Incompetence in Clintonland: How Team Hillary Blew the 2016 Election
The Clinton campaign made decisions in the 2016 election with breathtaking arrogance. The campaign conducted zero state polls in the last three weeks of the election. They refused to fire their chief of data analytics, Elan Kriegel, after getting Iowa and Michigan badly wrong in the primaries. And they doomed themselves by cancelling a deep dive poll of mid-western states four weeks before the election.
When news of Trump’s victory reached the White House on November 8, 2016, officials in the Obama Administration lit into Clinton, convinced that she was directly responsible for most — if not all — of the controversies that damaged her election chances:
‘No one forced her to underestimate the danger in the Midwest states of Wisconsin and Michigan.
‘No one forced her to set up a private email server that would come back to haunt her.
‘No one forced her to take hundreds of thousands of dollars from Goldman Sachs and other pillars of Wall Street for speeches.
‘No one forced her to run a scripted, soulless campaign that tested eighty-five slogans before coming up with ‘Stronger Together’.
In addition to those examples, the campaign’s arrogance resulted in several, lesser-known, strategic missteps. For example, they delegated virtually every campaign decision to data analytics and election modelling, they opened up few field offices, scheduled candidate visits to battleground states infrequently, and spent more than a billion dollars on personal attack ads against Trump that were, for the most part, ineffective .
2016 Clinton Campaign Headwinds
Going into the 2016 presidential election, Hillary Clinton’s campaign urgently needed to achieve two things: reverse her historic unpopularity and successfully navigate a poisonous media landscape. Clinton needed a campaign committed to a humble, strategic approach, but the organization she ended up with proved incompetent on both counts.
One major headwind for the campaign was that, after Trump, Hillary Clinton was the least liked major presidential candidate ever recorded in American political history. At the height of the election, polls showed Clinton was getting the lowest favorability rating she ever received in 25 years of public life. Just before the election, 52 percent of Americans viewed Clinton negatively. Exit polls found almost half of all 2016 voters (47 percent) didn’t think Clinton was trustworthy, four in 10 voters said she was corrupt (39 percent).
The overwhelmingly negative media environment would not improve her unpopularity either. Fox News ran a venomous, years long campaign against Clinton, broadcasting 330 stories about her throughout 2015. 300 of them were overtly negative. Eight of America’s most influential news outlets wrote coverage “negative in tone” about Clinton 84 percent of the time — compared to 43 percent for Donald Trump and 17 percent for Bernie Sanders. Overall, Clinton was covered more negatively: she received 62 percent negative and 38 percent positive coverage. Trump’s coverage was 56 percent negative and 44 percent positive. In the meantime, Trump earned $2 billion of free media coverage, more than any other candidate — maybe ever.
At the outset, the headwinds of 2016 were blowing fiercely against an unprepared Clinton campaign.
Clinton’s Crippling Dependence on Data Analytics
Out of the 130 million votes cast in 2016, Hillary Clinton’s loss came down to 77,000 votes in three states: 44,292 in Pennsylvania, 10,704 in Michigan, and 22,748 in Wisconsin. After the election, a common refrain from Clinton and her supporters was that FBI director James Comey cost her the election by making the crucial last days of the campaign about her email server.
Of course Clinton’s loss can be attributed to any number of factors including the FBI Director’s letters about her emails, the lack of a compelling economic message, or Russian hacking and interference. But the narrative painted by former Clinton campaign operatives is far more consistent: the Clinton organization followed a one-size-fits-all approach drawn entirely from pre-selected data which guided campaign Chairman Robby Mook’s decisions on field, television and everything else.
Inside Clinton’s analytics department there was a sign attached to the ceiling with the words: “statistically significant.” Kriegel had computers in the Brooklyn campaign headquarters programmed to run 400,000 election simulations every night, stress-testing the possible outcomes on November 8. This was the brain inside the Clinton machine, and it decided everything.
