According to a 2016 study by the Pew Research Center, 62 percent of Americans get at least some of their news from social media. On Twitter, a platform with 67 million monthly US users, nearly 60 percent of people look to their feeds to find out what’s going on in the world.
But when users look can have a significant impact on how they perceive major events like last year’s Republican presidential primary debates. In a new working paper titled “Make America Tweet Again” — a clever reference to then-candidate Donald Trump’s unprecedented mastery of the micro-blogging platform — data collected just before, during, and after three of those GOP contests shows how sensational tidbits tend over time to rise to the top of the Twittersphere, drowning out genuine policy discussions.
While the three GOP debates were actually taking place, tweets tagged with #GOPDebate reflected a balanced mix of tabloid chatter and substantive talk on the election’s major issues like immigration and healthcare. Immediately following the debates, however, policy talk all but disappeared as the conversation was dominated by sensational tweets concerning things like Trump and Senator Marco Rubio’s spat over the size of Trump’s hands — and Trump’s subsequent bragging about his other anatomical endowments. Similarly, while the heated exchange between Megyn Kelly and Donald Trump over his treatment of women briefly registered at the time on Twitter, it was only after the debate that the conversation truly took off.
Retweets also became increasingly popular, comprising some 80 percent of all tweets within twenty minutes of the debate’s conclusion as the most entertaining takes on the debate’s most sensational moments echoed through the Twittersphere, racking up retweets.
“Based on our findings, there may be — perhaps unfortunately — an advantage for candidates who lead with more sensationalist statements earlier in the debate,” says one of the paper’s four authors, Shiri Melumad, a doctoral candidate in marketing at Columbia Business School. This dynamic may have played particularly well for Trump, whose brash attention-grabbing style was virtually built for memes.
Melumad and coauthors Ron Berman, Colman Humphrey, and Robert Meyer, all of the University of Pennsylvania’s Wharton School, also examined how people’s general opinions of the candidates changed over time.
While the debates were happening, individual’s sentiments toward all of the candidates tended to be more negative than they were either before or after — perhaps reflecting Americans’ dismal up-close assessment of a highly unpopular slate of candidates. This was particularly true for Senator Ted Cruz, the authors note. As each of the three debates wore on, the sentiment of tweets mentioning Cruz grew increasingly dismal, only to rebound once the debate ended.
“If you’re interested in gauging voter opinion based on Twitter, it is critical to consider when certain content was Tweeted,” says Melumad.
While the authors find significant variation in the affective content of the Twittersphere before and after the debate, during the debates the topics of conversation appeared to have little impact on tweeters’ sentiment. For example, during the August 2015 debate, when moderator Megyn Kelly asked Trump about fraud allegations surrounding Trump University — a topic that roiled the live audience and might have inspired Trump’s critics to throw some digital shade — there was no corresponding real-time surge in tweet negativity.
While previous research has suggested that emotionally charged messages — ones meant to evoke strong positive or negative reactions — are most likely to go viral, the authors found instead that the most shared tweets in their sample were those that Linguistic Inquiry and Word Count (LIWC), a text-mining and analysis tool, aligned with drives of “achievement” (references to success and failure) and “power” (references relevant to status, dominance, social hierarchies). Both of those come across loud and clear in a widely shared tweet from conservative commentator turned NRA spokewoman Dana Loesch:
Rubio is DESROYING Trump. ‘He just says we’re going to ‘win win win,’ ‘Make America great’ …” #GOPDebate
What’s more, the data shows that in the crucial post-debate timeframe, such viral messages are more likely to come from individual Twitter users than from corporate accounts or traditional news sources like CNN that dominated the pre-debate conversation, highlighting the growing importance of personalities and “takes” in the modern media environment.
“Social media sites enable voters not only to express their opinions about candidates, but also to broadcast these opinions to potentially enormous audiences,” Melumad says, adding that many of the people who tweeted during the GOP debates had thousands of followers — and users were deeply engaged throughout the debate. #GOPDebate tagged tweets were generated at a rate of 201 per second from US-based accounts alone, compared to an average total of around 6,000 tweets per second globally, according to CIO magazine. That’s a tremendous wall of sound facing potential voters checking their feeds for a little context on the debate.
“Because of the platform provided by social media sites like Twitter, social influence is now playing a much larger role in voting decisions,” Melumad says. “During these debates voters are taking into account not just what candidates are saying and how they are saying it, but also what other voters have to say about it.”
Because of this — and because of Trump’s unique success with the 140-character limit — last year’s race for the White House has been called the nation’s first “Twitter Election.” There’s reason to believe it won’t be the last.
“While Trump has certainly pushed ‘political tweeting’ further into the spotlight, the trend we saw toward sensationalism is something that is happening across the political spectrum,” Melumad says. “In our data we saw it hold for other Republican candidates, as well as in the final presidential debates with Hillary Clinton.”