Rowman and Littlefield International

#NeverHillary vs #NeverTrump

Published on Wednesday 21 Dec 2016

The 2016 US presidential election was the most electric event in electoral history—in the fullest sense of the term. Social media played an unprecedented role in the campaigns, allowing the candidates to interact with and influence voters to a greater extent than ever before: via memes, targeted advertisements, hashtags, bots, and other mechanisms. One billion tweets were sent from the primary debates in August up to election day; over 75 million tweets were sent on election day itself. It’s no wonder that this has been called the ‘first true social media election.’

Trump himself has claimed that social media ‘helped him win’ the election, as it allowed him to directly connect with millions of voters. From a numerical perspective, Trump was undeniably more influential. By 8 November 2016, he had 13 million Twitter followers, and Clinton had 10.2 million. His account (@realDonaldTrump) was mentioned 450,000 times a day on average, whereas Clinton’s account (@HillaryClinton) was mentioned 250,000 times. Sentiment analysis has revealed that for both candidates, there were more negative tweets, but for Trump the ratio of negative to positive tweets was smaller.

Trump also dominated Clinton in the aftermath of all three debates. He and Clinton were mentioned 2.1 million times and 866,000 times, respectively, after the first debate; 1.7 million times and 770,000 times after the second; and 1.1 million times and 676,000 times after the third. All but one of the top nine tweeted moments from the debates focused on Trump (only three involved Clinton). They spanned both character and policy issues: his comments that he has ‘good temperament’ and that he is a ‘gentleman’ were as popular as his plans concerning Syria, his opinion of stop and frisk, and—paradoxically—his treatment of women.

These numbers do matter. What Trump said during the debates and throughout the campaigning period drove the discussion on social media and, consequently, on mainstream media as well. Through Twitter, Trump and his supporters were able to talk back against negative news media coverage—and be heard. According to John Naughton and Zeynep Tufekci, Trump’s influence illuminates how traditional media and the two political parties have lost control over public opinion and the boundaries of what can be said. Journalists used to be gatekeepers, and the term ‘Overton window’ denotes the range of policies deemed to be politically correct. The window has now been smashed by the prominence of social media.

So what, exactly, did Trump tweet? Why was it so resonant? It must be noted that Trump’s tweets are so notorious that an archive has been created for all 25,000+ of them: everything from 2009 until the present is instantly searchable. Both his supporters and detractors are fascinated by his rhetoric, his combination of authenticity, passion, and flamboyance. Trump truly aims to tell it like it is—or, rather, like he thinks it is (the difference between the two does not really matter). He takes a #nofilter approach, insulting everyone and everything from T-Mobile to Saturday Night Live to the cast of Hamilton to Whoopi Goldberg. This uncensored thought, coupled with unbridled feeling, engages and provokes his audience in equal measure. Indeed, Trump’s enthusiasm is boundless: 70% of his tweets include at least one exclamation point.

All of this makes him real, and personal, and vulnerable—not to mention highly entertaining. We have all had negative, politically incorrect thoughts that we dare not breathe aloud. Trump thus connects with us on a visceral level; the fact that he exposes his deepest prejudices, and at such odd hours of the morning, distinguishes him from virtually every other presidential candidate. And the timing could not be better—as trust in institutions is at a historic low, nothing could be more attractive than a maverick who represents the complete antithesis of the establishment.

Clinton, on the other hand, used Twitter in a far more traditional way: to promote her campaign and her proposed policies (her media team technically did most of the tweeting). From 27 September to 3 October, she posted 39 times more than Trump about registering to vote. Many of her tweets concerned the White House and polling; more broadly, her consistent use of the first person plural pronoun coupled with imperative verbs (‘we need to’, ‘we have to’) reflected her focus on the actions that Americans must collectively take for the betterment of their nation. It also reflected her professional politeness—she was less direct than Trump in asking the electorate to share their support. One of the most frequent verbs that occurred in Trump’s tweets was ‘donate’; he was far bolder in asking his supporters to give him money and spread his message.

