Scientists Jiebo Luo and Yu Wang, from the University of Rochester (USA), had not set out to predict who would win the 2016 elections that ended with Donald Trump as the 45th president in US history, but His exhaustive study –based on data science tools– of the followers each candidate had on Twitter offers key clues as to why the businessman has ended up in the mythical oval office of the White House .
“We wanted to understand how each candidate’s campaign was evolving and be able to explain why someone was winning or losing, ” says Luo, associate professor of Computer Science. According to their university, Luo and Wang summarized the findings collected in those fourteen months of study (from September 2015 to October 2016) in eight articles that were published during the presidential campaign.
They made observations such as that the more Donald Trump tweeted, the faster his following grew , even after failing to do very well in debates against the other Republican candidates and sparking controversy with proposals such as banning Muslim immigration from the US. country.
Investigators also found that when Trump accused Hillary Clinton of playing the female card, during the following week women were more likely to follow Clinton and less likely to unfollow her. However, the Republican’s comments did not affect the gender composition among his own followers.
On the other hand, they observed that the “gender affinity effect” seen in other elections – which implies that women are inclined to vote for people of the same gender – did not appear to be working in Clinton’s favor when the primaries came to an end. . The percentage of women who followed Clinton on Twitter was no greater than what they assumed in the case of Trump. Moreover, although it was more likely that those who stopped following both candidates were women, the phenomenon was “particularly pronounced” in the case of Clinton, they point out.
Discontent among Sanders supporters
At the same time, several prestigious media polls suggested that some of the supporters of Bernie Sanders – the Democratic candidate that Hillary Clinton faced in the primaries – could leave the ranks of this party and end up voting for Trump if Sanders stayed. out of the presidential race. Luo and Wang found evidence that seems to confirm this by noting that the number of Sanders followers who also followed Trump had grown – while the number of followers who also followed Clinton was falling. The followers who ‘shared’ Sanders and Trump were also mostly men – around 64% -.
“In the end, although we chose not to make any predictions about it, we were not at all surprised that Donald Trump won, ” Luo explains. But why did these researchers take notice of Twitter? As early as 2008, Barack Obama’s use of social media in the presidential race helped establish them as very powerful tools for candidates in reaching a much larger number of potential voters, as well as receiving feedback from them. and be able to attack your rivals.
Since then, there has been a growing interest among data scientists to analyze elections based on social media posts. Twitter, in particular, is a very rich data source because the millions of tweets that are posted every day are easily accessible .
Image: Gage Skidmore via Wikimedia / CC