The campaign’s dependency on analytics grew over the course of 2016, bypassing the expertise and instincts of traditional media buyers, calculating the “cost per flippable delegate,” and outputting what states, television markets, networks and shows to buy. Mook spoke glowingly of Kriegel and described his influence expansively: “His hand has guided almost every aspect of what we do” Mook said. He was hardly exaggerating.
Nearly every strategic campaign decision made by the Clinton campaign was decided by a data-fed algorithm. The Clinton Campaign was so dependent on its data modelling that algorithms decided every media buy in the primary against Bernie Sanders ($60 million). Data analytics so dominated the Clinton campaign that once while Clinton was flying to a battleground state for a rally, the algorithm recalculated its recommendation about the rally’s location and Hillary diverted her plane accordingly.
Data modeling inside Clinton’s campaign made decisions on things like:
· The cities and states where the campaign should spend money and where it should open field offices
· When the campaign should send emails to supporters as well as the contents of the emails themselves
· What zip-codes and neighborhoods volunteers should canvass, and what phone numbers they should dial
The warning signs that the data modeling might not be as good as some thought were early and clear, but the campaign failed to take corrective action. Clinton’s data operation predicted she would win the Iowa caucuses by 6 points and that she would win the Michigan primary. She ended up winning Iowa by two-tenths of a point and lost Michigan to Sanders.
The campaign’s algorithms also missed critical changes in the general electorate. It failed to identify the importance of Michigan (again) and Wisconsin; weaknesses that only became apparent until it was too late to act. According to one internal estimate, the Clinton campaign spent about 3 percent as much in Michigan and Wisconsin as it spent in Florida, Ohio and North Carolina — most voters in Michigan hadn’t seen a Clinton tv ad until the final week of the campaign. The bottom line was that Clinton’s data analytic myopathy failed to anticipate what eventually sunk her candidacy: that she would lose the white women vote by 10 percent and that the vote in union households would split nearly 50–50.
The data revolution in campaigning is remarkably valuable in modern electoral politics, but the Clinton campaign was negligently over dependent on analytics and repeatedly ignored alarming messages from Democratic Party officials, pointing out the campaign’s failure to reach persuadable voters. As a result, the campaign paid little attention to qualitative focus groups and ignored feedback from the field.
Clinton’s Soft Ground Game
A campaign’s “ground game” refers to the ability to organize local communities of supporters to drive up voter turnout. Campaigns use techniques like phone banking and door knocking which are coordinated out of the local field office. These offices also register voters and organize transportation to the polls.
Studies have shown a positive relationship between field offices, voter turnout and a bump in the candidate’s share of the vote. Recent candidates for president do roughly a half percentage point better in counties with field offices than in counties where they don’t. Separate studies have shown an increase in voter turnout among people who live closer to field offices, generally an advantage for Democratic candidates.
In 2016, Clinton minimized her ground game and opened very few field offices, limiting outreach to crucial communities. Nationally, the Clinton campaign opened 537 field offices — 410 fewer than Obama had in 2008 and 252 fewer than he had in 2012. In Wisconsin, for example, Clinton opened 40 field offices. In 2012, Obama had 69.
One former Obama operative said that “there’s this illusion that the Clinton campaign had a ground game. The deal is the Clinton campaign could have had a ground game. They had people in the states who were willing to do stuff. But they didn’t provide people in the states with anything until GOTV.”
The reality on the ground for many of the states that decided the election (Pennsylvania, Michigan and Wisconsin) was that Hillary was struggling terribly with enthusiasm and low popularity. Instead of addressing these problems, the campaign underinvested in things like on-the-ground campaign literature, door knocking, and commitment to vote cards. These traditional ground game techniques would have signaled to voters in those states that the campaign had momentum, that Clinton wasn’t taking their votes for granted. Nonetheless, the command structure inside the campaign dismissed concerns about their on-the-ground engagement and believed that television, limited direct mail and digital marketing were the only valuable tools.
The Clinton campaign’s refusal to prioritize door knocking to drum up support meant the campaign was not hearing from voters they assumed were already likely Clinton voters. This left no information to check the models against. Of the actual real-time voter information collected by the campaign’s volunteers and canvassers, most of it was piled up in bins at campaign HQ never to be touched.
