Voter data by TargetSmart Communications.
In late February, Barack Obama stood before a room of his party’s governors at a Washington fund-raiser and offered a new explanation for the Republican rout that claimed the jobs of more than 750 Democratic officeholders around the country in 2010. At the time, the president had described the outcome, simply and indelibly, as a “shellacking,” but here he ventured a deeper analysis. “We know how to win national elections,” he told the crowd. “But all too often it’s during these midterms where we end up getting ourselves into trouble, because I guess we don’t think it’s sexy enough.”
Beyond the narcissism implied—the suggestion that any ballot without his name on it lacks a certain magnetism—Obama was onto something. Current conventional wisdom holds that Democrats’ prospects this November are grim. After the obligatory acknowledgment that the party in the White House almost always loses ground in off-year elections, the most commonly cited reasons are situational—the botched Obamacare rollout, a zealous conservative base, the fact that these midterms follow a redistricting process largely controlled by Republicans, the preponderance of competitive Senate races in states that lean red. And yet Obama’s diagnosis of Democrats’ midterm woes comes closer to the truth. The party is suffering from a chronic condition, not a short-term malaise.In fact, the very phenomenon that sustained Obama’s own victories is the one that may doom his party in midterm elections for the rest of his life. The dynamic so cripples Democrats’ off-year performance that in 2018, a President Jeb Bush or Rand Paul could see Republicans actually pick up seats.
A decade ago, Obama memorably rebutted the trope that the United States could be neatly cleaved into a red and a blue America that pits coastal liberals against inland traditionalists. But in one very measurable and consequential sense, there are two Americas. There is the America that votes in presidential elections, which has helped Democrats win the popular vote in five out of the last six cycles and supports the view that Hillary Clinton can continue that streak should she run. Then there is the America that votes more regularly, casting ballots in both presidential and midterm years, which led to the Republican wave in 2010 and gives its party’s leaders reason to be so sanguine about their odds this time around.
There are about 127 million people in that first category, and among their number is the ascendant coalition—young and diverse, urban and mobile—that now gives Democrats a huge advantage in presidential races. But only 78 million of those people, or about 40 percent of the country’s voting-age population, belong to the group that goes to the polls every two years, and those regular voters carry a considerably more conservative cast. (The number of unregistered voters is almost as large.)
Over the past four years, the consequences of this schism have made themselves clear. A Democratic president is handed a progressive mandate by a convincing electoral-college victory. But he has his agenda unilaterally obstructed by a Republican House empowered by the right-leaning midterm electorate—an electorate that also disadvantages Democratic Senate candidates and sustains Republican governorships and state legislative majorities. Indeed, Democrats are facing an inverse of the four-decade span in the late twentieth century when the party controlled the House of Representatives and largely dominated the Senate but suffered through three two-term Republican presidencies. The bad news for Democrats is that the imbalance could take a generation to work itself out naturally. The good news is that, thanks to a newly nuanced understanding of the voting brain, they know exactly what it will take to fix it.
In 1981, a young political scientist at the University of Georgia named James E. Campbell received a query from Newt Gingrich, who represented his state in Congress. Could the professor prepare a report for the Republican National Committee about the 1934 midterms? At the time, that election was the answer to a trivia question: the only midterm since the Civil War during which the party controlling the White House gained seats in Congress. But Gingrich thought Republicans developing his party’s 1982 strategy could learn from a more detailed account. Campbell, who had briefly worked as a Republican staffer on Capitol Hill, returned with an explanation that flattered his client’s sense of the moment: When Franklin Roosevelt introduced the New Deal during his first two years in office, he set in motion a partisan realignment that insulated Democrats from a midterm backlash. Gingrich and other ambitious Republicans hoped that Ronald Reagan’s election would prompt a similar realignment in their favor.
Case studies of historical anomalies have limited career value for political scientists, whose discipline focuses on rules, rather than exceptions. After submitting his report, Campbell decided to turn his attention to the pattern itself. What was the real reason, as he put it, that “like clockwork,” midterm elections reduce the number of seats held by the president’s party?
