The Digital Democracy Project (DDP) was set up to help build the international evidence base on the impact of these trends with a robust Canadian case study whose methods could be applied in other locations. The project consists of three phases. The first was a two-day workshop for journalists about disinformation threats, held in May 2019 in Toronto. The second, and the focus of this report, was researching and monitoring the digital media ecosystem in real time ahead of the Canadian federal election on Oct. 21, 2019. The third and final phase, beginning in early 2020, will involve further research and consultations with experts and public representatives to develop policy recommendations.

We launched this phase of the project in August 2019 and continued collecting data until the end of November. This work builds on the growing field of study of election integrity and, in particular, on the study of the spread and influence of media exposure (both online and offline) and of disinformation and toxic content on the behaviour of voters. Using a novel approach that combined online data analysis with a battery of representative national surveys, we sought to contextualize and better understand developing patterns of online activity with measures of media consumption, trust and partisanship.

Overall, our findings suggest the Canadian political information ecosystem is likely more resilient than that of other countries, in particular the U.S., due to a populace with relatively high trust in the traditional news media, relatively homogenous media preferences with only a marginal role for hyperpartisan news, high levels of political interest and knowledge, and — despite online fragmentation — fairly low levels of ideological polarization overall. While we do find affective polarization, which involves how individuals feel about other parties and their supporters, we find less polarization on issues, which has been a key point of vulnerability in other international elections.

Despite some worries about automated activity being used to game trending hashtags on Twitter or the presence of a few disreputable online outlets, our research suggests their impact was limited. While there remain significant blind spots in the online ecosystem caused by limited data access for researchers, based on the communication we could see, we did not find evidence of any impact attributable to a co-ordinated disinformation campaign.

Looking forward, however, we find evidence to suggest potential future vulnerabilities, most of which are related to growing partisanship and polarization, as well as the segmentation of the populace into online information environments that reinforce existing world views.