Trump’s dominance of Twitter is further illuminated by the fact that Clinton tweeted twice as much about him as he did about her. Brandwatch analysis has shown that from 1 September to 30 October, 15.9% of Trump’s tweets directly mentioned Clinton, whereas 31.4% of Clinton’s tweets directly mentioned Trump. This is not surprising when one considers that Trump’s tweets are often about himself and express his personality; he brags about his victories and thanks his fans. The first person singular pronoun is far more pervasive in his tweets, as a contrast to Clinton’s plural pronouns. Several of his most frequently tweeted phrases were ‘I will be interviewed’ and ‘I will beat Hillary’. Although he certainly castigated Clinton (using the adjective ‘crooked’ an inordinate number of times), she embarked in far more negative campaigning against Trump.

As meticulously manicured as most of Clinton’s tweets were, she did have a few unexpectedly resonant moments. She penned the most popular tweet in the campaigning period: ‘Delete your account’. This was in response to Trump’s impertinent tweet about Obama’s endorsement of Clinton’s campaign. Here Clinton is far snarkier—and less professional—than usual. By slipping out of character and speaking the street language of social media, she became more authentic and appealing. Not to mention attention-grabbing: articles in the New York Times, the Guardian, and numerous other news channels were written about this Twitter exchange.

Only after the campaign period was over did Clinton fans show their numbers on Twitter (much as EU supporters became more vocal after the referendum). The most popular political tweet of 2016 was a quote from Clinton’s concession speech on 9 November that reassured women and girls that there was still hope. It has received over 639,000 retweets and 1.1 million likes. These staggering metrics are not surprising, given the firestorm of grief and rage triggered by the election result—powerful emotions that had not previously been ignited among Clinton supporters.

To fully account for Trump’s success on Twitter, we must take into consideration the users who supported him and their position in the social network. They were, on average, more influential than Clinton’s supporters as measured by social media metrics: their follower to friend ratio, the amount of engagement their tweets generated, the accounts they interacted with. Interestingly, they were not as influential in real life as Democratic users—more of them were ‘ordinary’ people, as opposed to institutions, politicians, or journalists. This further underscores the tension between traditional organisations and social media.

It is also important to examine the role and impact of automated content, which is becoming more and more pervasive in political contexts. According to the Project on Computational Propaganda (COMPROP) at the Oxford Internet Institute, ‘Political bot activity reached an all-time high for the 2016 campaign.’ Approximately one out of every five election-related tweets was from a bot; bot activity increased during the debates. By election day, Trump bots were five times more active than Clinton bots. These automated accounts were used in a ‘deliberate and strategic’ manner: pro-Trump campaigners carefully timed their tweets during the debates and hijacked pro-Clinton hashtags (like #ImWithHer and #uniteblue). They also used hashtags like #TrumpTrain and #CrookedHillary to spread fake news. Many of the bots were not produced by Trump’s campaign—rather, they were from political action committees (PACs) and extremely fanatical individual supporters (such as Palmer Luckey).

So what does all of this mean for future elections? Now that we’ve had our first Twitter candidate, are we likely to see the first Instagram or Snapchat candidate in 2020? Social media has undeniably increased the complexity of the campaigning landscape—it is one more channel through which voters can be influenced. It is also an extremely useful laboratory for ideas: they can be tested quickly, and the most resonant can be incorporated into offline messaging. Although Twitter and Facebook have definitely reduced marketing costs, traditional media is still important—Trump spent about $300 million on television advertisements, which is over $80 million more than what John McCain spent in 2008. It is, however, about half of Clinton’s budget.

Future candidates can learn a valuable lesson from Trump and his attention-arresting tweets. There is just one hurdle: authenticity is, by definition, hard to fake. And a bot certainly won’t do the trick.


Yin Yin Lu is a DPhil Candidate at the Oxford Internet Institute and Balliol College; her research focuses on hashtags, rhetoric, and resonance in new media. She is also the founder and convenor of the #SocialHumanities research network, through which she chaired a panel discussion on the US election on social media.