Most of Campbell’s fellow scholars subscribed to the “referendum” theory of midterm elections. Presidents, the theory went, are swept into office on hopes that they inevitably fail to fulfill; in midterms, the voters who swung their way swing back in disapproval. (This thesis was developed by Edward Tufte, now better known as the philosopher-king of infographics.) Journalists and politicians still favor this interpretation, since it’s easy to view the messy array of federal, state, and local elections as a reflection of the familiar face in the Oval Office. Democrats were walloped in 2010, goes the thinking, because the country soured on Obama’s stimulus and health care bill; Republicans withstood such a fate in 2002 because George W. Bush had successfully rallied public opinion in the wake of 9/11. Every midterm outcome is supposed to say something meaningful about what the country believes.
Before Tufte came along, however, another interpretation of midterm outcomes prevailed. This theory was known as “surge and decline,” and it was introduced in 1960 by University of Michigan political scientist Angus Campbell. (The Campbells are not related.) Angus Campbell had been part of the team that produced the canonical book The American Voter, and in understanding midterm dynamics, he thought it made sense to separate the electorate into two groups: core voters who regularly cast a ballot and had developed partisan loyalties, and peripheral voters who are less concerned with politics but may be activated if the stakes seem high. Campbell’s peripheral voters are swing voters in the loosest sense: Not only do they alternate between parties, but they also drop in and out of the electorate. Presidential campaigns, Campbell argued, “bring a surge of peripheral voters to the polls,” and the candidate who draws more of them wins. In the next congressional election, most of those peripheral voters stay home, and so the party in the White House fares worse.
James E. Campbell ultimately concluded that both models are right, but they apply to separate constituencies. For independent-minded voters without strong loyalties, off-year congressional elections are indeed a referendum on the incumbent president. Partisans, however, follow a surge-and-decline dynamic—if they become disenchanted, they won’t defect to the other party. They simply don’t vote in the next election.
In 1994, Gingrich gave James E. Campbell and his fellow political scientists another confounding case study. Under the then–minority whip’s strategy, Republicans won 54 House seats, taking back the chamber for the first time since 1954. Republicans also captured eight Senate seats, giving them a majority there as well. None of the models—neither referendum nor surge-and-decline nor Campbell’s hybrid—predicted such a wave. With the notable exception of Gingrich himself, Campbell observed in his book The Presidential Pulse of Congressional Elections, “[N]o one expected Democrats to lose so many seats” that year.
Up until then the political science models had assumed that both parties’ coalitions contained a similar mix of regular and sporadic voters. But the political realignment that began in 1994 has disrupted that balance. Today the Republican coalition is stacked with the electorate’s most habitual poll-goers—or “Reflex” voters, as we will call them. The Democratic Party claims the lion’s share of drop-off voters, or “Unreliables.”
Unmarried women, for instance, are one of the most consistently Democratic blocs, and they are also among the least likely to vote in non-presidential elections. Latinos, too, have become an indispensable part of the Democratic coalition, making states like Colorado, North Carolina, Virginia, and potentially Arizona newly competitive. But while Latinos’ total presidential votes tripled from 1988 to 2012, their midterm participation has declined by about seven points. Obama won reelection in part because in 2012, African Americans for the first time turned out at a higher rate than whites. In 2010, black turnout trailed white turnout by nearly five points, and the election outcomes reflected that divergence.
Socioeconomic factors help to explain why turnout patterns vary so much by age, race, and marital status. The higher rates of mobility and displacement experienced by minorities (and the young of all races) unmoor them from the localized social networks that raise awareness and interest in civic affairs. Moving introduces transaction costs, too: registering at a new address, identifying the correct polling place, and so on. Conversely, owning a home tends to make individuals more politically invested in their communities and thus more likely to vote regularly. It is no coincidence that whites, who have high rates of property ownership, are also the electorate’s most reliable off-year voters—their turnout dropping by only two percentage points over the last quarter-century, to 48.6 percent during the last midterms.
