. In the transformed electorate produced by negative partisanship, they still profile to break in favor of the Democratic presidential candidate in 2020. In final 2018 balloting, Democrats nearly swept these “soft Trump” districts. Yes, independents also played an important role in this shift—but again, thanks largely to negative partisanship, it’s a different role than you might think.

In today’s polarized era, election outcomes increasingly reflect, in quite a direct sense, the partisan composition of the electorate. With a few notable exceptions, the side that wins such contests will also be the side that makes up a plurality of the overall electorate—including independent leaners.

And, in turn, the voting public’s partisan makeup is determined by its demographic makeup. Is the electorate in question more white or nonwhite? More educated or noncollege-educated? More male or female? And perhaps most important, younger or older?

I’ve analyzed voter file data to demonstrate the powerful effect that turnout has on the demographic (and thus partisan) composition of congressional electorates. The “Wave in the Making” chart compares turnout of demographic groups in 2018 in California’s competitive House district 45 to the turnout of these same groups in the 2014 midterms.

Again, 2018 saw historic turnout surges, dramatically reshaping that midterm electorate from what it was in 2014, an election cycle that Republicans dominated. As you can see, turnout surges in 2018 broke within demographic groups that heavily favor Democrats, such as young voters, Latinos, and women. If we look closely at how California’s 45th congressional district in Orange County flipped to the Democrats, we can chart how turnout surges among these groups changed the electorate in that district. In 2014, the 45th district electorate heavily favored the GOP—and by 2018, that same electorate had achieved a competitive measure of partisan balance.

And this point opens, in turn, onto another crucial trend within the voter file data: Turnout among independents surged, too. In district after district, the turnout among independents skyrocketed in 2018 over 2014 rates. This suggests something very important—and heretofore largely overlooked— about the voting behavior of independents: When preferences among this voting group swing wildly between the parties between cycles, some of that movement is the product of turnout surge and decline. In other words, when their votes swing, this shift is not at all static; it is caused by different independents expressing different preferences. And independents are also very sensitive to negative partisanship.

This understanding of independent voting behavior changes, well, everything. It suggests that although some independents are moved by persuasion, others—and likely a great many others—are moved by the same forces that motivate their partisan counterparts to go to the polls.

And here we encounter still another vital shift in the electoral landscape. Although turnout surges powered Democratic victories in every 2018 House race (apart from a handful of California races), most Democratic campaigns struggled to match Republican turnout, even with a wave at their backs. Although Democratic turnout surges are proportionately much larger than the GOP’s typical turnout gains, the margins of victory for these winning Democrats relied heavily on a turnout surge of independents. That meant that they won their House seats by narrower margins than they would have if the Democratic turnout had been stronger.

Some of this difference, no doubt, is because of the California Democratic Party’s efforts to improve access to voting. Even as red states have erected new roadblocks to voting, blue states like California and Oregon are using Democratic majorities to tear similar obstacles down. You can readily see the results of these efforts when you compare the Democratic coalition turnout in California to the levels reached in every other state’s 2018 House races.

But turnout and demographics don’t win elections by themselves; variant outcomes also track to a striking degree to differences in strategy. Candidates who ran as liberal Democrats rather than more traditional “Blue Dog” campaigns outperformed their more moderate counterparts in terms of base turnout. This sharper partisan appeal has allowed them to come close to equalizing the partisan composition of the district. Like their more liberal counterparts, Blue Dog candidates still experienced large turnout surges among independents. But the winning margins they enjoyed proved to be tighter than those in districts featuring Democratic candidates who saw their party hit the same percentage level—in a larger pool of Democratic coalition voters—that Republican turnout had.

This suggests these candidates may have paid a penalty by not focusing more effort on increasing turnout among their own voters and concentrating instead on independent and even opposition-party voters. This strategic choice is framed, in part, by the idea that consciously downplaying signs of Democratic affiliation will avoid upsetting Trump voters, and thereby will not inflate GOP turnout. But the data shows that Republican turnout surged in these Blue Dog districts, while Democratic turnout lagged compared to levels in other districts. In other words, my research suggests that there may be no benefit, only cost, to such a strategy.

To hear many Democratic leaders tell it, ignoring Trump was the secret to their success in 2018, but the voter file data suggests otherwise. Democratic gains, strong though they were, may actually have been handicapped by a strategy that failed to exploit the party’s best asset: the electorate’s angst about Trump. What’s more, the turnout of the Republican base was just as strong in these districts as in districts where candidates were more liberal and did talk about Trump. So the strategy of not rousing any partisan blowback in the general election doesn’t appear to have yielded the advantage of a suppressed opposition vote.

However, one party did frame 2018 as a referendum on Trump—and it wasn’t the Democrats. Again disregarding every traditional piece of conventional wisdom for presidential conduct in a midterm, Trump leaned into the referendum effect. While Democrats all but disowned Obama in 2010 and 2014, Republicans hugged Trump closely—in no small part, of course, because they were given little choice in the matter. Trump rallied for 36 GOP House and Senate candidates and tweeted about an “invasion” staged by caravans of immigrants approaching the country’s southern border, deep-state coups, and Russia witch hunts. He also exhorted his supporters to remember that he was on the ballot.

This looked to be a recipe for disaster, and it should have been, according to traditional campaign models. But Republican turnout in 2018 actually improved over the party’s 2014 turnout, even though the GOP controlled all three branches of government. The punditocracy’s fable of a key corps of disaffected Republicans breaking for the Democrats in 2018 is wrong—in much the same way that Nancy Pelosi is mistaken in her narrative of health care pragmatism swelling the blue wave. Republicans delivered for Donald Trump, as Donald Trump instructed them to. They were simply outnumbered in certain places—and almost everywhere they were outnumbered, they lost because of the turnout surge within the Democratic voting coalition.

What do all these negative partisan trends portend for 2020? Even though most analysts continue to look at 2016 as a frame for 2020, that reflex bespeaks another failure to size up a dramatically shifting electorate. The complacent voting ranks of 2016 who believed (at the energetic prompting of polls and pundit forecasts) that Donald Trump could never be president have been replaced by the terrified electorate of 2020. These voters know all too well the hazards of granting great power to a figure like Trump and view the president as a Terminator-like political figure who simply can’t be stopped. After threading the 2016 Electoral College needle, Trump has acquired an almost mythic aura of invincibility for most people, election analysts included. And in some key ways, the unsettled character of this political moment—together with the structural inequities of the Electoral College and a battery of GOP-orchestrated voter suppression drives—has inoculated President Trump from political gravity. But as Andrew Yang would tell you, math matters, and Trump has a basic math problem. As the electorate is currently constituted, there are more potential Democratic voters out there than there are Republican, and not just in California. There are more in the Midwest and in the Sun Belt. There are so many more in Virginia and Colorado that both states have moved off the swing state map.

The 2020 election will be a battle of the bases, with nothing less than the country’s survival as a functional democracy on the ballot. Partisanship is a hell of a drug—especially when it’s cut with a heavy dose of existential fear.