An effective ground game might have, for example, showed some of the white male union members the campaign expected to be likely Clinton were in fact leaning toward Trump. In other words, a stronger ground game may have revealed a substantial realignment of white working-class voters, a normally reliable constituency for the Democrats, who swung heavily for Trump. As it turned out, between December 2015 to December 2016, 18% of all white working-class Democrats left their party and started identifying themselves as “leaning Republican” in their survey responses. This remarkable shift went undiscovered and never made it into Clinton’s modelling.
On election day, Clinton’s team was shocked to have received only 78% of white working-class voters that had previously voted for Obama. Trump received 18% of Obama’s 2012 vote. Exit polls showed 45% of all Obama-Trump voters identify as “Republican leaning.” An effective ground game would have caught this pattern and a competent campaign would have responded accordingly.
Clinton’s Infrequent Personal Campaign Visits
Campaign visits by a candidate can have a statistically significant, county-sized, impact on voting preferences. These visits also have the bonus effect of generating a lot of local and state media attention. While some candidates’ visits can have measurable impacts on voter preferences, other candidates’ visits do not. In 1948, visits by candidate Harry Truman had a significant impact on the county-wide vote while events featuring Dewey, Truman’s opponent, had no impact at all. In 1996, Bill Clinton’s campaign stops made a difference. But his rival, Bob Dole, no matter where he campaigned, his visits had no impact on voters.
It might surprise some to learn that Clinton was actually a good campaigner. Data shows Clinton’s in-person campaign visits had an overall positive impact on her share of electoral support, just like those of Truman and her husband. Visits by Trump and third-party candidate Gov. Johnson, however, failed to register any similar impact on vote percentage. But again, this data clearly was never properly considered by the campaign’s algorithms.
Unfortunately for Clinton’s chances, the campaign failed to understand the importance of candidate visits, reflecting a stunning lack of confidence in their own candidate.
Clinton ended up having held eighteen (18) fewer personal events than Trump in Florida, North Carolina, Pennsylvania, Ohio, Michigan and Wisconsin (78 vs 96). By election day, Clinton became the first major party presidential nominee since Nixon in 1972 not to visit Wisconsin during the election.
Clinton’s Failed TV Advertising Strategy
Clinton spent over $1 billion on the least-issues focused television advertisement campaign run by a Democratic presidential candidate in decades. The Clinton campaign relied almost exclusively on negative, personal attack ads against Trump. According to one analysis, 90 percent of Clinton’s ads went after Trump individually and contained little to no policy, while 70 percent of Trump’s ads were estimated to have contained at least some policy.
As it turned out, attacking Trump personally and not talking about policy was a wholly ineffective strategy. A recent study played over a dozen different ads for 2 million likely voters to find out which ads were least and most persuasive. The study found that personal anti-Trump ads were among the least persuasive. But ads that highlighted core Democratic policies on the economy and healthcare were found to be three to four times as persuasive as personal attacks on Trump.
Conclusion
The thrust of the Clinton campaign strategy was to maximize base-voter turnout and invest less in trying to persuade undecideds. In the final analysis, this strategy was a catastrophic failure. 2016 was a persuasion, not a turnout election. Not only did turnout not really matter (it played only a modest role in Trump’s victory) the biggest driver of Trump’s gains was persuasion: he flipped millions of white working-class Obama supporters to his side. Amory Beldock, a former Clinton campaign field organizer in Miami Dade explained why “mobilization over persuasion,” was a failed strategy:
The Clinton campaign’s singular focus was on rebuilding the Obama coalition of minorities and millennials. They used advanced data analytics but ultimately failed to mobilize an apathetic voting bloc. It was the fatal combination of identity politics and data obsession that dealt Team Hillary the final blow.
This is part one of a multipart series on the 2016 and 2020 elections. Part II will examine whether Joe Biden’s organization in 2020 has heeded lessons from the string of incompetent strategies that were embraced by Hillary Clinton’s 2016 presidential campaign.