At the same time, voting (or not) tends to be self-reinforcing. Records show that people who have cast ballots tend to maintain the habit, while campaigns and parties naturally prioritize potential supporters with spotty voting histories over those with no history of voting whatsoever. “It’s not rocket science that if a person votes in every election, you don’t need to knock on their door,” says Kevin Arceneaux, a Temple University political scientist. The inverse is just as true. One who has never voted is far less likely to receive a visit from a campaign volunteer.
Add it all up, and the Democrats’ midterm conundrum comes to look like an actuarial one. “If twenty years ago, you said the midterm electorate is older, I would have said, ‘Yahoo! Glad to hear it,’ ” says Celinda Lake, a Democratic pollster. “But now the Roosevelt seniors are dead and the Reagan seniors are voting.” Increasingly, those older voters are backing the same side: In 2000, Al Gore won the youngest and eldest bands of the electorate by slight margins; in 2012, the over-50 vote broke for Mitt Romney by 12 points.
There are also simply more of those older voters overall. Since Obama’s first appearance on a presidential ballot, the population of Americans over the age of 55 has increased by nearly 13 million. By 2022, it will have increased by another nine million. People tend to grow more conservative as they age, but as a cohort, Generation X—whose oldest members will soon reach their fifties—is appreciably more conservative than the Millennials who follow them. “When the Millennials are fifty-five, they’re going to vote more Democratic,” Lake says, not exactly cautioning patience. “That’s thirty years away.”
Until very recently, the entire campaign apparatus was structured to cater almost exclusively to winning over people already likely to cast ballots. Politicians, who tend to have an outsized sense of their persuasive powers, like to imagine elections as a grand debate before a riveted citizenry. Journalists are willing collaborators, eager to see themselves as chronicling a contest of conflicting ideas and narratives. Media consultants (who make their money on advertising commissions) and pollsters (whose work fixates on what those ads should say) set budgets and strategies tilted overwhelmingly toward swaying the existing electorate. The result has been a political establishment united under a broad bipartisan consensus: Elections are won and lost through ad wars.
That consensus is now being undone, particularly among Democrats and their allies. In the late ’90s, political scientists began using field experiments to test the effectiveness of basic campaign methods. Their work was gradually embraced by groups on the left, who in 2007 formally adopted a more empirical approach to electioneering with the launch of a secretive research consortium called the Analyst Institute. The Democratic National Committee and Obama for America were active participants, and one of the political scientists who helped to launch the Analyst Institute, Notre Dame professor David Nickerson, became director of experiments for Obama’s reelection. What began as isolated academic inquiries has built to a revolution in campaign tactics.
The problem with political ads, from the empiricists’ perspective, is that there’s little proof of their impact. Despite the colossal sums spent on them, it’s nearly impossible to measure the effect of any individual TV or radio spot, because it is hard to randomize its placement and calculate its reach. Individually targeted tactics like direct mail, phone calls, and canvass visits, however, are easy to test, particularly when measuring methods to register and turn out voters. Over the last 15 years, hundreds of experiments have yielded a clear understanding of which of those methods perform best. The most effective techniques now appear on a frequently updated Analyst Institute best-practices sheet that has become mandatory decor in Democratic field offices.
Accordingly, field operations have been transformed from busywork for volunteers into the most rigorously scientized corner of the trade. All the research suggests that the most effective form of outreach is also the most seemingly old-fashioned: a conversation on a doorstep between a potential voter and a well-trained volunteer. Experiments have even pointed the way toward the best kinds of volunteers; canvassers can be most successful when they’re reaching out to non-voters of the same ethnicity or from the same zip code. Productivity has been tabulated, too. Surveying a decade’s worth of experimental research for the 2008 edition of their book Get Out the Vote, Yale professors Don Green and Alan Gerber calculated that a typical canvasser can complete six encounters per hour. Assigning a monetary value to that labor makes it possible to put a price on democracy. One new vote mobilized through fieldwork costs $29.
