Big data project to investigate online abuse of candidates during federal election

With a federal election approaching, Canadians are bracing for an uptick in political chatter on social media. So are university researchers from around the world.

Heidi Tworek, an assistant professor in UBC’s history department, and Christopher Tenove, a postdoctoral fellow in political science, are among them. Their new project is one of 18 that will analyze huge amounts of data harvested in real time from social media during the upcoming election as part of the Digital Ecosystem Research Challenge.

We spoke with Tworek and Tenove about what they hope to learn.

What motivated you to get involved with the Digital Ecosystem Research Challenge?

CT: We are interested in how social media platforms are shaping political participation in democracies these days. One of the ways that might be happening is by exposing people to incivility, abuse, hate speech and other kinds of problematic communication that make it difficult for some people to participate—particularly if they belong to groups that are often stigmatized or abused. So we’re part of this collaborative project to understand what social media platforms are doing to democratic politics.

What specific questions will you try to answer?

HT: How has social media changed the game in terms of trolling and harassment of political candidates? What does that really look like in an election in 2019? And we’re trying to answer those questions both quantitatively and qualitatively. We’ll be using data from Twitter, including all tweets from candidates or directed at them, and combining that data with interviews with political candidates. We’ll ask them how they are experiencing this, what kind of social media they see as harassment, and how their offices are actually dealing with it.

How much will you focus on the psychological impact?

CT: In preliminary discussions I’ve had with previously elected candidates, I’ve been struck by the fact that a really big part of it is just their ability to communicate publicly to their constituents and to the general public. Their social media accounts are less of a private thing. Some campaigns might have a full-time communications person assisting them, and the candidate might not even be the first person to read messages on their account. So it’s less about the way in which people are targeted individually and have to respond individually, and more about an organizational issue that has to be dealt with.

Shouldn’t the social media platforms deal with it?

HT: That’s the second part of this, the ongoing question of responsibility that platforms have for the content they host. How far and how quickly are they attending to speech that contravenes their own terms of service? Particularly during an election campaign, how quickly things are being done becomes very key. The other important thing for a country like Canada is the question of context, and understanding how a combination of words that may not seem abusive to somebody outside Canada may in fact be very abusive and harmful to, say, an Indigenous person in Canada. That’s one thing we’re going to look out for, to see whether some types of abusive speech are not picked up by platform moderators because they’re not sensitive to that Canadian context.

How prepared do you think candidates are for this kind of abuse?

CT: That’s a question we are going to be asking. I would say from my discussions with a number of people that the larger parties for the 2015 election were trained on social media strategies, but the preparation around both cybersecurity and dealing with harassment and abuse was not as well thought out or tested. It’s something all parties have been trying to develop, especially since the U.S. experience in 2016. One contribution we hope to make is to learn from candidates and those who assist them, find out what works, and give them more evidence about what works and what might not.