Measuring Emotion in a Volatile Election
Co-authored by Jason Brush and Michael Cobb, and originally published in The Huffington Post.
The 2016 US presidential election has played out on social media like none before. In particular, Donald Trump’s unprecedented rhetoric seems to have found its perfect outlet with Twitter. His short, punchy accusations and provocations—a natural extension of his public presentation style—has been polarizing: invigorating his base while at the same time outraging Hillary Clinton supporters.
But with the revelation that a third of pro-Trump tweets are generated by bots, how effective has his use of Twitter been? And how has Trump’s use of Twitter compared to Clinton’s?
Twitter’s use by both candidates has little to do with policy—of Clinton’s 907 tweets in September, only 25% were directly related to a defined policy (contrasted to 37% dedicated to attacking Trump). Trump fared substantially worse, with only 2% (yes, only two percent) of his 755 tweets related to policy.
Instead of discussing policy, social media has been used by the candidates to inflame emotions on both sides. And this is key, because studies show that we make voting decisions like we do many of our decisions in life: guided by emotion. Given how politically polarized our country is, and how each side views this election as a referendum on American identity itself, it’s not surprising that emotion is used by the candidates to rally support on social media.
The importance of emotion, and the central role social media has played in this election, led us to question how successful the candidates have been in their use of emotion on social media—Twitter in particular.
- How often did each candidate tweet with measurable emotion (i.e., joy, trust, anticipation, surprise, fear, disgust, anger, or sadness)?
- When they tweeted with emotion, which emotions did they express?
- And when they tweeted with emotion, did the emotions they tweeted correspond to the emotions tweeted by the public about the election?
Looking at Twitter between July 13th and October 10th, we classified tweets from the public, as well as Trump and Clinton’s handles, per the following emotions: joy, trust, anticipation, surprise, fear, disgust, anger, or sadness.
For each day, we identified when the measurable emotions in the candidate’s tweets matched those emotions in the general public’s tweets—three things stood out:
First, Clinton tweeted with emotion far more than Trump. Despite Republicans have regularly denounced Clinton as being emotionally cold—with RNC Chair Reince Priebus going so far as to mansplain that smiling more would help her image—Clinton has tweeted 43% more emotional tweets than Trump.
Second—and even more telling—the emotions of her tweets matched the public’s more than twice as often than Trump’s.
Lastly, the emotions in Clinton’s tweets matched a far greater range of the public’s emotions than Trump’s.
In short, on Twitter, Clinton shows far more empathy than Trump.
Looking more closely at three key moments during the campaign thus far—the Democratic National Convention, the first debate, and the leak of Trump’s taped lewd conversation with Billy Bush—we see examples of how the dominant emotions in Clinton’s tweets more clearly map to the dominant emotions being expressed by the public.
After the Access Hollywood tape leak, the tweeting public reacted to this video intensely with disgust rather than anger:
Clinton mirrored the emotional tenor of the public (anger and disgust) (“this is horrific”) then pivoted to fear (“we cannot allow this man to become president”):
Trump tried to deflect through anticipatory urgency:
But, when addressing the scandal itself, he’s emotion-free:
But has Clinton’s empathy been effective? The idea that she is “better” at Twitter than Trump might seem counterintuitive. However, we can tell that it helps with her base: when Clinton tweeted emotional content, it was retweeted by her followers a remarkable 41% more than her average content was.
And has Trump’s non-empathetic approach worked for him? At first glance, it might seem like his pugilist style works just fine: when looking at the public’s emotional conversation surrounding the election, he is mentioned in 79% of the tweets. A “yuge” number, as he would say.
Though being the center of attention isn’t always a good thing. When anger is expressed toward Clinton on Twitter, Trump followers account for 47% of the conversation. But when anger is expressed towards Trump himself, Clinton followers account for only 36% of the conversation. This suggests that people are much more broadly aligned in negative sentiment toward Trump than Clinton.
In contrast, Clinton succeeds with her use of emotion on Twitter by using the same approach that worked so well for her husband, so many years ago: I feel your pain.
Through his lack of emotive empathy, Trump has another message: You are in pain, and whatever that may be, I am the solution.