Few candidates, however, inspire volunteer corps large enough to sustain such an ideal mobilization campaign, and many voters live behind doors that are simply not reachable. High-quality phone banks whose salaried callers can mimic a volunteer’s sincerity (those specializing in fund-raising solicitations seem to do best) are too scarce to handle millions of election-season calls. The solution has been direct mail, a relic of twentieth-century electioneering whose economics nonetheless match twenty-first century imperatives. The start-up costs to print leaflets and attach postage are relatively small and scale naturally, making direct mail equally suitable to communicate with a small pool of voters for a city-council race or a vast presidential universe. And unlike in digital communication, matching an analog mailbox to the identity of the voter who owns it is a straightforward process.
More importantly, it turns out that some of the psychological power of face-to-face encounters can be translated to paper. Todd Rogers, a Harvard psychologist who served as the Analyst Institute’s first executive director, dug into behavioral-science journals for insights on human motivation that could be tested in campaign contexts. Experiment after experiment has since confirmed the effectiveness of subtle prods that trigger what Rogers has called a citizen’s “basic need for belonging.” Addressing the recipient as “a voter” or “the type of person who votes” (a message born of a theory known as identity salience) produces a small increase in turnout. So does asking people to commit to a plan for when, where, and how they will vote (implementation intentions). Emphasizing that many other people will vote in an upcoming election (social-norms theory) has been proven more effective than bemoaning those who don’t show up. Added together in a single nonpartisan get-out-the-vote letter, the messages can boost an individual’s likelihood of voting by about one-third of a percentage point without increasing costs. Factoring in printing and postage, new votes can be created this way for $71 each.
Gerber and Green have achieved even more striking results by sending out letters that threaten to distribute neighbors’ vote histories before and after Election Day. But few political organizations are willing to be associated with such a blunt approach. Rogers wanted to see if he could find a more palatable way to deploy social pressure to push people to the polls. So before the last midterms, he enlisted one of the Analyst Institute’s most reliable research partners to let him run another test.
In 2010, the America Votes consortium planned to send 800,000 pieces of mail in targeted congressional districts. Rogers, working with his colleague John Ternovski, randomized those letters so that half featured the proven language and half included that message plus an additional sentence in the upper right-hand corner: “You may be called after the election to discuss your experience at the polls.” (A control group received no mail at all.) Rogers and Ternovski were testing the potential of a new concept—self-integrity—by threatening accountability for potential voters who valued civic engagement. Their simple adjustment increased the letter’s impact by more than 50 percent and generated about 1,500 votes across the experiment. The cost of a new vote dropped to $47.
Such results undercut the popular belief that Unreliable voters are driven to the polls by passion—either about a given candidate or the general political climate. Pollsters imbue this so-called intensity gap with near-prophetic powers: In mid-October 2012, for instance, the Politico–George Washington University Battleground Poll reported that Republicans led Democrats by a ten-point margin among those calling themselves “extremely likely” to turn out. But that didn’t prevent Obama’s reelection, of course. Similar findings about this year’s midterms (Battleground has Republicans up by seven points in the enthusiasm category now) will likewise reveal little about the returns come November.People cast ballots for reasons that have nothing to do with their excitement level. For Unreliable voters, specifically, it often takes a psychologically potent encounter to jolt them out of complacency.
If Democrats fail to see midterms as sufficiently sexy, the problem may lie not with the party’s rank-and-file but with its donors and activists. The strategists engineering the party’s campaigns now have at their disposal databases containing the names of every Unreliable voter in the country, as well as guidance on where, how, and when they can be reached. (Democratic analysts have developed predictive models to anticipate which voters are most likely to actually open and read their mail.) Volunteers who live near those passive sympathizers can be dispatched; when in-person contact is unfeasible, carefully crafted letters can be sent instead. But all of these increasingly powerful tools also require money and manpower. This is why it’s not intensity scores on polls but rather the bustle of field offices and the sums on fund-raising reports that are the best guide to the Democrats’ midterm prospects. When those indicators sag, says Mike Podhorzer, the AFL-CIO’s political director and chair of the Analyst Institute’s board, “the effects are cascading.” For a party populated with Unreliable voters, the midterm imperative is clear: Raise the dollars and secure the volunteer commitments. Then go and turn out those who are already on your side but won’t show up without a friendly nudge.
Obama’s first presidential campaign certainly did not suffer from a resource crunch. On top of the $750 million raised from donors across the country, his candidacy generated unprecedented volunteer enthusiasm among African Americans and young whites, the same groups that require the most work to mobilize. Staff organizers trying to rouse Unreliable voters in big cities and college towns never had to resort to paid canvassers or import activists from elsewhere to knock on doors. And Obama was popular enough (and Bush-era Republicans were in such disfavor) to do well among swing voters, claiming more votes among those who described themselves as moderates or independents than Al Gore and John Kerry.
As the 2010 midterms approached, Obama’s Organizing for America campaign-in-exile was embedded with the Democratic National Committee and eager to reproduce the magic. But this time the math wouldn’t cooperate. The first strategic decision a campaign usually makes is setting a “win number”: just over half of the total votes projected to be cast in the race, enough to guarantee victory regardless of an opponent’s performance. Next, a campaign assesses how many votes it can count on from those who are certain to cast ballots. (Effectively, the number of Reflex voters in the electorate in question, multiplied by the share of the vote the candidate scores in the polls.) Then it sets out to close the gap between that figure and the win number by mobilizing its party’s Unreliables and registering sympathetic new voters. The tricky part is that the core vote, and thus the gap, keeps fluctuating as some voters’ views on the candidates change.
During the last midterms, presidential-battleground states like Pennsylvania, Wisconsin, Michigan, and North Carolina held plenty of turnout targets—often young and minority voters who had been registered and turned out for the first time by Obama. But the operatives overseeing the party’s strategy didn’t have the resources to mobilize them again. “It was tough to have the kind of volunteer energy to push us over the edge,” says Liz Brown, the former deputy executive director of the Ohio Democratic Party. Meanwhile, as Democrats’ poll numbers sank, their support among Reflex voters shrunk. With those same dynamics leaving fewer and fewer likely supporters to target among infrequent voters, Democratic candidates went into a death spiral.Not all Unreliables are equal, and algorithms can weigh the hundreds of variables attached to a voter’s record to predict an individualized “turnout score,” framed as a 0-to-100 percent probability of casting a ballot. In 2010, Democratic campaigns were forced to reach out, largely in vain, to solidly liberal Unreliables with turnout scores that indicated they were less likely to vote at all.
A trope of 2014 campaign coverage is that Democrats are embracing progressive concerns—a minimum-wage hike, contraception, immigration reform, the villainy of the Koch Brothers—to boost turnout in this year’s midterms. Little of the research informing Democratic tactics supports that explanation. Gerber and Green note that, while traditional messaging about candidates, parties, issues, and policies can change opinions, it does little to influence a person’s likelihood of actually voting, which is of course what matters.No experiments since have challenged that finding, says Green, now at Columbia.
The real reason Democrats have embraced a progressive agenda has not been to energize their own base but to lure Reflex voters from the other side. Obama and his party’s candidates talk about the minimum wage in the hope that working-class whites skeptical of Democrats on other matters will become more ambivalent about voting Republican. Democrats’ renewed interest in women’s issues—including a defense of Planned Parenthood and embrace of equal-pay standards—is also designed with defections in mind. In 2012, the Obama campaign’s entire direct-mail program on women’s issues was targeted at reliable voters who leaned Republican: Field experiments in the first half of that year had showed that the messages were most persuasive among voters whose likelihood of voting for Obama previously sat between 20 and 40 percent.
One of the few bright spots for Democrats in 2010 came in Colorado, where John Hickenlooper was elected governor and interim Senator Michael Bennet won a full term, both victories coming in first-time statewide campaigns. There were many ways to interpret Bennet’s victory—his Republican opponent, Ken Buck, was an extreme and hapless right-winger—but those in the senator’s circle were most encouraged by the situation on the ground. While the Democratic Senatorial Campaign Committee (DSCC) was running TV ads coast to coast on behalf of its candidates, the Colorado race drew one-third of the committee’s entire $7.5 million national field budget. After Election Day, it was easy to assess the return on that spending. Even as Democrats in other states failed to mobilize their way out of unpopularity, in Colorado they hit their win number: Bennet beat Buck by 15,444 votes.
Bennet is now the chair of the DSCC, and his 2010 chief of staff, Guy Cecil, its executive director. Much of their job this year is to protect Democratic seats in states whose politics have been unfriendly to Obama: Arkansas, Louisiana, West Virginia, Alaska. Many of those places do not have strong Democratic organizations or lefty volunteer cultures, and have never experienced the full force of the party’s modernized voter-contact operation. Through a dramatic strategic shift of resources to field-based mobilization—named the “Bannock Street project,” after the location of the Denver office that housed Bennet’s 2010 campaign—Democrats are hoping to reshape the electorate in key midterm states by getting Unreliables to act more like Reflex voters.
The project will entail a total of 4,000 paid workers, most of them working in coordination with other Democratic campaigns. Cecil promises staffing levels and ground coverage comparable to what Obama deployed to his battlegrounds in 2012. “We are not hiring these staff to do all the voter contact,” says Cecil. “We are hiring staff to recruit and manage volunteers” who, per the research, can do that more effectively. In Georgia college towns, that will mean dedicated campus organizers. In the Detroit metropolitan area, it will require tracking down so-called “in-state movers” who have relocated since their last vote, often across municipal boundaries, to ensure they register at their new addresses. In Alaska, a 130-person field team will include staffers posted in remote villages where they will be paired with representatives of local communities and tribal leaders to mobilize rural Unreliables who rarely get much direct campaign contact. In North Carolina, even as their lawyers fight electoral reforms in courts, Democrats will mine public records to identify those who may have developed a habit of casting a ballot on now-obsolete early voting days and convince them to vote under the new rules.
The DSCC’s unprecedented bet on field operations will “only solve our problem,” Cecil concedes, “if the election is a close one.” But, he adds,“if control of the Senate comes down to one state and it will cost one million dollars or two million dollars to reach forty thousand voters, we’re going to spend the money.”
Cecil is far more parsimonious when it comes to his group’s media budget. It is a smarter investment, Cecil has concluded, to spend much of the first half of the year building the infrastructure to support volunteer mobilization rather than delivering a temporary shock to the opinions of Reflex voters.As a result, Democratic candidates will be left to stand by as Republicans and well-funded outside allies like Americans for Prosperity outspend them on advertising. The justification for Cecil’s gambit is found in social-science research that demonstrates the complexity of changing voters’ opinions in any enduring way. In their book The Gamble,George Washington University’s John Sides and UCLA’s Lynn Vavreck demonstrated that the persuasiveness of TV ads aired during the 2012 election dissipated within a day; after five days, it had worn off entirely.
Such findings ought to prompt a reconsideration of how we think about the laps of the horse race. The “it will all come down to turnout” meme misapprehends get-out-the-vote operations as a form of ratification—the final frenzied push to ensure that the people whom candidates have persuaded all year actually cast a ballot. The new playbook on the left, as made evident in the Bannock Street budget, inverts that logic: Democratic Senate campaigns will be designed to mobilize their way into contention, then persuade their way across the finish line. The risk is that November arrives and Democrats are so unpopular that the Unreliables they need to mobilize become too costly and the Reflex voters they need to persuade too far out of reach—“the tipping point where you can’t recover,” as Cecil puts it.
Democrats should not be too worried about the inbound negative ads: There will be millions of Unreliables in Senate battlegrounds this fall who would never vote for a Republican. And once mobilized, a reluctant voter’s ballot counts the same as any other’s. But the enthusiasm and interest of the activists and donors upon whom that mobilization depends can certainly waver. It’s that crucial group of Democrats who most need patience, as well as a confidence in new truths about voter behavior that opinion surveys cannot yet measure—and may not reveal themselves until all the votes are counted.
Sasha Issenberg, a fellow at UCLA’s Luskin School of Public Affairs, is the author of The Victory Lab: The Secret Science of Winning Campaigns.